Академический Документы
Профессиональный Документы
Культура Документы
VOLUME 1
FIRST EDITION
Second Printing, with revisions
All rights reserved. Apart from any fair dealing for the purpose of private study,
research, criticism or review, as permitted under the Copyright, Designs and Patents
Act 1988, no part of this publication may be reproduced, stored in a retrieval system
or transmitted in any form or by any means electronic, electrical, chemical, mechani-
cal, optical, photocopying, recording or otherwise, without prior written permission.
All enquiries should be addressed to the Publisher.
DIR Publishing Ltd retains sole and exclusive right to produce, or to authorize others
to produce, printed-circuit boards or kits for the transverter designs published in this
book. Amateur constructors are authorized to build the equipment designs in this
book for their own use only.
Whilst every care has been taken in the production of this book, the Publishers
cannot be held legally responsible for accidental errors or consequences arising
therefrom.
ISBN 0952046806
ii
EDITOR
Ian White G3SEK
AUTHORS
Roger Blackwell G4PMK
David Butler G4ASR
Geoff Grayer G3NAQ
Gnter Hoch DL6WU
Sam Jewell G4DDK
John Nelson GW4FRX
Dave Powis G4HUP
Dave Robinson G4FRE/WG3I
Ian White G3SEK
John Wilkinson G4HGT
iii
2017 Digital Replica Edition
The VHF/UHF DX Book was written in the early to mid-1990s by a team of experienced
VHF/UHF DXers and equipment developers, in an effort to pass on our knowledge
and stimulate further developments.
Looking back, that plan worked very well... for a time. But as we entered the digital age,
information of lasting value became trapped on the printed pages and was lost
to the next generation of VHF/UHF DXers.
To keep that information alive, this digital replica of the Second Printing
dated 1995 has been released by kind permission of the copyright owner TRPublishing and its
proprietor Trevor Preece G3TRP.
As with all older books, the challenge for the reader is to separate the parts that are of lasting
value from other parts that have become dated. But we make no apology
for the latter; they remain an accurate snapshot of VHF/UHF DXing in its heyday.
These designs are over 20 years old and many of the components
are no longer available.
More modern designs are available today, so seek them out and use those instead.
iv
CONTENTS
CHAPTER 1
Introduction
CHAPTER 2
VHF/UHF Propagation
CHAPTER 3
Operating
CHAPTER 4
Assembling your Station
CHAPTER 5
Receivers and Local Oscillators
CHAPTER 6
Transmitters, Power Amplifiers and EMC
CHAPTER 7
Beam Antennas and Feedlines
CHAPTER 8
144MHz
The Suffolk 144MHz Transverter 8-2
The W1SL 144MHz Power Amplifier 8-31
Antennas for 144MHz DX 8-37
CHAPTER 9
50MHz & 70MHz
A High-performance 50MHz Transverter 9-2
The Cray High-Performance 70MHz Transverter 9-20
Solid-State Power Amplifiers 9-34
Antennas for 50MHz and 70MHz 9-39
CHAPTER 10
432 MHz
A DXers Transverter for 432MHz 10-2
A Low-noise 432MHz GaAsFET Preamplifier 10-25
The K2RIW 432MHz Power Amplifier 10-27
Antennas for 432MHz DX 10-34
CHAPTER 11
Power Supplies and Control Units
CHAPTER 12
Test Equipment and Station Accessories
v
vi
CHAPTER 1
INTRODUCTION
1 1
INTRODUCTION
INTRODUCTION
by Ian White G3SEK
1 2
INTRODUCTION
1 3
INTRODUCTION
constructional side. And theres no better all sizes, and on the equally important topics
motivation for a DX-chaser than the thought of matching the antenna to the feedline and
of missing some DX! making it work properly. There is also a
comparison of computer-predicted perform-
DESIGNING FOR DX ance of a wide variety of commercial antennas
If you want your station to work at its full for 144MHz and 432MHz, contributed by
potential, you have to understand how it Rainer Bertelsmeier DJ9BV, the Technical
works. Its very tempting to jump straight into Editor of DUBUS magazine.
the details of circuitry and hardware, com- Chapters 47 have thus explained how your
pletely overlooking the fact that all the bits station ought to perform, and the basic
and pieces eventually have to work together as principles of RF equipment design. Chapters
a system. 812 then show how its done in practice.
System design is what transforms a collec-
tion of boxes, cables and antennas into an EQUIPMENT DESIGNS
effective station. Chapter 4 explains the Chapters 8, 9 and 10 present modern equip-
general concepts and encourages you to ment designs for working DX on 50MHz,
develop the capabilities of your station to 70MHz, 144MHz and 432MHz. Every item is
make difficult DX contacts using weak signals. real, tried and tested.
Chapter 5 expands these ideas in the area of No previously published transverter designs
receiver design, and explains the techniques met our performance standards for modern-
of noise and intermodulation analysis which day DX operating, so here are four all-new
are used to design the transverters in Chapters designs. The transverters were developed by
8, 9 and 10. Dave Powis G4HUP (50MHz), Dave Robinson
Chapter 6 is about Transmitters, Power G4FRE/WG3I (70MHz) and John Wilkinson
Amplifiers and EMC. John Nelson G4HGT (432MHz). Sam Jewell G4DDK
GW4FRX reminds us that a poor-quality developed the 144MHz transverter and co-
signal is a problem for everybody else and he ordinated the whole transverter project.
pulls no punches about the poor quality of Because these new transverters follow the
many present-day transmissions. Well known system-design principles of Chapters 4 and 5,
for his own impeccable signal, John clearly they are much more future-proof than
explains the causes of inferior signals and how normal. Although the transverters are already
to remedy them. GW4FRX builds quality into close to optimum for todays operating
every aspect of his power amplifiers, and conditions, the information in Chapters 4 and
Chapter 6 contains a wealth of sound advice 5 will help you to analyse the effects of any
on amplifier design and construction. He then changes you may wish to make in the future.
explains how to commission and operate a Each transverter is divided into modules, so
newly-built amplifier without hazards or heart you can easily make changes if new devices
attacks. Finally he tackles the increasingly become available to do the same jobs better.
important question of how to run high power These transverter designs will help to bring
whilst living amicably with your neighbours. your station close to the leading edge and
Until recently the highly successful DL6WU keep it there.
family of long Yagis was virtually unknown in Chapters 8, 9 and 10 contain all the infor-
the UK so were pleased to welcome Gnter mation you need to build your own transvert-
Hoch DL6WU as the contributor of Chapter 7 ers and get them working. But that isnt all:
on Antennas. He begins with a general each designer has also explained in detail how
survey of VHF/UHF antennas, which clearly he developed, constructed and commissioned
shows why the Yagi has become a firm his own transverter. All four transverters are
favourite. The chapter continues with advice based on the system design principles out-
on choosing, designing and installing Yagis of lined in Chapters 4 and 5 so they show a
1 4
INTRODUCTION
distinct family relationship. Yet the circuitry, we dont even agree completely amongst
construction and alignment details reveal four ourselves but thats entirely appropriate
peoples quite differing approaches to the task because VHF/UHF DX is a developing subject
of designing a transverter and making it work. area.
By comparing and contrasting their individual Although we dont claim to have covered
accounts, you can learn how design decisions the whole of VHF/UHF, the information in
are made along with a great deal more about this book has lots of applications to related
the art of practical RF electronics. areas. In particular we hope that amateur-
As well as the new transverters, Chapters 8, satellite users will find a great deal of interest.
9 and 10 contain designs for power amplifiers And even local communications such as
and antennas for the respective bands. Some simplex FM, repeaters and packet radio can
of these are well known in other countries but benefit from the RF engineering design
have never been published before in the UK. principles described here.
We have also included updated versions of
classic 144MHz and 432MHz power amplifiers OVER TO YOU
published many years ago. GW4FRX returns This book offers you a flying start into VHF/
in Chapter 11 to deal with Power Supplies UHF DXing. Theres still a lot to learn and a
and Station Control and gives practical new generation of ideas and techniques to be
designs and advice on integrating your developed. So now its over to you...
equipment into a safe and functional trans-
mitting station. Good DX!
Chapter 12 is devoted to Test Equipment,
because measurements go hand-in-hand with
design to keep your station at peak perform-
ance. Roger Blackwell G4PMK takes delight in
building equipment which almost equals the
performance of professional testgear, at a
fraction of the cost. Chapter 12 is his selection
of the most useful testgear and accessories for
the VHF/UHF DX station.
HOW IT STARTED
I started this project because no existing
amateur-radio book seemed to capture the
reality of VHF/UHF DX, with its unique blend
of advanced technology and personal skill.
Even the more specialized VHF/UHF hand-
books are too general; they contain enough
information to start an interest in DX, but not
enough to satisfy it.
In recruiting the other authors, my aim has
been to build up a complete picture of VHF/
UHF DXing through contributions from
specialists in particular areas. Although weve
tried to write the book on VHF/UHF DX,
what youll read are not bland accounts in the
detached style of a textbook; they are often
personal views on controversial topics. You
may not always agree with what we say and
1 5
INTRODUCTION
1 6
CHAPTER 2
VHF/UHF
PROPAGATION
2 1
VHF/UHF PROPAGATION
VHF/UHF
PROPAGATION
by Geoff Grayer G3NAQ
2 2
PROPERTIES OF RADIO
WAVES
2 3
VHF/UHF PROPAGATION
during a complete oscillation cycle. Although wave is shared equally between the electric
we could refer to the strength of either the and magnetic fields and flows along the
magnetic field (H) or the electric field (E), we direction of propagation. Its magnitude is
usually mean the latter. In space they are given by:
related by the formula E = HZ0 which is E2
analogous to the equation V = IZ for electrical W =EH= watts per m2
Z0
circuits. (Remember that an electric field is the
result of a potential difference, and a magnetic
field results from a flow of electric current). Z0 VELOCITY, FREQUENCY,
is known as the characteristic impedance of free AND WAVELENGTH
space.
All types of electromagnetic waves move with
PHASE the same velocity in totally empty space, often
The phase gives the position of the wave in its called free space. This velocity is convention-
oscillation cycle as an angle round a circle. Its ally known as c, and its value is almost exactly
value can therefore lie between 0 and 360 300,000km per second. To bring this down to
degrees (0 to 2 radians). The phase of a wave a more familiar distance scale, EM waves
is changing continuously at any point, and travel about 300mm or one foot in one
can therefore only be defined relative to itself nanosecond in free space, and for practical
at another point, or relative to the phase of purposes in air also.
another wave at the same point. The frequency f and wavelength (lambda)
The electric and magnetic fields of an EM of any wave are related to the phase velocity v
wave are in phase with each other, so Z0 is a by:
resistive impedance whose value is 377. v = lf
POWER
The power (W) carried by an electromagnetic
2 4
INTERACTION OF RADIO
WAVES WITH MATTER
2 5
VHF/UHF PROPAGATION
between their phases. There can still be a A common misconception is that the
phase shift between them, in which case their velocity of an electromagnetic wave cannot
amplitudes will not be completely additive, possibly exceed its value in free space. As with
but if this phase shift remains the same, the most misconceptions, this is partly true. The
waves are coherent. power and information carried by an EM wave
There is a continuous range of coherence, propagates at its group velocity which never
from completely coherent, meaning that the exceeds the free-space value c. But if you could
phase relation between waves or wavelets is see the individual oscillations, you would say
completely constant (as was the case with the that the wave appears to propagate at a
radiating dipole) to completely incoherent, a velocity measured by the distance the wave-
situation in which the phase relationships are front advances in a given time. Since a
both random and varying. In this latter case, wavefront is actually a surface joining points
contributing wavelets add together in a of equal phase at any given instant, this
statistical way, so that if n equal wavelets were apparent velocity is called the phase velocity.
contributing, the result would be (n) times a During a single cycle, the wavefront advances
single wavelet, rather than n times as it would by one wavelength; hence the phase velocity
be if they were coherent. When the number of is calculated by multiplying the frequency
contributing wavelets becomes very large, the (which remains constant) by the wavelength
difference between (n) times and n times within the medium.
becomes highly significant, so the degree of If the phase delay of the re-radiated wave in
coherence is a very important factor in the propagating medium (compared with the
determining signal strengths. phase of the incident wave) is between 0 and
180, it is said to lag and the phase velocity
GROUP VELOCITY AND appears to decrease. But if the phase delay is
PHASE VELOCITY in the range 180 to 360, this is equivalent to
Although we conventionally speak of the a phase advance of 180 to 0, so the phase
velocity of light as if it were a single quan- velocity appears to increase. Fig. 2.3 illustrates
tity, two velocities are required to describe an these effects. Fig. 2.3b shows a delay of /6
electromagnetic wave propagating through (60) relative to the reference phase (Fig. 2.3a).
matter the group velocity and the phase Fig. 2.3c shows a larger delay of 5/6 (300),
velocity and these are generally different. which is equivalent to a phase advance of 60.
If you try to push a heavy object it does not The phase velocity in most materials is less
respond immediately, because of its mass or than c, but it exceeds c in metals, inside
inertia and because of restraining forces such as waveguides and importantly for propagation
friction. In the same way, their mass and the in the ionosphere. Only in empty space are
forces exerted by other charged particles will the phase and group velocities exactly identi-
prevent the electron oscillators in the propagat- cal, both being equal to c.
ing medium from immediately following the This is a difficult concept, but the distinc-
electric field driving them. Their oscillations tion between group and phase velocities is
will have a phase difference relative to the very important; failing to grasp it can lead to
incident wave, and so too will the wavelets that some bizarre misconceptions! Group velocity
they re-radiate. To an outside observer, the is the correct quantity to use when timing
wave appears to progress through the medium radar signals and moon echoes, when making
with the same frequency and amplitude, but vertical ionospheric soundings or when
with a phase that is continually being updated establishing how long it takes for signals to
by the process of energy absorption and re- arrive by an unknown propagation mode; the
radiation. The effect is to change the velocity reason is that all these imply the flow of
and hence the wavelength of the wave com- information or power. But when considering
pared with free space. wave properties such as refraction and
2 6
INTERACTION OF RADIO WAVES WITH MATTER
2 7
VHF/UHF PROPAGATION
2 8
INTERACTION OF RADIO WAVES WITH MATTER
V REFLECTION
Increasing
ionization
Any sharp change of refractive index will
density produce some reflection as well as refraction.
As well as the main refracted rays, Fig. 2.5
Path of ray
shows small reflected rays coming back out of
the surface. The amount reflected depends on
2 9
VHF/UHF PROPAGATION
2 10
INTERACTION OF RADIO WAVES WITH MATTER
oscillating and re-radiating the radio wave. If dense and electron collisions are more
some of the energy temporarily held by the frequent. If the D layer becomes highly
oscillating electrons is transferred to other ionized, for example by an intense solar flare,
particles before it is re-radiated, the propagat- it also becomes highly absorptive and pre-
ing wave will be attenuated since some of its vents HF radio waves from reaching the E or F
energy is absorbed. Usually this energy is layers above. Ionospheric absorption is much
converted into a rise in temperature of the lower at VHF and above because the probabil-
absorbing material. ity of energy loss by collision decreases as the
The attenuation in insulating materials is frequency increases.
generally small the electrons bound in atoms In poor electrical conductors the electrons
are little affected by the passing wave. Air is a can move more freely than in insulators, but
good insulator, so absorption in the lower some of the wave energy will be dissipated as
(non-ionized) atmosphere is very small at all heat by the electrical resistance. In this case,
frequencies. At 1GHz the loss in traversing the absorption increases rapidly with frequency.
standard atmosphere (page 2-13) by the Most solid natural materials contain water,
shortest path directly upwards is less than which makes them lossy conductors anyone
003dB, increasing to 023dB at the horizon unfortunate enough to have an antenna in
where the path through the atmosphere is the loft will know how signals are reduced
maximum. Even for very weak moonbounce when the roof is wet. Foliage and other green
signals, these absorption losses are negligible. vegetation are also made up largely of water.
In the upper ionosphere (the F layer) the Wood, although a fairly good insulator in a
probability of electron collisions is low, so dry state, is a lossy conductor when filled with
absorption is again low at all frequencies. sap. The absorption of a 15m thickness of
Lower down, the atmosphere becomes more various types of vegetation may be estimated
Fig. 2.8.
Approximate
absorption
due to 15m
of foliage of
various
kinds
2 11
VHF/UHF PROPAGATION
Fig. 2.9.
Diffraction,
showing the
variation in
signal
strength in
the first few
wavelengths
above and
below the
edge of the
obstacle
(not to
scale)
roughly from Fig. 2.8. The data can be scaled halved and hence the power is reduced to
for other thicknesses; multiply the attenuation one-quarter a 6dB loss.
in Fig. 2.8 by log10 (actual thickness in metres Within the shadow region the signal falls
divided by 15). off continuously, but there is not the sharp
Absorption losses also take place during cutoff which one associates with a beam of
reflection from a conducting surface. Reflec- light (actually, light does undergo diffraction
tors re-radiate radio waves because currents are at a knife-edge, but the effect is not normally
induced in their surfaces, and if they are not noticeable). At low angles just above the
perfect conductors there will be resistive losses shadow region, the signal level varies by up to
which increase with frequency. At VHF/UHF, 3dB.
poor conductors such as mountains or Diffraction is one of the mechanisms which
buildings act mainly as very lossy reflectors. can aid VHF/UHF propagation well beyond
Losses accompanying reflection from the the visible horizon [1], but the loss due to
ground depend upon its conductivity, and diffraction around an object increases with
this largely depends on its water content both the scattering angle and the frequency.
chalk hills and deserts make poor reflectors. Thus a solid obstacle which has a scarcely
On the other hand, the sea is a good electrical noticeable diffraction loss on 50 or 70MHz
conductor and therefore a good reflector. may appear as a significant barrier at 13GHz
and above, so a clear take-off becomes more
DIFFRACTION and more important at higher frequencies.
Reflection, refraction, and absorption are all
closely related, but diffraction arises from a That concludes our discussion of the basics
different mechanism (Fig. 2.9). Normally the of propagation. Now to see how they apply in
signals re-radiated during the passage of an various propagation media.
EM wave through matter will cancel in all
directions except straight ahead. If part of a
wavefront is removed by an obstacle, cancella-
tion will be incomplete and some signal will
propagate into the shadow region. In line
with the edge of the obstacle the situation is
symmetrical, with half the wavefront missing
on one side and the other half present on the
other. Not surprisingly, the signal amplitude is
2 12
THE PROPAGATING
MEDIA
2 13
VHF/UHF PROPAGATION
THE IONOSPHERE
The ionosphere forms the outer mantle of the
earths atmosphere (Fig. 2.10). It consists of a
variable composition of charged atoms (ions)
and free electrons, together with non-ionized
atoms in an overall electrically neutral mixture
known as a plasma. The ionization results
mainly from solar electromagnetic radiation
but is also caused by the impact of atomic and
sub-atomic particles of solar and cosmic-ray
origin, together with some larger particles of
matter in the form of meteors (page 2-5).
Although the total density of matter
decreases with height, the number of elec-
trons per unit volume continues to increase
up to a peak at 200400km as shown in Fig.
2.10 on the left. Beyond that height the
decreasing density takes over as the outer
atmosphere merges into interplanetary space.
Three major divisions of the ionosphere are
recognized, known as the D, E and F regions.
Within these regions are layers, although
these are not just slabs of ionization but local
maxima within the trend of increasing
ionization with height. Moreover, these layers
are not fixed but change continuously in
position and density with the time of day,
season of the year, state of the solar flux, and
the solar wind.
The movements of the ionosphere are very
Fig. 2.10. The lower and upper atmosphere, complex. In addition to wind patterns, which
showing the troposphere and the regions of
still occur even in the outermost regions of
the ionosphere. The logarithmic scale of
height compresses the upper portion of the the atmosphere, there are electromagnetic
diagram. At the same scale as the lowest forces as the charged particles move through
part, the whole diagram would be tens of the earths magnetic field, and also electro-
metres high! static forces between the charged particles.
THE D REGION
increases.) At greater altitudes the rate of The D region lies between 60 and 90km high.
decrease of N, and hence the refraction, Its main feature is that it absorbs radio waves,
becomes less and less; mathematically the particularly at HF, because the electrons are
decrease is exponential. A model based on the frequently colliding and dissipating the
averaged atmospheric conditions is known as energy of the waves instead of propagating
2 14
THE PROPAGATING MEDIA
2 15
VHF/UHF PROPAGATION
2 16
THE PROPAGATING MEDIA
2 17
VHF/UHF PROPAGATION
2 18
THE PATH FROM TRANSMITTER TO RECEIVER
power PT watts with a gain GT over isotropic in with the capture area of the antenna, because
the direction being considered, the power per the area over which an antenna of fixed gain
unit area S available at a distance R metres is collects energy depends on the wavelength.
(PT GT) divided by the surface area of a However, the properties of the space through
sphere of radius R: which the wave propagates are not frequency-
PTGT dependent.
S= watts per m2
( 4pR )
2
AN EXAMPLE PATH-LOSS
The signal power PT intercepted by a CALCULATION
receiving antenna of gain GR over isotropic is Although we said at the beginning of this
equal to the power density S multiplied by the section that free-space loss contributes in most
capture area A of the antenna. From Chapter 7 cases to the overall path loss, most types of
the capture area of an antenna with gain GR terrestrial propagation are dominated by other
over isotropic is: sources of attenuation. For example, assume
that you are listening to someone in an
G
A = R sq. wavelengths orbiting spacecraft at a distance of 1000km.
4p
How strong would the astronauts signal be
compared with that from your friend on the
G l2
A = R m2 hill across town? Lets assume that they are
4p
both using 1W hand-held 144MHz transceiv-
So the received signal power PR is simply S ers with near-isotropic antennas having 0dBi
multiplied by A: gain.
2 Your friend is line-of-sight at 10km, and on
l
Pr = GTGRPT watts your S-meter he is 60dB over the noise level.
4pR
Assuming his signal strength is governed only
To obtain the space loss, we must first by the free-space path loss, this loss is:
reduce PR to the power which would be
L1 = 20 log10 10 + 20 log10 144 + 32 45 = 95 6dB
received using isotropic antennas, which we
do by setting GT and GR equal to 1. Then the while from the astronaut it is:
free-space path loss L is simply:
L 2 = 20 log10 1000 + 20 log10 144 + 32 45 = 135 6dB
2
P 4pR
L= T = The difference is only 40dB, so the astro-
PR l
nauts 1W would be quite a respectable signal,
Of course, this formula only applies in the 20dB over the noise level over a 1000km free-
far field of the antenna, where R is much space path. In contrast, the additional losses
greater than , so L will always be a number over terrestrial paths of 1000km are so great
much greater than 1. that it would require an exceptional DX
For practical purposes it is more convenient opening to hear a 1W hand-held.
to express the path loss L in decibels and to We could have found the answer of 40dB
use kilometres rather than metres for R. Also more quickly by returning to basics. Using the
we tend to use to use frequency rather than inverse-square law, the ratio of the signals is
wavelengths. Making all these adjustments, (1000km/10km)2 = 10000, which is of course
the formula for path loss becomes: 40dB.
L = 20 log10 R + 20 log10 F + 32 45 dB PATH LOSS WITH CHANGE
where R is in km and F is in MHz. OF DIRECTION
Note that the expressions for free-space path Line-of-sight propagation is usually not very
loss involve frequency or wavelength. The interesting to DXers. All but the most local
term involving wavelength was introduced contacts (or contacts with spacecraft) will
2 19
VHF/UHF PROPAGATION
FADING
Why do signals fade? Well, obviously in some
cases because the conditions which led to
propagation have gone away. But there are
other causes of fading, often regular or quasi-
regular in character. The most common is
interference between signals arriving from the
same station by different propagation paths.
It is often an oversimplification to speak of
just one signal path, because rays travelling in
slightly different directions can arrive together
at the receiving antenna. In general they will
arrive with a difference in phase. If the
difference between two paths is in the range
from 0 to 90, the net effect is an increase in
2 20
THE PATH FROM TRANSMITTER TO RECEIVER
signal strength (constructive interference); if the bounce; refer to the sections dealing with
phase difference is in the range from 180 to these modes for further details.
90 the net effect is a decrease (destructive One of the striking features of the VHF/UHF
interference). Any change of amplitude or bands is the difference in their fading charac-
phase will change the situation, causing the teristics over both short and long paths.
net signal to rise or fall. Such changes can be Fading on different bands can be compared by
brought about by a variation in the delay listening to a multi-band beacon. 50MHz and
along the contributing paths, for example due 70MHz are both notorious for long, slow
to a change in the refractive index. Complete fading, whereas fading on 432MHz and above
cancellation between two paths occurs only can be very rapid and fluttery. Some reasons
when the amplitudes of the waves are identi- for this are advanced later in this chapter
cal and also the phase difference is exactly under Tropospheric Forward Scatter.
180. Nevertheless, to quote [1]:
Another example of multi-path interference The propagation mechanisms that cause
causing periodic fading occurs when the severe fading and phase effects on horizontal
frequency of the waves on one path is shifted and low ... angle paths are not yet well
slightly compared with the other, due to a understood.
motion of the reflecting surface (the Doppler
effect). The two waves then beat together, the
fading rate being the difference between the
two frequencies. The most common example
of this is the flutter which occurs when a
signal arrives by reflection from a moving
aircraft, as well as by the more direct path.
The flutter slows down and momentarily stops
as the aircraft passes directly overhead or
directly across the signal path at right-angles,
since at that moment the aircraft is moving
neither towards nor away from you; it then
speeds up again as the aircraft moves away.
Aircraft flutter can equally well be explained
in terms of changes in path length [2], and
this alternative view is entirely compatible
with the Doppler-shift explanation; it is
simply another way of calculating the phase
velocity.
A further mechanism for fading is the
changing polarization direction of arriving
waves, due to motion of the reflector or
changes in the medium along the path. An
example of this is Faraday-rotation fading of
moonbounce signals (page 2-62). A discussion
of how polarization affects fading can be
found on page 2-10.
Some propagation modes always involve
multiple paths and movements of the reflect-
ing surface, and are therefore intrinsically
subject to fading. These modes include
meteor-scatter, auroral scatter and moon-
2 21
VHF/UHF PROPAGATION
[1] Brackets around a band mean that although the mode of propagation may exist in principle on that band, contacts
are rare and/or extremely difficult.
2 22
TROPOSPHERIC PROPAGATION
ENHANCED TROPOSPHERIC
REFRACTION
The refractive index of the troposphere was
introduced on page 2-14, where we saw that
the radio refractive index (N) normally
decreases with height. Since refraction occurs
whenever there is a change in refractive index,
bending of radio waves around the curvature
of the earth is the norm rather than the
exception within the troposphere. In the
optical analogy, it is like an almost-permanent
mirage which enables you to see beyond the
horizon. The refractive index profile for the
Standard Atmosphere described on page 2-14
is shown in Fig. 2.16a and it extends the radio
horizon beyond the geometric horizon by
approximately one-third (Fig. 2.17a). Within
this line-of-sight plus one-third distance,
signals are almost as strong as in free space.
In practice, real atmospheric conditions
seldom approximate to the Standard Atmos-
phere. Refractive-index gradients between
dN/dh=0 and dN/dh=79 units/km are
considered normal. If dN/dh decreases more
rapidly than 79 units/km, super-refraction is
taking place (Fig. 2.16a) and the radio horizon
is extended (Fig. 2.17b). Sub-refraction condi-
2 23
VHF/UHF PROPAGATION
tions occasionally occur when the refractive UHF, where other effects take signals well
index increases with height (dN/dh is positive, beyond the refractive radio horizon.
Fig. 2.16a), bending the waves away from the A well-sited VHF/UHF station with reason-
earth so that signals become abnormally weak able equipment can make contacts over many
or disappear altogether; in effect, the radio hundreds of kilometres under average tropo
horizon becomes closer (Fig. 2.17c). conditions, even though the line-of-sight plus
To find out whether refractive tropospheric one-third rule would imply a range of only a
propagation exists between two stations, it is few tens of kilometres by normal refraction
necessary to see if the refracted rays strike the from most locations. Some books imply that
ground or encounter any other obstacles under normal conditions only simple refrac-
along the path. This can be done by graphical tion is taking place, but this clearly cannot be
ray-tracing [3, 4], which involves drawing a the case. In fact several mechanisms contrib-
vertical cross-section of the relief along the ute to this enormous extension of range,
path, using an earth radius modified to take including diffraction around obstructions and
standard refraction into account. Short-range more importantly forward tropospheric
ray tracing may be useful at microwave scatter which we deal with later.
frequencies but it is rather academic at VHF/ Similar misconceptions are sometimes found
regarding contacts over even longer distances,
which are attributed to super-refraction. This
too is inaccurate; to make the signal follow the
Fig. 2.17. Radio ray trajectories for (a) curvature of the earth, dN/dH would need to
Standard Atmosphere; (b) super-refraction; maintain just the right value over the whole
(c) sub-refraction. D marks the optical or
path, which is a very improbable state of
geometric horizon
affairs. In fact the mechanism responsible for
long-distance tropo DX is tropospheric ducting.
TROPOSPHERIC DUCTING
A duct is formed when a radio waves are
trapped between two boundaries. The atmos-
pheric refractive index does not always vary
smoothly with height; discontinuities occa-
sionally occur, forming a layer in which
refraction is much greater than elsewhere.
Such a layer will refract the wave back towards
the earths surface, forming the upper bound-
ary of a duct whose lower boundary is the
earths surface itself, and this is known as a
surface duct. Alternatively the wave may
become trapped between two discontinuities
in the troposphere, the upper one bending the
wave towards the earth, and the lower one
bending it back up again. This is known as an
elevated duct.
These discontinuities of the refractive index
are still very small, of course, so only waves
which are at glancing angles to this surface get
reflected. In this respect the duct resembles an
optical fibre, which is transparent when
viewed crossways but will trap light waves
2 24
TROPOSPHERIC PROPAGATION
launched into its ends. A duct may also be refractive index is to make it decrease more
compared to a microwave waveguide, in that a sharply than normal at the inversion bound-
duct will not propagate signals whose wave- ary (see equation on page 2-13), bending the
length is too long in relation to its vertical radio waves downwards and trapping them
depth. The minimum duct thicknesses are between the boundary and the ground.
approximately: Near-surface inversions are often formed
when the ground cools rapidly under clear
f Min. duct skies, in turn cooling the layer of air in
50MHz 6m 400m
contact with the ground. This is called a
70MHz 4m 300m
144MHz 2m 200m radiation inversion and usually results in fog or
432MHz 07m 100m mist. Surface ducts do not propagate well over
13GHz 023m 50m land, the ground being a relatively lossy
reflector covered with obstacles which absorb
Although the minimum tropospheric-duct and scatter radio waves. However, ducts occur
thickness increases with wavelength, the quite frequently over the sea, and since the
increase is not strictly proportional as it would sea (when calm) is a good reflector, sea ducts
be in a metal waveguide. The minimum can propagate over very long distances with
thickness is about 65 at 50MHz, but about little loss. For example, the 2500km path from
220 at 13GHz. The difference is due to the the Isles of Scilly to the Canary Islands (EA8) is
irregularities of refractive index which make reported to open about once a year on
the boundaries fuzzy. The upper frequency 144MHz. A particular feature known as an
propagated by a duct is determined by evaporation duct is very common over the sea,
scattering losses due to these irregularities. and is formed by the rapid change in humid-
Because the waves within a duct are con- ity which occurs just above the surface. Sea
fined to a two-dimensional layer, theoretically ducts rarely penetrate far inland, so you need
the signal only reduces linearly with the to be located either on an island or close
distance (1/d instead of 1/d2 as in free space enough to the coast to be able to beam into
page 2-18). In spite of losses due to inefficient the open end of the duct. Such ducts are often
coupling of signals into and out of the duct, only about 15m high, limiting propagation to
together with leakage and scattering caused by the microwave bands, but deeper evaporation
irregularities, the attenuation at ranges up to ducts capable of propagating lower frequen-
1000km can be comparable with the free- cies do occur occasionally.
space value and ducted signals are often
extremely strong. Propagation is obviously ELEVATED DUCTS
best between stations which are in the duct Elevated ducts form when a double disconti-
itself, or can beam directly into its open ends. nuity in refractive index occurs (Fig. 2.16c). In
this case the waves are bent upwards from the
SURFACE DUCTS lower boundary of the duct, and downwards
Surface ducts occur when a steep negative from the upper boundary. Thus the waves can
gradient in refractive index forms immediately be guided over irregular terrain until the duct
above the ground or sea (Fig. 2.16b), trapping is broken by mountains or other changes that
the radio waves by refraction from above and the duct cannot follow. Elevated ducts form at
by reflection from the surface below. In a heights typically between 450 and 2000m,
normal atmosphere, the air temperature and and are the origin of the longest overland
dew point and hence the refractive index tropo openings. Even longer distances can be
all decrease with height above ground. When worked via elevated ducts over the sea.
the temperature increases with height instead Although ducts approaching 400m deep are
of decreasing in the normal way, this is called occasionally formed over warm seas, ducts
a temperature inversion. The effect on the radio over land in temperate climates are generally
2 25
VHF/UHF PROPAGATION
much shallower. This is borne out by the between specific areas. This is occasionally so
observation that while tropo ducting may marked that the opening resembles sporadic-E
occur many times a year on the 144MHz propagation. However, a sporadic-E cloud will
band, there are very few records of tropo propagate signals in many directions, while
ducting at 70MHz or lower frequencies in the tropo ducts tend to occur in specific direc-
UK. It also explains why strong ducted signals tions. Also, ducts can be very stable, lasting for
can be found on the 432MHz or microwave hours or even days, in contrast to the more
bands at times when ducting on 144MHz is transient behaviour of sporadic-E. Although
much weaker or even absent. less well-developed leaky ducts do not
Well-developed elevated ducts produce a produce the strongest signals, their imperfect
skip effect which carries the signals over the boundaries do permit stations at intermediate
heads of stations beneath the duct. Ducted distances to couple in, thus providing a wider
propagation thus tends to be strongest geographical spread of DX.
2 26
TROPOSPHERIC PROPAGATION
2 27
VHF/UHF PROPAGATION
refraction can occur during the passage of points, this is not a practical forecasting tool
either type of front, but those caused by cold for radio amateurs because the data are not
fronts are more marked. Enhancements occur available soon enough. As an aid to catching
because the incoming cold, dry air is more tropo openings, a barometer is a must a
dense than the warmer air ahead of the front recording type, if you can afford it. National
and tends to undercut it. The warm air is and continental weather maps are readily
therefore driven upwards, creating a tempera- available from TV and newspapers, and even
ture inversion along the line of the front. So more immediate information is available from
the best direction of propagation is along the weather satellites or HF facsimile broadcasts
front as it passes over your location, but the for quite a small investment in equipment.
opening will inevitably be short-lived. Look for a large, stable high-pressure area with
Forecasting tropo openings is no easier than light winds.
forecasting the weather itself. While it is Direct observation of the sky wind, clouds,
possible to construct vertical atmospheric transparency can give a lot of information
profiles from weather observations at specific about your local tropo situation. Inversion
layers can sometimes be seen as a dark line of
trapped smoke just above the horizon. In
Fig. 2.19. DX operators maximum ranges by anticyclonic conditions a layer of
tropospheric ducting, taken from the Top stratocumulus with a base below 1000m may
List in DUBUS 1989/1 indicate the presence of a surface duct, while a
base above 1000m could suggest an elevated
duct.
Finally, dont ignore the messages given on
broadcast TV and VHF radio apologising for
interference. This is usually caused by strong
tropo signals from other countries bad news
for TV viewers but an opportunity for us!
MAXIMUM DISTANCE
SPANNED BY TROPO DUCTS
Fig. 2.19 shows the maximum tropo distances
worked by operators reporting in DUBUS 1989
for the 144, 432, and 1296MHz bands. The
maximum range appears to fall off with
frequency, but to some extent this may reflect
reduced band occupancy and path-loss
capability on the higher bands. From the UK,
tropo paths to the north and to Scandinavia
are much rarer than towards Spain, southern
France, Italy, Switzerland, Austria, and
Czechoslovakia. Mountain ranges often limit
the distance workable from the UK to around
1000km. On the other hand, sea ducts are
limited only by the scale of the weather
pattern. The 4000km path from California to
Hawaii is regularly spanned by sea ducts,
producing world-record tropo DX. During an
opening between the Canary Islands (EA8)
and the British Isles and Norway on 8/9
2 28
TROPOSPHERIC PROPAGATION
2 29
VHF/UHF PROPAGATION
2 30
AURORAL SCATTERING
2 31
VHF/UHF PROPAGATION
Fig. 2.21. Solar interactions with the earths magnetic field (not to scale). The earths
magnetic field is pushed away from the day side by the solar wind. Note the magneto-tail
containing trapped charged particles, extending away from the earth on the night side
of these electrons modulates the scattered can occur well away from the position of the
wave, producing a characteristic hiss caused original aurora.
by random Doppler shift. So SSB is difficult to
read and you need to speak slowly and clearly ORIGINS OF AURORA
(Chapter 3). This technique works well on 50 As befits such a wondrous and strange phe-
and 70MHz but the Doppler shift increases in nomenon, the explanation of aurora is not
direct proportion to the frequency. CW gives simple. It is still the subject of research and I
far better communication than SSB for auroral can only give a rough outline here. The
contacts on 144MHz, and is almost essential electrons which cause the aurora originate
at 432MHz. In addition there may be a shift in from the variable streams of atomic particles
the centre frequency of the signal, due to the which are continuously emitted from the sun.
collective motion of the secondary electrons This solar wind is composed mainly of elec-
which form currents around the hemisphere, trons and protons (hydrogen nuclei) with a
so dont be surprised if you need quite a large small proportion of highly ionized atoms of
RIT shift in order to resolve the other station the elements that make up the sun (chiefly
especially on 432MHz. helium), and neutral atoms. Together these
The visible and radio manifestations of an form a plasma which is electrically neutral
aurora may not always be evident at the same overall.
time. Visual effects range from 400km in During their journey from the sun towards
height down to 100km, whilst radio reflec- the earth, the charged particles are influenced
tions are limited to the bottom of this range. by the interplanetary magnetic field. This applies
Thus an aurora can be visible while its radio- particularly to the electrons as a result of their
reflecting part is below the horizon. Some- much smaller mass. The path of a moving
times, however, a radio aurora is observed electron is bent when it travels through a
without visual effects being reported. Apart magnetic field because it experiences a force
from the obvious reason of clouds or daylight which is at right-angles to the field and also to
preventing visibility, ionospheric currents the direction of motion. This is familiar as the
transport the secondary electrons around the force which makes an electric motor work, or
earth, so that scattering from these currents moves the electron beam in a TV tube. As a
2 32
AURORAL SCATTERING
result, the charged particles of the solar wind radio auroral effects are found; the details of
travel in spirals along the field lines linking the the acceleration mechanism are still the
sun and the earth. But that is not the whole subject of debate and research.
story moving charged particles also generate It is important to understand the implica-
their own magnetic fields which modify those tions of what I have just described. The direct
already present. The effect is to distort the pre- cause of an aurora is not solar activity. It is
existing field lines as if the solar wind was some event which triggers a change of configu-
exerting a pressure upon both the interplan- ration in the magneto-tail. If we wish to detect
etary magnetic field and the field of the earth. auroras or better still, be warned in advance
What causes the interplanetary magnetic we need to look for these triggering events.
field in the first place? It is dominated by the
field of the sun; but unlike the earths mag- THE WHEN AND WHERE
netic field which resembles that of a two-pole OF AURORAS
bar magnet and is quite steady, the suns field The field lines from the magneto-tail connect
is more complex and changing. The sun has with the earth to form the auroral ovals
several magnetic poles called coronal holes around the north and south magnetic poles.
where the field lines leave the sun and The northern oval rotates around the mag-
connect into interplanetary space, and these netic north pole (presently located near Thule
are also where charged particles can escape. in NW Greenland, although its position is
The suns magnetic field therefore switches slowly moving) and the accelerated electrons
direction several times per rotation as a and other particles are guided along these field
number of coronal holes pass by. These lines to create an aurora. The pressure of the
changes take place at sector boundaries; within solar wind pushes the field lines towards the
each sector, the suns magnetic field is in the dark hemisphere, so night-time auroras
same direction. generally extend further south than those
For a long time it was thought that auroras occurring in the daytime.
were caused directly by electron streams The daily rotation of the magnetic pole
originating from cataclysmic events on the about the geographic pole combines with the
sun such as sunspots, flares or solar solar-wind effect to make the auroral zone
prominences. From satellite measurements of approach and recede twice a day (Fig. 2.22). As
the solar particle flux, we now know that this seen from the magnetic latitude of the UK, the
is not correct but these obvious ideas can oval comes close around 1800 local time (Fig.
take a long time to disappear! 2.22d) and then slightly recedes before
The view now accepted is that charged approaching even closer around local mid-
particles continually emanating from the sun night (Fig. 2.22a). This gives two times of day
in the solar wind become trapped in the when aurora is most likely to occur, as can be
earths magneto-tail, which is formed by field seen in Fig. 2.23.
lines which stream from the earth away from Fig. 2.23 shows another phenomenon.
the sun (Fig. 2.21). An increase in the solar There is a dip in the probability of finding a
particle flux stretches the magneto-tail until it radio aurora at local magnetic midnight, ie
snaps and reconnects into a new and more when the longitude of the magnetic pole and
stable configuration. When this happens some the sun differ by 180. In the UK this falls
of the trapped particles are ejected out into between 2200 and 2300 UTC. While the aurora
space, while those in the retained part of the is taking place, two huge electron currents
magneto-tail are propelled towards the earth (kiloamps) flow away east and west from the
by the contracting magnetic field lines. point nearest the sun; but the currents cannot
However, this alone is not sufficient to meet on the night side of the auroral oval
accelerate the particles to the energies neces- because of the repulsion between like electric
sary to penetrate to the E region where the charges. Instead both currents turn away
2 33
VHF/UHF PROPAGATION
N N
a b
N N
Fig. 2.22. The position of an auroral oval at six-hourly intervals during the day. The geomagnetic
north pole (shown by a dot) rotates about the geographic north pole in the centre of each picture,
and the direction of the solar wind is shown by an arrow. The result is that the auroral oval
approaches and recedes twice a day. Rings show the accessible range from London and Edinburgh
2 34
AURORAL SCATTERING
towards the equator, leaving a relatively quiet show this reversal [9].
region of the ionosphere around magnetic There is also an annual variation of the
midnight. This is the Harang discontinuity, position of the oval. In the northern winter
named after the Norwegian scientist who first the magnetic pole is on average tilted further
described it. From your viewpoint on the earth, from the direction of the sun, so the oval
the ionospheric current is flowing from the east moves to lower latitudes. This results in peaks
before the discontinuity but afterwards from in auroral activity around the equinoxes,
the west, and a magnetometer recording the although Fig. 2.24 shows that auroras can
deflection of the earths magnetic field will occur at any time throughout the year.
The frequency with which auroras occur
decreases very rapidly with distance south-
Fig. 2.23. Occurrence of radio aurora in the wards from the auroral oval. Hence there is a
UK by time of day, showing peaks in late very large difference between the northern
afternoon and around local midnight, and the
Harang discontinuity around 2200-2300h and southern UK. As can be seen from Fig.
2.22, at the optimum approach of the auroral
oval (around midnight) stations in Edinburgh
can access the auroral ionization when it is
below the horizon from London.
2 35
VHF/UHF PROPAGATION
Fig. 2.25. Isoflectional map for a station in London. The contour marked 0 joins locations
where the auroral reflection would be most favourable, if ionization is present at E-layer
heights. The same map would apply to field-aligned ionization
favourable reflection geometry can often be London. For E-layer heights of around 100km,
heard from southern England in relatively only the region closer than 1100-1200km is
weak auroras which provide contacts with within range. Notice that you can beam
Scotland and Northern Ireland. Some stations north-east along these contours towards an
will regularly be much stronger than others area over southern Scandinavia where the
who are using higher power but are less contours are widely separated. This means
favourably situated. that there is a large area over which the field
To explore these effects, in 1979 I developed alignment is favourable, and most auroral
a computer program for the Scientific Studies contacts from the southern UK are indeed
Committee of the RSGB (now renamed the made with the beam in this direction. If the
Propagation Studies Committee). At certain aurora comes far enough south, into the
points in the E layer, a radio beam originating region across northern England where the
from your station will intersect the earths contours bunch together, even local stations
magnetic field at right-angles. Contours at similar latitudes will become strongly
joining these favoured locations can thus be auroral. Similar isoflectional maps can be
drawn on maps, which I named isoflectional constructed for other locations, but the shape
maps [10]. The computer program plots these of the contours will be different.
contour lines as seen from a specified station Note that these maps only consider the
location. If a similar map for another station is geometry for favourable auroral reflections.
superimposed, the intersection of the con- They say nothing about the location of the
tours shows the preferred reflection region, auroral ionization or even whether there is
and hence the directions towards which the any! Very often the aurora will not be suffi-
antennas should be pointed for these to ciently extensive or come far enough south to
stations to effect a contact. As an example, Fig. provide any significant ionization at the
2.25 shows this plot for a station located in optimum locations. In such cases the signals
2 36
AURORAL SCATTERING
2 37
VHF/UHF PROPAGATION
2 38
SPORADIC-E
2 39
VHF/UHF PROPAGATION
on this subject I will not deal with equatorial- Wallops Island, North Carolina during 1976
zone Es any further. for days when it exceeded 4MHz. Most of
these days occur between May and August;
TEMPERATE-ZONE Es the rest of the year is represented only by the
FREQUENCY BEHAVIOUR small hatched area. The data from Wallops
The apparent sudden onset of an Es opening is Island were used because that particular
largely due to our habit of listening on one ionosonde took samples every 15 minutes
frequency band. By monitoring the frequency during that year, unlike most ionosondes
range above 28MHz you can follow the which sample at hourly intervals. Even so, the
upward trend of the Es MUF as the ionization data may still be pessimistic excursions to
intensity increases. However, the further the even higher frequencies may occur between
MUF rises, the less likely it is to continue to sampling periods, and the ionosonde only
rise. Fig. 2.29 plots the highest E-layer critical measures the situation immediately overhead
frequency (f0; see page 2-16) achieved at while more intense Es may be occurring
elsewhere.
Fig. 2.29 shows a rapid tail-off of Es prob-
Fig. 2.29. Critical frequencies f0E measured ability as the frequency increases; re-plotting
above Wallops Island (North Carolina, USA) the data on a logarithmic scale suggests that
during 1976
the smoothed probability curve is heading
towards a frequency limit. This implies an
upper limit to the ionization density that can
be produced by the Es mechanism, as the
observations confirm.
PATTERNS IN
TEMPERATE-ZONE Es
Despite the label sporadic, which implies
that its appearance is irregular and unpredict-
able, there are predictable patterns in the
occurrence of Es. It is primarily a midsummer
phenomenon, occurring in the in the north-
ern and southern hemispheres alternately.
The times of year at which Es openings have
occurred at 60MHz and 144MHz in north-
west Europe are shown in Fig. 2.30, together
with the times of reception of a 28MHz
beacon for comparison. (The 60MHz data are
reasonably representative of 50MHz.) All
frequencies show a similar peak between early
May and the end of August, the maximum
occurring during the first week in June.
Indeed, some people say that openings on
144MHz from the UK are almost guaranteed
within a few days around this peak though
the years 1989, 1990, and 1991 showed that
this is not necessarily the case. Clearly the Es
season varies greatly from year to year, just
like the British summer!
Claims have also been made that the
2 40
SPORADIC-E
2 41
VHF/UHF PROPAGATION
Fig. 2.32.
Two almost
identical Es
openings on
successive
days. Dots
indicate the
region
where
reflection
takes place,
near the
midpoints
of the paths
lar day, there is an enhancement of about than the atmospheric gases which form the
30% in its probability of occurring the next regular E and F layers. These metal atoms need
day over the same path, and an enhancement much less energy to become ionized than do
of about 10% in its probability of occurring on the atmospheric gases, and they also take
the third day [15]. If we make the analogy of a longer to recombine. Es layers are often seen
capacitor discharging through a resistor, this to descend from about 120km high to about
would imply a decay time-constant of one 100km during the day. Often two or more
day. At present, however, it is only a guess thin layers (less than 1km thick) form at
that similar probabilities might apply at VHF. particular heights. The ionization within these
layers is not uniform, and if you could look at
SPORADIC-E CLOUDS the E layer at a particular radio frequency you
Investigations by rocket probes, radar and would see discrete clouds which appear to
ionosondes reveal the following picture. Es get smaller as the frequency increases. At
layers contain a high concentration of ionized 144MHz you would see only the highest
metals such as iron and magnesium, rather concentrations of ionization measuring
Fig. 2.33.
Wedge-
shaped
footprint
from a
reflecting Es
cloud
2 42
SPORADIC-E
Fig. 2.34.
DX
operators
maximum
ranges by
Es, taken
from the
Top List in
DUBUS
1989/1
perhaps tens of metres across, and this is what opening, it may be that the ionization which
makes 144MHz Es so geographically selective. had become concentrated into Es layers is
Since signals at the Maximum Usable beginning to spread out vertically again,
Frequency are bent through the smallest providing a near-vertical reflecting surface
angles, they will provide the longest-distance which may also become field-aligned. If so,
propagation. For this they use clouds on the this back-reflection may be very similar to the
far horizon (Fig. 2.12b). A distant blob of post-Es type of FAI propagation (page 2-48).
high ionization will appear almost as a point
when viewed from the antenna, and the MAXIMUM Es DISTANCES
refracted wave will take the shape of a cone. The maximum geometric range corresponding
The intersection of this cone with the ground to E-layer heights of 90 to 120km is 2130 to
gives a wedge-shaped footprint of propaga- 2450km (page 2-15). For comparison, Fig. 2.34
tion which is familiar from Es openings (Fig. shows the maximum distances on 144MHz
2.33). These clouds of ionization often move claimed by operators in the DUBUS DX lists.
or disperse quite rapidly, so contacts need to In good agreement with prediction, the
be made quickly! distribution drops rapidly at 2400-2500km.
Some extension beyond this limit may be
SIDE AND BACK REFLECTIONS credibly explained by good sites with distant
Occasionally Es contacts are reported with horizons, with a little help from tropospheric
large deviations of antenna direction from the refraction. However, this does not account for
great-circle direction. Most often these occur the significant tail of contacts extending to a
on the lower-frequency VHF bands, particu- minor peak at 3300250km, which is not seen
larly 50MHz. This is because the bending in other E-layer modes such as aurora or
angle at lower frequencies is larger for a given meteor-scatter at this frequency. The DUBUS
ionization density, so the angle of the re- lists and Fig. 2.34 represent peoples best DX,
fracted cone of rays covers a wider region on while Fig. 2.35 shows the distances of ordi-
the ground. See also the section on iono- nary Es contacts sampled randomly from
spheric tilts on page 2-51. reports of major 144MHz openings during
Sometimes both antennas need to be 1987 and 1988. The most probable distance
pointing in roughly the same direction, and worked is between 1400 and 1900km, so the
for this type of back-reflection we need a 3000km+ peak in Fig. 2.34 clearly represents
different explanation. Since these conditions double-hop Es. There are also a few examples
seem to occur towards the end of a long Es of double-hop contacts (eg [17]) at times when
2 43
VHF/UHF PROPAGATION
MULTI-HOP Es
The time of year and day when Es is most
likely to occur depends on your geographic
Fig. 2.36.
Derivation of position. The best time of year is clearly the
the most summer, regardless of whether you live in the
probable northern or southern hemisphere. Fig. 2.31
time of day showed the most likely times of day, which
for were given in terms of local time. In terms of
transatlantic Standard Time (GMT or UTC), the plots must
50MHz Es
(curves), and be shifted forward by one hour for every 15
actual times longitude east of the Greenwich meridian, or
of contacts back an hour for every 15 west. The depend-
(histograms). ence upon local time shows that temperate-
See text for zone Es takes its daily timing from the
details rotation of the earth, and therefore that solar
radiation probably is responsible for generat-
ing the ionization. However, since Es occurs
sporadically rather than every day, there must
be other factors involved too as discussed in
the next section.
Multi-hop paths are likely to be formed from
hops of the most common distance, ie
1650250km. By dividing the great-circle path
into segments of this length, you can find the
most probable number of hops. Read off the
latitude of the midpoint of each hop, and shift
the characteristic daily frequency curve of Fig.
2.31 to the corresponding times in GMT.
Finally, multiply each point on these curves
together to produce a curve of the overall
probability. In practice you need only work out
enough points to sketch in the curve [18].
An example prediction is shown in Fig. 2.36
for a two-hop 50MHz Es path from the UK to
2 44
SPORADIC-E
the eastern seaboard of the USA. Figs 2.36a generally agreed that meteors provide the
and b are the smoothed hourly distributions, metallic raw material for Es, they clearly arent
moved westwards (ie later in the day) to the the whole story.
required positions for the two hops, and (c) is Correlations between Es and many other
the combined probability curve. Compare this factors have been claimed. Lets consider a few
prediction with the actual hourly distributions which have received the most publicity. Firstly,
of transatlantic Es openings: Fig. 2.36d gives the solar cycle; this is known to have little
data for 1987, while Fig. 2.36e gives the effect on the E-layer MUF. The radiation
cumulative data for 1982-85. The agreement is absorbed in the E layer is primarily ultra-violet,
good, showing that middle to late evening in which like visible light is almost independent
the UK is the most hopeful time. A similar of the solar cycle. A recent analysis of many
prediction for the even longer path from the years of data claimed a very weak dependence
FY7THF beacon shows equally good agree- of VHF Es on the twelve-month mean sunspot
ment with the actual times of reception, and number [19]. However, my own analysis of
looking in the other direction suggests late openings on 144MHz between 1972 and 1991
afternoon as the best time for the double-hop suggests a weak negative correlation with
path to Cyprus and the Middle East [18]. sunspot numbers; the probable reason for this
is explained in the next section.
CAUSES OF TEMPERATE-ZONE There have been suggestions that Es forms
Es FACTS AND FICTION preferentially over mountains. If you look at a
There have been many suggestions about how relief map of Europe, mountains such as the
Es is formed, but not all of them stand up to Vosges, Eifel and Jura can indeed be found at
scientific analysis. In this section I will try to around half the most probable Es distance
sort out the facts from the fiction. from central England. For longer paths, which
As mentioned earlier, Es layers are observed maybe deserve higher mountains, the Alps
to contain high concentrations of ionized and Pyrnes together with their lower
metals. These relatively heavy atoms tend to extensions form an almost continuous arc
gravitate towards the earth, and it is supposed covering all popular directions from the UK.
they are being continuously replaced by debris However, this is rather a parochial view, and
from meteors, many of which have a high perusal of DUBUS magazine shows spectacular
metallic content. From this it has been 144MHz Es contacts with no land above 200m
suggested that the level of Es follows the anywhere near the path midpoint! Es on
annual variation of meteor rate, which peaks 50MHz certainly does not require high
around the same period as Es in the northern ground, as many transatlantic contacts have
hemisphere. But the Es season in the southern shown. If mountains do have any effect on Es
hemisphere is in their summer too, which is it is clearly rather small.
six months out of phase with the peak of the What about thunderstorms? This idea looks
world-wide meteor rate. Hence it is not promising, because it would explain why
surprising that analyses of meteor and Es data mountainous regions might be favoured
have failed to reveal a convincing correlation. because storms tend to develop there during
Another apparent connection between the summer months while not ruling out
meteors and Es is the enhanced rate of meteor the possibility of Es over lower-lying terrain.
reflections observed on 144MHz when the Es Furthermore there is a mechanism to explain
is approaching the critical frequency for that the connection, namely that strong convec-
band. The explanation is simple the pre- tion cells (cumulonimbus clouds) can develop
existing Es ionization simply adds to that high upward velocities which break through
arising from the burn-up of meteors at E-layer the tropopause at about 12km, causing gravity
heights, producing an enhanced rate of waves which propagate up to 100km altitude,
audible meteor bursts. So although it is where they form standing waves which could
2 45
VHF/UHF PROPAGATION
2 46
SPORADIC-E
both the earths magnetic field and the wind good. The very fact that intense Es is sporadic
direction. The force producing this drift is the suggests that it requires several unrelated
same as that which deflects the electrons in factors to be present, all at the same time.
your TV tube to scan the screen, and the If any one of them is missing, you will
direction of drift will reverse if the wind search the bands in vain.
direction is reversed. The drift in the vertical
direction is determined by the horizontal AURORAL Es
component of the earths magnetic field. If Auroral-zone sporadic-E differs from the more
the wind moves with different velocity at familiar temperate-zone Es by being generally
different heights (the technical definition of found at higher latitudes, at night as well as
shear), electrons at different heights will drift during the day, and at other times besides the
at different speeds, and hence will bunch into summer. This is because the ionization
layers. If the wind direction actually reverses, originates from incoming auroral particles
the layer will be stabilized at the height of the rather than solar ultra-violet radiation, hence
reversal. the time and place of auroral Es tends to
This wind-shear mechanism probably follow that of the aurora. Usually auroral Es is
applies to all types of Es as a means of concen- formed from the ionization remaining after an
trating the ionization into layers. The mecha- auroral storm and its associated geomagnetic
nism clearly operates best under stable disturbance have subsided, though it can
geomagnetic conditions and will be upset by precede an aurora if sufficient ionization is
auroral activity; hence the negative correla- already present from particle precipitation.
tion with sunspot number mentioned earlier. The mechanism which concentrates the ions
Auroral Es (see later) normally occurs immedi- into a layer sufficiently dense to reflect VHF is
ately before or after an auroral storm when probably wind shear, as described above for
the magnetic field is not so highly disturbed, temperate-zone Es.
and is facilitated by the massive ionization Although auroral Es is much more common
spread throughout the E region at such times. at more northerly latitudes, after a big storm
However, wind shear at 100km is difficult to the ionisation will spread southwards and
detect and measure, and its relationship with becomes accessible at mid-latitudes. For
more readily observable objects or events has example, during the huge aurora of 13 March
been the subject of much speculation. 1989, 50MHz auroral-E signals from Finland
To summarize, the best scientific authorities appeared in the UK as the auroral oval was
even including the inventor of the wind- still expanding and passing overhead. This
shear theory still do not have a complete aurora spread so far south that intense auroral
theory for the formation of Es [21]. We know Es formed in traditional temperate-zone Es
that the raw material to be ionized is provided territory, giving almost T9 signals on 144MHz
by meteors; and that the energy for ionization from Italy, Yugoslavia and Bulgaria and the
comes directly from the sun in the case of surrounding area as late as 0200 hours! On the
temperate-zone Es, and from auroral particles same day, very similar auroral Es was observed
in the case of auroral Es. These long-lived ions in the USA [22].
are concentrated into irregular thin layers by
wind shear combined with the earths mag-
netic field. But we do not yet understand
these factors well enough to predict the
occurrence of Es. The weather at E-layer
height bears little relation to the more readily
observable conditions down here in the
troposphere, so the prospects for amateurs to
predict Es from ordinary weather maps are not
2 47
VHF/UHF PROPAGATION
Fig. 2.38. Isoflectional map for a station in Geneva. Comparing this with Fig. 2.25, the
prospects for FAI contacts between Geneva and London seem unfavourable
2 48
F2 PROPAGATION
2 49
VHF/UHF PROPAGATION
Fig. 2.39.
Widely
scattered
points show
that the
correlation
between
sunspot
number and
solar flux is
only
approximate
MEASUREMENTS OF SIGNIFICANCE OF
SOLAR ACTIVITY SOLAR ACTIVITY
The traditional measure of solar activity is the Contrary to continuing popular belief, the
sunspot number (SSN) which is discussed in day-to-day MUF does not follow solar activity
more detail in the next section. However, a up and down. The ionosphere acts like a
better measure of solar activity as it affects the capacitor, taking time to charge up and also
ionosphere is the solar flux unit (SFU). This is a time to decay. The capacitor analogy can be
simple measurement which is immediately taken further, since the rate of production or
available, unlike the SSN which requires recombination of the ions depends on the
processing. The solar flux is measured at number already present; this results in an
several frequencies, the value at 2800MHz exponential charge or decay curve, just like a
being considered the best measure. There is a capacitor. The value of SSN or flux which is
correlation between the SSN and the flux (Fig. normally used in programs to predict HF
2.39), but the widely scattered points show propagation is in fact a mean value averaged
that this is an approximate relationship and over several months; three, twelve and
not the simple curve shown in Fig. 2.39, thirteen months are popular periods.
which is often assumed in home-computer Rather than being beneficial, sudden in-
MUF predictions. creases in solar activity can produce deleterious
Current solar-flux and magnetic-disturbance effects such as sudden ionospheric disturbances
data can be obtained by phoning the Geo- (SIDs), polar-cap absorption events (PCAs) and
physical Data Centre at Boulder, Colorado fade-outs. The SID and PCA have little effect at
directly on (010) 1 303 497 3235. The infor- VHF, but the fade-out does indicate a decrease
mation from Boulder is also broadcast on the in the MUF. Sudden increases in solar activity
WWV standard-frequency transmissions, and can also trigger aurora (page 2-37).
are usually available on the DX packet-cluster
network. The RSGB plans to broadcast solar SEASONAL VARIATIONS IN
and geophysical data several times a day from THE F2 LAYER
the UK on a frequency just above 38MHz. Although the ionization in the F2 layer is
2 50
F2 PROPAGATION
caused by solar radiation, the maximum MUF. In fact, the daytime increase in MUF
electron density is not found at latitudes results partly from the pressure of the solar
where the sun is directly overhead but in wind on the ionosphere, which results in a
regions 1015 north and south of the mag- lowering and compression of the F region.
netic equator. The MUF decreases either side This produces a higher ionization density and
of the maximum by about 0.8MHz per degree also a more favourable geometry for reflection
of latitude, as stations at higher latitudes (Figs 2.21 and 2.40).
know to their cost. These regions of maximum The MUF may show a peak considerably
MUF do not follow the seasonal movement of before or after local noon. Sometimes two
the sun north and south of the geographic daily peaks occur, as with sporadic-E (page 2-
equator, but merely change in relative inten- 41) but closer together. Statistically, however,
sity. The F2 MUF peaks around the equinoxes the paths with the highest MUF are still most
when the two regions are equally illuminated, likely to be those at which local noon occurs
rather than at mid-summer in either hemi- at the midpoint. In other words, look for
sphere. When approaching the solar maxi- openings to Australasia and the Far East just
mum, the later (autumn) equinox gives higher after dawn, Africa around mid-day, and the
MUFs, while during the decline towards the Caribbean and then the USA in the afternoon
solar minimum the spring equinox is better. and evening. If in doubt, point your beam
towards the sun!
DAILY VARIATIONS IN THE
F2 LAYER TILTS, CHORDAL HOPS, AND
The largest change in MUF is its daily varia- NON-GREAT-CIRCLE PATHS
tion. However, we know that this cannot be At dawn and dusk, when the change in height
entirely due to the increased ionization of the electron concentration is taking place,
around the sub-solar point, because changes the ionosphere is tilted from east to west.
in solar flux on a timescale of hours do not This tilt can launch chordal hops involving two
result in a significant increase in the global or more ionospheric reflections but no
intermediate ground reflection (Fig. 2.40). The
MUF for chordal hops is higher by virtue of
Fig. 2.40. Effect of F2 layer tilt in launching
chordal hops which involve no intermediate the shallower angles of incidence and there
ground reflection (not to scale) are no ground-reflection losses, so the chordal
hop is probably the favoured mode of long-
distance F2 propagation on 50MHz. For a
given number of ionospheric reflections, a
chordal hop covers no greater distance; for
example, two maximum-distance conven-
tional hops joined by a very shallow ground
reflection will cover essentially the same
distance as a maximum-distance chordal hop
passing just above the ground.
Chordal hops are not confined to twilight
propagation, however. Every change in
electron density with latitude or longitude can
be thought of as a tilt, in that it changes the
effective height of the layer. In this sense
sporadic-E clouds are full of tilts due to the
lumpy nature of the ionization. Also the F
layer changes density with latitude, as de-
scribed above. These effects are important for
2 51
VHF/UHF PROPAGATION
2 52
TRANS-EQUATORIAL PROPAGATION
2 53
VHF/UHF PROPAGATION
2 54
IONOSPHERIC SCATTERING
2 55
VHF/UHF PROPAGATION
2 56
METEORSCATTER
Notes
ZHR indicates zenithal hourly rate of meteors.
Favoured directions and times are valid for latitudes around 50N. Local times are valid for any longitude. For optimum
reflection geometry on shorter paths, both stations should offset their beams by 5-15 towards the direction shown in
brackets.
Table updated and revised from the RSGB Operating Manual, using additional data from the BAA Handbook 1990 (1989)
and Meteor Showers: A Descriptive Catalog by G W Kronk (Enslow 1988).
Many minor showers are also listed in these sources and may be worth trying.
2 57
VHF/UHF PROPAGATION
2 58
METEORSCATTER
DISTANCES BY MS
Distances worked by MS are what one would
expect from E-layer reflection. On 50 and
70MHz the majority of returns come from
between 88 and 100km high, corresponding Reflecting signals off the moon normally
to a maximum range between 2100 and called EME (earth-moon-earth) or moon-
2250km. Unfortunately, 70MHz contacts of bounce has always been considered a rather
those distances from the UK must generally be esoteric mode, because it is only just possible
crossband. A plot of maximum claimed using amateur equipment. However, EME is
distances on 144MHz reported in DUBUS one of the most reliable modes of DX propa-
implies a similar reflection height, though gation and is the only one to give world-wide
there are a few claims extending to 2800km coverage on the higher VHF/UHF bands
(Fig. 2.45). It can be argued that the maxi- without using satellite repeaters. It is also one
mum range should be greater on the lower of the few modes in which the path loss is
bands, since they can use the less dense trails fairly well known in advance. When the
commencing at greater heights. moon is closest to the earth the round-trip
Not shown in Fig. 2.45 is one isolated claim path loss is 2515dB at 144MHz, 261dB at
of a distance of 3200km being worked from 432MHz, and 2705dB at 1296MHz. From this
Europe, and there are a few similar examples information and the station parameters, the
from the USA. The propagation mode for expected signal/noise ratio can be estimated.
these extremely long distances remains In practice there are considerable fluctuations
unexplained. Double-hop MS does not fit the in signal strength (mostly downwards!) which
evidence; nor does single-hop using meteors we describe below.
burning up at extreme heights. The most
likely explanation seems to be single-hop MS TRACKING THE MOON
assisted by some other mode. Obviously the moon must be above the
horizon for both stations, although atmos-
pheric refraction may mean that signals may
still be heard with the moon a few degrees
below the geometric horizon. You then have
to locate the moon, point your antenna at it
and follow its motion through the sky.
The traditional method of locating the
position of the moon you certainly cant rely
on seeing it from the UK is to use the
Nautical Almanac. The moons positions are
given in Celestial co-ordinates used by
astronomers, known as Right Ascension (or
Hour Angle) and Declination. These have to be
converted to the earth-based co-ordinates of
azimuth and elevation for your particular
2 59
VHF/UHF PROPAGATION
Fig. 2.46. Equivalent noise temperature of the sky at 136MHz, in kelvins. Principal feature is the Milky
Way, which appears S-shaped owing to the map projection used
2 60
MOONBOUNCE
location an ideal task for a personal compu- BEST TIMES FOR EME
ter, which can also be used to predict the Since EME signals will always be very weak, all
Right Ascension and Declination so you dont possible steps must be taken to reduce noise,
actually need the Almanac at all. As well as both external and system-generated. Night-
telling you the moons azimuth and elevation time is best because sun noise will be absent
at suitable time intervals, a good moon- and man-made noise reduced. This implies a
tracking program will also calculate the times full moon, rather than a new moon with the
when you have a mutual moon window with sun nearby. Apart from this, the phase of the
any chosen part of the world, the level of moon is not important the entire moon is
cosmic background noise and the Doppler still there, even if you cant see it all.
shifts caused by the relative motion of the The other major noise source is the galactic
earth and the moon [26]. background which follows the contours of our
The dominant motion of the moon, as seen galaxy, as shown in Fig. 2.46 [27]. Although
from the earth, is due to the rotation of the the Milky Way is roughly planar, it appears S-
earth itself. One rotation in 24 hours amounts shaped on the diagram because of the projec-
to 25 of arc every 10 minutes, with a small tion used to map the celestial sphere onto the
correction for the moons own orbital motion. page. The noise temperatures in Fig. 2.46 are
The angular diameter of the moon seen from at 136MHz; multiply all values by about 087
the earth is about 05, which is far less than for 144MHz. The multiplying factor for
the beamwidths of most amateur antennas so 432MHz is about 006, though reference [27]
it is generally sufficient to make aiming also includes a 400MHz map which would be
corrections every 510 minutes at most. a more accurate starting point. During the
At its the highest elevation, midway course of a lunar month, the moon drifts
between rising and setting, the moon moves across the celestial background shown in Fig.
only in azimuth and no elevation correction is 2.46, moving sinusoidally between declina-
needed for an hour or more. Similarly, the tions of approximately +28 and 28. How-
motion close to moonrise and moonset is ever, it moves only slightly against that
predominantly in elevation. background during a single day. The best
times of the month are thus when the moon
GROUND GAIN is in front of a quiet region, towards the top
Even with no elevation adjustment, EME centre of Fig. 2.46. Since the moon will then
contacts can still be made when the moon is be high in the sky for stations in the northern
near the horizon. At low elevations, unclut- hemisphere, the quietest days of the month
tered ground in front of the antenna can will also provide the maximum moon-time for
provide up to 6dB extra gain. For example, I the majority of the worlds EME stations.
have contacted W5UN many times on The moon is in a slightly elliptical orbit
144MHz with the moon as high as 20, using around the earth, so its distance varies a little.
a single horizontal 16-element Yagi. However, The free-space part of the path loss (page 2-19)
at other angles the ground-reflected signal therefore varies by about 1dB between perigee
subtracts to give a minimum, and Chapter 7 when the moon is nearest to the earth and
shows how the angles of these maxima and apogee when it is furthest away. The period of
minima depend upon antenna height. the apogee-perigee cycle is not quite the same
Ground gain is mainly useful on 144MHz as to the moons orbital period; and in fact all
where cosmic noise usually exceeds the the various aspects of the moons motion drift
thermal noise from the ground (Chapter 4). with respect to one another, over periods
On the higher bands the reflection character- ranging from one month to several years.
istics of the ground deteriorate, and any extra EME conditions are at their most promising
gain is more than cancelled by the ground when full moon, night-time, perigee and high
noise. northerly declination all coincide. This will
2 61
VHF/UHF PROPAGATION
happen next during 1999-2000 [28]. In any is called Faraday rotation, and is due to the
other year and month, optimum times for earths magnetic field which causes the
schedules are something of a compromise. electron oscillators to be twisted away from
Perigee (when the signal is strongest) is not the electric-field direction of the incident
necessarily the best time; a low sky-noise wave, and hence to radiate the new wave with
temperature is more important with todays a different plane of polarization (see page 2-5).
sensitive receivers. When both stations are using linear polari-
zation, what is the likely effect of these
DOPPLER EFFECT polarization changes? The power in the radio
As a result of the relative motion of the moon wave is proportional to the square of its
seen from the station, the received frequency electric field (page 2-4), so the attenuation AP
is shifted due to the Doppler effect. The due to a linear-polarization offset angle of
relative velocity is mostly due to the rotation between the incoming signal and the antenna
of the earth, so the Doppler shift is upwards in is:
frequency as the moon is rising and moving
A p = -20 log10 (cosq ) dB
towards you, passes though zero when the
moon is almost due south, and goes progres- So the polarization offset required to cause a
sively LF towards moonset. The shift is loss of 1dB is only 27. Because Faraday
proportional to the operating frequency, rotation makes all angles of incoming polari-
amounting to a maximum of a few hundred zation equally likely in the long term, it
hertz on 144MHz and up to 1kHz on 432MHz. follows that mismatched polarization could
This effect needs to be allowed for when cause a noticeable deterioration (>1dB) in
searching for your own echoes or for weak weak EME signals for about 70% of the time,
stations on scheduled frequencies, especially with serious effects on the success rate of
when using narrow receiver bandwidths. The marginal contacts.
simplest way to deal with Doppler shift is to
set the RIT to cancel the frequency offset on FARADAY ROTATION
your own echoes and then to use the main Faraday rotation on the outward and inward
tuning control for all frequency setting. If you passages through the ionosphere is additive,
tune in a station to the same pitch as your and can combine with the spatial rotation to
own echoes, he will hear your EME signal on give apparent one-way propagation [29]. Even
the same frequency as his own echoes. if you can hear your own echoes, it doesnt
follow that other station can hear you and
POLARIZATION OF the opposite applies too! The number of
EME SIGNALS rotations depends on the lengths of the two
Imagine a horizontally-polarized wave, slanting paths through the ionosphere, and
transmitted at the moon from the UK and also on the levels of ionization and
reflected back without any change of polariza- geomagnetic field along those paths.
tion. At a longitude of 90 east or west it will The magnitude of Faraday rotation, and
arrive with nearly vertical polarization. This hence its rate of change, decreases with
spatial polarization rotation depends on the increasing frequency. At 144MHz the time for
relative longitudes of the two stations and the the polarization to rotate through 90 is
position of the moon, and could in principle typically half an hour, which is not too long
be calculated and corrected by rotating the to wait if the orientation is unfavourable. At
antenna(s) by this amount. However, the 432MHz, however, Faraday rotation can be
chances are that the signal will not arrive as more of a nuisance because it changes more
predicted because the plane of polarization slowly, and can become stuck for long
will also be rotated each time the electromag- periods.
netic wave passes though the ionosphere. This Although stations with fixed linear polariza-
2 62
MOONBOUNCE
tion suffer the most from Faraday rotation, timescale of libration fading is a few seconds
they also usually need its help on interconti- at 144MHz, and a second or less at 432MHz;
nental QSOs to overcome the offset due to however, on all frequencies it can be suffi-
spatial polarization. But Faraday rotation is an ciently rapid to break up individual Morse
unreliable ally and can lead to many unsuc- characters.
cessful schedules, sometimes with good
signals one way but not the other, until luck LONG-DELAYED ECHOES
and persistence eventually yield a two-way Though not strictly moonbounce, the puz-
contact. Since linear polarization is the zling phenomenon of long-delayed echoes
standard on 144MHz and 432MHz EME, the deserves a mention here. Rare but well-
best practical option on these bands is fully- documented cases of echoes being heard with
rotatable linear polarization. much longer delays than the 25-second earth-
However, stations with rotatable linear moon-earth path have been recorded by
polarization report that received EME signals stations carrying out EME tests [30]. Similar
sometimes seem spread over a range of effects are observed more frequently on HF
polarization angles, and the expected cross- but possible mechanisms have been suggested
polarization null is not observed. Part of this for these cases, such as trapping of signals in
polarization spreading is due to the geometri- the ionosphere for several turns around the
cal effects of reflections from the irregular earth. As far as I am aware, no credible
surface of the moon; but the larger and more explanation has yet been put forward for long-
variable component depends on ionospheric delayed echoes on VHF/UHF.
conditions, probably because the propagation
velocity through the ionosphere depends on
the alignment between the plane of polariza-
tion and the direction of the earths magnetic
field. Thus signals that were initially linearly-
polarized can become spread out, while
circularly-polarized signals (the standard on
13GHz and above) can lose their true circular-
ity. In both cases there can be a loss in signal
strength of up to 3dB.
LIBRATION FADING
As well as Faraday fading, EME signals suffer a
much more rapid form of fading called
libration fading. Although the moon always
shows the same face to the earth because its
rotational and orbital periods have become
locked together through tidal forces it still
rocks a little on its axis. This motion is called
libration, and one consequence is that
considerably more than half of the surface of
the moon becomes visible from earth over a
period of time. Another consequence of
libration is that the path lengths of signals
backscattered from various parts of the
moons surface are always changing, leading
to quite rapid fading and occasional brief
enhancements of several dB. The typical
2 63
VHF/UHF PROPAGATION
REFLECTIONS OFF far from one end of the path and hence not
too high. This limits the useful range to only a
2 64
RECOGNIZING
PROPAGATION MODES
2 65
VHF/UHF PROPAGATION
ing that the lower VHF bands are less affected REFERENCES
by a screened location. And you may still be
able to work the world using moonbounce; a [1] Recommendations and Reports of the CCIR,
quiet environment with enough space for the 1986, Vol.5: Propagation in Non-ionized
antenna is more important for this mode of Media. ISBN 92-61-02741-5.
propagation. [2] P.W. Sollom G3BGL, A Little Flutter on
VHF. RSGB Bulletin, (a) November 1966, p.709;
BE YOUR OWN
(b) December 1966, p.794.
PROPAGATION EXPERT
[3] Microwave Handbook, Volume 1. RSGB,
Science says the first word on everything,
ISBN 0-900612-89-4.
and the last word on nothing. (Victor Hugo)
If youve read this chapter carefully, you [4] J. Priedigkeit W6ZGN, Ray tracing and
have probably thought of a lot of questions VHF/UHF Radio Propagation. QST, November
which it hasnt answered. Hopefully this will 1986, p.18.
lead you to further reading; it may not be easy [5] R.G. Flavell G3LTP, Studies of an Extensive
to find the references you need, but they can Anticyclonic Propagation Event and some
usually be obtained though inter-library loans, Short-term Enhancements Observed at VHF
a purposeful visit to a specialist library or by and UHF. Radio Communication, February
borrowing from a friend. However, textbooks 1984, p.128.
wont provide the answers to all the questions [6] Julian Gannaway G3YGF, Tropospheric
you may want to ask. Radio propagation is Scatter Propagation. Radio Communication,
still not fully understood, and this is particu- September 1981, p.710 (based on reference 2).
larly true of some of the short-lived VHF/UHF
[7] G. Roda, Troposcatter Radio Links. Artech
modes.
House, 1984, ISBN 0-89006-293-5.
By frequently observing your own favourite
bands and above all, by using them you [8] Commission of the European Communi-
can learn more about their practical behaviour ties, COST Project 210 Final Report, Annex 2.16:
than many so-called experts. And you can Development of the BTRL/Yeh Troposcatter
work plenty of DX at the same time! Interference Prediction Model. EUR 13407 EN
(1990).
[9] D.J. Smillie GM4DJS, Plotting of Magnetic
Deviation and Aurora. Radio Communication:
(a) February 1992, pp.51-53; (b) March 1992,
pp.31-33. NB The circuit described can be made
much less sensitive to RF and transient interfer-
ence by reducing the amplifier bandwidth to about
1kHz.
[10] C.E. Newton G2FKZ, Radio Auroras. RSGB,
1991. ISBN 1-872309-0308.
[11] RSGB Amateur Radio Call Book.
[12] H. Hatfield, Recording Jamjar Magnetom-
eter. British Astronomical Association note
ON3 (February 1983).
Technical Topics: Kilner-jar magnetometer.
Radio Communication, July 1988 p.522.
Russ Wicker W4WD, Simple Magnetometers.
Proc. 22nd Conference of the Central States VHF
Society. ARRL, 1988. ISBN 0-87259-209-X.
2 66
REFERENCES
[13] D.O. Pettit, A Fluxgate Magnetometer. [25] R.G. Cracknell ZE2JV (G2AHU) and R.A.
BAA Journal vol. 94 (2), p.55 (1984). Whiting 5B4WY (G3UYO), Twenty-one years
R. Noble, A Simple Magnetometer. Electronics of TEP. Radio Communication (a) June/July
World & Wireless World, September 1991. 1980, p.626; (b) August 1980, p.785.
[14] A. Hopwood, Riding the Solar Storm. [26] John Morris GM4ANB, Amateur Radio
Electronics World & Wireless World, March Software. RSGB, 1985. ISBN 0-900612-71-1.
1990. Suitable programs include those by VK3UM,
[15] G.H. Grayer and M. Harrison, Spatial and WA1JXN and G3SEK for IBM-compatible
Temporal Correlation in Bistatic Communica- machines, available on disks from G4ASR.
tions via 28MHz Sporadic-E. Proc. Fifth Int. Some programs are also available for the BBC
Conf. on HF Radio Systems and Techniques, IEE, micro, Spectrum and others all enquiries to
July 1991. IEE Conference Publication 339. G4ASR.
[16] Ernest K. Smith and Sadami Matsushita, [27] R.E. Taylor, 136MHz/400MHz Radio-sky
Ionospheric Sporadic E, pp.194-214. Pergamon, Maps. Proc. IEEE, April 1973, p.469.
1962. Reprinted in Eimac EME Notes AS-49-25, and
[17] Michael R. Owen W9IP, The Great also in Recommendations and Reports of the
Sporadic-E Opening of June 14, 1987. QST, CCIR, 1986, Vol.6: Propagation in ionized
May 1988, p.21. media. ISBN 92-61-02751-2.
Emil Pocock W3EP, Sporadic-E Propagation at [28] Michael R. Owen W9IP, Orbital mechan-
VHF: a review of progress and prospects. QST, ics of the moon. Proc. 21st Conference of the
April 1988, p.33. Central States VHF Society. ARRL, 1988. ISBN 0-
87259-066-X.
[18] Geoffrey H. Grayer G3NAQ, Sporadic E
and 50MHz Transatlantic Propagation during [29] Tim Pettis KL7WE, Spatial Polarization
1987. Ham Radio Magazine (USA), July 1988, and Faraday Rotation. Proc. 22nd Conference of
p.10. the Central States VHF Society. ARRL, 1988.
ISBN 0-87259-209-X.
[19] G.A. Moraitis, Temporal variations of
MED F0Es. National Observatory of Athens [30] H.L. Rasmussen OZ9CR, Ghost Echoes on
Ionospheric Institute, preprint (1989). the Earth-Moon Path. Nature, September 4
1975, p.36.
[20] Jim Bacon G3YLA, An introduction to
sporadic E. Radio Communication, 1989 (a)
Fig. 2.10 is reproduced by permission of the RSGB,
May, p.37; (b) June, p.37; (c) July, p.51.
and Fig. 2.20 by permission of the Rutherford
[21] J.D. Whitehead, Sporadic E layers: History Appleton Laboratory.
of Recent Observations. COSPAR 1988.
J.D. Whitehead, Recent Work on Mid-latitude
and Equatorial Sporadic-E. Journal of Atmos-
pheric and Terrestrial Physics, vol.51 (5), p.401
(1989).
[22] Emil Pocock W3EP, Auroral-E propaga-
tion at 144MHz. QST, December 1989, p.28.
[23] T.F. Kneisel K4GFG, Ionospheric scatter
by field aligned irregularities at 144MHz. QST,
January 1982, p.30.
[24] H.H. Sargent III, A Prediction for the Next
Sunspot Cycle. Proc. 28th IEEE. Vehicular
Technology Conference, Denver, Colorado, March
22-24 1978.
2 67
VHF/UHF PROPAGATION
2 68
OPERATING
3 1
OPERATING
OPERATING
3 2
GENERAL OPERATING TECHNIQUES
3 3
OPERATING
youre doing, dont stop thinking! If an operator can get something out of a
Your first objective is to get into contact pile-up, even if its only one letter, he can use
with the person whos just finished calling it as a hook to pull out the station he wants.
CQ. But what if someone else is calling too? For example he might say QRZ the station
To adapt the old saying: One caller is com- with Kilo in the call?, or QRZ the Kilo
pany, twos a pile-up. Pile-ups sometimes get station? If that could have been you, call
totally out of hand, but at their best they are a again. If not, keep quiet and wait for your
game of friendly rivalry ruled (mostly) by skill. next chance.
The key to getting your callsign recognized in And dont call unless youre sure the other
a pile-up is simply this: station is listening! This sounds obvious, but
its amazing how many stations continue to
Put yourself in the place
call a station who is actually transmitting, or
of the other operator
worst of all continue to call when clearly a
What is he or she doing, thinking, right at QSO has begun with someone else. In a heavy
this instant? (Chances are that its a he, pile-up for a weak DX station it isnt always
though CW brings some nice surprises.) Hes obvious whats going on, and everybody
probably trying to hold the microphone, to makes mistakes but nobody ever spoiled
finish writing the details of the last contact in their on-air reputation by deciding not to
the log, and to gulp a cup of coffee, all at the press the PTT.
same time. He wont have a fourth hand free Success in pile-ups isnt easy, and there are
to tune the receiver, so call on exactly the no guaranteed solutions. If you arent doing
frequency hes already listening to. too well, the important thing is to try and
Call him by spelling only your own callsign, understand why not and maybe change your
just once. He will assume that anyone on the tactics.
frequency is calling him, and he already
knows his own callsign, thank you. Call as THE NEXT CHANCE
soon as he goes over to receive, then listen. If If you didnt make it last time, you can always
he comes straight back, your QSO is under try again at the end of the QSO thats going
way. If not, and you can hear other longer- on now. The station who called CQ generally
winded stations still calling, you can slip your retains the frequency when the QSO is over,
callsign in again. and a competent operator will immediately
A good operator will probably catch your tell everybody that hes ready for further calls.
callsign or somebodys first time. But not Let him show you the correct timing. Mean-
even the best operator can do that every time, while you should have placed yourself
and some people seem to need several repeti- accurately on the frequency of the station
tions before anything sinks in! In any case, being worked, ready to drop in your callsign
the other operator may be suffering a lot more at exactly the right moment. Well be talking
interference than you and, if youre running some more about this as the chapter
less than legal-limit power, your signal may continues.
not be so strong as he is with you. Possibly he
has a pile-up of stations of roughly equal CALLING CQ =
signal strength, all calling at once. In Europe TAKING CHARGE
he will probably be working in a foreign If nobody ever called CQ, there would be
language and if its English hes doing you a precious few contacts made. But before you
big favour. So never underestimate the other do, there are three things you need to con-
persons operating ability if he doesnt catch sider. First of all, your chances of raising some
your callsign first time. If he needs to call real, rare DX are pretty slim, especially if
QRZ? (meaning please call again) then youre just one of many stations in your
simply do that, same as before. locality. Wouldnt you do better to listen
3 4
GENERAL OPERATING TECHNIQUES
3 5
OPERATING
acknowledgement of your signal report to him, nice QSO. HB9QQ signing with G4ASR, and
and his report to you. Follow this with your HB9QQ is calling QRZ DX, over.
own acknowledgement, and the basic QSO is Note that theres absolutely no doubt about
complete. who is staying on the frequency, or when hes
So a typical bare-bones QSO would go like ready for another call, so the foundations are
this (where a callsign is printed in bold, it already laid for the next QSO.
should probably be spelled out phonetically) As I said, if you called CQ, you generally
CQ DX, CQ DX from HB9QQ, HB9QQ, retain the frequency when the QSO is over;
HB9QQ, over. but there are exceptions. If youre called by a
G4ASR station which is much more rare or exotic
G4ASR from HB9QQ hello David, youre than yourself, your QSO will probably attract-
5 and 9 in JN47, over. ing a pack of locals who also want to work
HB9QQ from G4ASR roger, thanks Pierre, him so its courteous to everyone if you give
youre 5 and 9 also, in IO81MX, over. up the frequency. If youre being called by
From HB9QQ roger ... more DX, you can always announce where
From the moment I hear that roger from you intend to go, eg:
Pierre, our basic QSO is complete and valid, G4ASR signing with HB9QQ and going
and wed be entitled to exchange QSL cards twenty-five HF, twenty-five HF.
even if the band went flat at that very instant. Ill say more about this and other special-
Since we arent super-DX to each other, and ized operating techniques later in this chapter.
have obviously worked several times before, Now lets look at some other general aspects
wed probably have a brief chat about condi- of VHF/UHF DX operating.
tions. Wed probably also ask about activity
from rare locator squares in each others CW OPERATION
locality, wish each other good DX and then The most effective mode for difficult and
sign off and go looking for some! long-distance VHF/UHF work is undoubtedly
Thats all very idealized and simplified. CW. Chapter 4 shows that CW gives you far
What youd actually do would depend on the more ability to overcome the losses of long
precise details of the situation. Obviously you paths than any telephony mode, mainly
dont need to go through the formalities in owing to the narrower bandwidth but also
such a brisk and businesslike way with Fred in because of the simple nature of the signal at
the next town; even Pierre and I started to any given moment you only need to decide
relax as we each realised that the band was whether the carrier is on or off. There is no
wide open. But at the other extreme, where valid reason why a CW capability shouldnt
signals are in and out of the noise and be part of every VHF/UHF station, but a
communication is only just possible, there is surprising number of operators make no use
considerable potential for misunderstandings of it. Yet CW is the key to success with many
and mistakes. This applies particularly to weak-signal transmission modes such as
meteor-scatter and moonbounce (see later meteor-scatter or moonbounce. During an
sections), so formal QSO procedures are aurora, while SSB operators are struggling to
absolutely mandatory for these modes. work the semi-locals, the CW DXers are
working the rare stations with relative ease.
SIGNING OFF Theres a saying that A little CW goes a long
When the QSO draws to a close all the way; well, a lot of CW goes further still! And
sooner if there are other stations waiting the once youre comfortable using it, Morse is
station who called CQ always has the last actually fun. Until youve acquired the skill, it
word. In the example above it was HB9QQ, so needs a major leap of faith to believe that
he ends with something like this: but its true.
OK, Seven Three, David and thanks for the In general the good CW operator is the one
3 6
GENERAL OPERATING TECHNIQUES
whose Morse is easy to read, and doesnt try to be recovered with a better signal/noise ratio
impress by sending fast. The optimum speed than FM when the signals are weak.
of sending depends entirely on circumstances, Ive already touched on the question of
and when conditions are poor its sensible to when to spell letters phonetically. Basically its
send more slowly than when the signals are a balance between speed and intelligibility,
loud and clear. For DXers, another major and you judge it best by experience. Appropri-
virtue of CW is that the abbreviations used on ate internationally recognized phonetics can
the amateur bands jump clean over the be most helpful under certain circumstances,
spoken language barrier; fluent amateur CW especially in weak-signal work or during
operators anywhere in the world can under- sporadic-E openings when conversing with
stand each other perfectly. For an excellent VHF-only stations who dont normally have
introduction to CW DXing on any band, read QSOs in English at all. The ICAO phonetic
The Complete DXer by past-master W9KNI [1]. alphabet (Alpha, Bravo etc) should probably be
Many VHF/UHF CW operators find it the first you should try, since thats what the
difficult to put their transmitter or transceiver other operator is most likely to be expecting,
exactly on another stations frequency. Unless but it isnt optimum for all letters in all
you do, you could well be calling as much as languages. Its up to you to devise the best
2kHz off frequency. If the other station is combination of phonetics which works for
using a 250Hz CW filter on receive, he just your own callsign.
wont hear you. The following technique will Better still, if you find yourself working into
enable you to zero-beat your frequency to that a country where English is not frequently
of the other station. First, turn off the RIT and spoken, try to use the native language. Even if
XIT (incremental tuning or clarifier) and your accent is terrible and you can only
centre the IF passband tuning. In almost any stumble through a contact, give it a go. You
transceiver built today, the audio frequency of dont have to learn the whole language; there
the CW monitor is meant to match the offset are an increasing number of books and tapes
of the transmitted CW note. So the trick is to aimed specifically at helping you to make
turn off the transmitter VOX and send a string amateur radio QSOs [2]. Provided you use the
of dashes while still receiving. Adjust the correct language (eg French doesnt go down
transceiver frequency until the audio tone of well in Flemish-speaking northern Belgium),
the station you wish to zero-beat is identical your efforts will be very much appreciated and
in tone to the CW monitor note. When you very effective in terms of DX contacts. When
transmit, you will be exactly on that stations TA finally comes through on 144MHz spo-
frequency. If you wish, you may now use the radic-E, the first to crack the pile-up will be
clarifier to adjust the pitch of the received the operator who can call in pidgin Turkish!
signal to your liking. This technique obviously Theres a VHF DXer in Southern England who
relies on the offset of your CW carrier being is willing to attempt a basic QSO in just about
the same as the sidetone note, and in most any language. Not only does he have the QSLs
rigs youll need to adjust the frequency of the to prove it, but hes also made a lot of friends
CW oscillator crystal to make certain that it is. in the process.
3 7
OPERATING
unexpected, so dont add to it by using fancy did, youd better move away some distance to
operating tricks or by borrowing procedures get clear of the pile-up youre leaving behind.
that dont belong on the amateur bands. Making a successful CW QSY can be very
For working DX, just keep your operating difficult, especially in crowded conditions
style simple and predictable, and stick close to close to the CW calling frequency, so thats all
the standard pattern and procedures of an the more reason to spread out and tune
amateur-band QSO. around when you know the band is open.
3 8
GENERAL OPERATING TECHNIQUES
from most national amateur radio societies or DXing. Obviously you dont like missing the
from specialist newsletters. Avoid transmitting DX, but take comfort in one thought: nobody
within the beacon sub-bands interference goes to the trouble of arranging a good VHF
reduces the value of the beacon service to station for only a single evenings operation.
others and spoils your personal reputation. There will usually be another chance some
other time. Therefore an important virtue a
PLAYING THE GAME DXer must possess is long-term patience.
Whenever anyone tries to do something really Now lets look at another pile-up situation.
challenging, theres always a temptation to During an opening, you hear many stations
bend the rules written or unwritten. And trying to work OK1KRA on 144MHz. Youve
theres always a temptation for other people already worked that square and country, and
to accuse those who are successful of having in fact the White Mountain Radio Clubs QSL
somehow cheated. VHF/UHF DXing is no card has been on the wall for ages. So what do
exception, so lets look at some of these you do? Do you switch on the high-power
questions. amplifier and go for it? Have you forgotten
the anguish that you suffered when you were
DID YOU REALLY WORK HIM? chasing your first contact with Czechoslova-
Keen intuition can be as important as keen ears kia? You have two choices: leave the amplifier
in copying weak signals. Theres no dividing on and go looking for something rarer, or turn
line between the role of the ears and that of the it off and indulge in a little of what the
brain. But intuition is not the same as guess- Russians call radio sport. You can set your
work. If you think you know what youre own handicap, and only you know whether
hearing, you still need to check it rigorously, you did it with ten watts, four hundred watts
point for point, against what youre actually or two kilowatts. Taking a multi-dB advantage
hearing. Almost all DX operators are scrupu- into a pile-up and winning is nice; but it isnt
lously honest about this, and are quite pre- always fair.
pared to scrap a difficult contact unless every-
thing about it is definitely correct. SKEDS ARE THEY FAIR?
It follows that everyone will have a certain Some people consider it unfair to make VHF/
percentage of failed DX contacts and the UHF schedules (skeds) via the HF liaison
more you try to achieve, the higher that frequencies or by telephone. But anyone
percentage is likely to be. An operator who whos tried the really challenging modes like
seems to be 100% successful is either not long-distance troposcatter, meteor-scatter and
trying for the real DX or else attracts the moonbounce will confirm that skeds are
greatest suspicion! Anyone can make an legitimate and often necessary. On these
honest mistake in claiming a contact, and modes, the other station is in and out of the
thats what QSL cards are for. But operators noise at the best of times. Although a QSO
who show well, lets call it excessive may be possible when both beams are lined
optimism are well known on the DX up and both stations are looking for each
grapevine. And even if they eventually grow other on the same frequency, its stretching
out of it, there are some very long memories... coincidence to expect that many such QSOs
will happen by chance. In fact its only
ETHICS IN PILE-UPS through making skeds that weve developed
Working a new country or a much-wanted these modes at all.
locator square can be very exciting. Losing out People who criticize skeds are often ignorant
in a pile-up isnt as much fun. But losing is a of the stringent requirements for a complete
part of DXing. We all have to lose sometimes. QSO. If you and I make a sked, we may know
If we managed to crack every pile-up success- each others callsigns in advance and we may
fully there would be no challenge left in even be able to guess what report were going
3 9
OPERATING
TROPO DX
Since tropospheric propagation is the most
common DX mode, Ive already used it in the
previous section to outline and discuss general
operating techniques. However, there are a
few special tricks and topics to tropo as well.
Unlike most other VHF/UHF propagation
modes, tropo openings can last for days and
sometimes weeks. So you can often relax your
operating procedures, compared with other
modes which may only last for seconds. You
can even exchange real locations and talk
about the weather, just like 35MHz!
Long-distance working on SSB takes place
within the appropriate sub-section of the
band concerned. In Europe and Africa (IARU
Region 1) this equates to 350kHz of sub-band
on 144MHz, and in contests and extensive
openings this space fills up very rapidly.
During openings it wont be necessary to use
the calling frequency as you might when band
conditions are flat. In fact, you should make a
firm resolution not to go anywhere near the
calling frequency during an opening, except
to listen for DX; and let me tell you in
3 10
SPECIAL OPERATING TECHNIQUES
advance that the chances of hearing anything of JPCs S9, but so what? Result one new
interesting will be slim verging on none, locator square and maybe a new country as
because of the pandemonium in that small well. Its really satisfying when you try some
part of the band. of these more subtle operating tactics for the
If youre an average suburban station in a first time and find that they really do work.
well-populated locator square and you want to But what if you have to run with the pile-
work similar stations a few hundred kilome- up? Lets assume youve tuned the band
tres away, by all means call CQ in a big tropo carefully and hear a massive dogfight in
opening. But if you want to work anything progress. Somewhere underneath it is the one
new or rare, the only way to accomplish this station which will give you DXCC on
is to listen, listen and listen some more. Tune 144MHz. What sure-fire technique can you
up and down the band, checking out every use to work him? Answer: there isnt one. DX-
interesting-sounding signal, and dont move chasing is very similar to fishing what works
away until you have confirmed that either its in one situation may be a total failure next
not DX or its not in a square you require. It time. But even though there are no guarantees
can also be useful to monitor what the well- of success, knowledge of a few techniques can
sited stations in your area are working, to give help. Effective radiated power obviously helps
yourself an idea of how propagation is going too, but its nothing like as important a
and what chance you have of working the precisely-timed call on the right frequency.
really long-distance stations. What is the right frequency? In some
instances it will be the frequency being used by
PILE-UPS the operator whos just ending a QSO with
You may conclude from this that the only your wanted station. Tail-ending dropping in
way to work rare DX is to attach yourself to a your own callsign just as he completes his last
pile-up and rely on your brilliant operating transmission is accepted by some operators,
technique. Sometimes that may be true, but but is open to grave abuse if used unskillfully.
its also important not to fall into the syn- Unfortunately there are too many operators
drome that HF DXers call pile-up fixation. who know about tail-ending, and too few who
You can keep pounding away, determined to know how to do it properly. Also, there arent
work the station come what may. Maybe you enough DX stations who are prepared to take
make it, maybe you dont but either way charge of their pile-up and create some sem-
you spend half an hour trying. The experi- blance of order by telling callers exactly what
enced operator may make only two or three they want them to do. So tail-ending is best
calls, and if that fails will leave the second avoided unless thats the only way the DX
VFO on the pile-up and go looking elsewhere. station is accepting the next call. If hes not
The chances are that other stations located in accepting tail-enders, a call just off the fre-
the same wanted squares will also be heard. quency of the last person contacted can often
Another facet of pile-up fixation is the 59+ produce results, on the basis that if everyone
syndrome an irrational urge to work all the else is zero-beat, your call will stand out by
strong stations. For example, a cursory sweep being different. This applies particularly to CW
around the band suggests that the only station operating, though calling on CW can some-
audible from Austria is OE3JPC. Since his times give good results in an SSB pile-up too. It
signal is strong, he has attracted a very large makes your call stand out, even if the station
pile-up. However, a more careful tune of the doesnt want a CW QSO, or cant copy CW!
band finds OE3OKS calling CQ at a perfectly I take it all back there is just one sure-fire
workable S4. As all the other local stations are way to beat a pile-up. That is to be in the other
locked in mortal combat over OE3JPC theres stations log before the pile-up starts. In other
no competition for OE3OKS and you bag him words, you need to be tuning around so that
first call. He might have only been S4 instead youre right there when the DX station pops
3 11
OPERATING
up. So we keep coming back to the DXers something, do it straight away before you let
golden rule listen, listen, listen. If you dont the station go youll never get a better
hear them you dont work them! chance.
Comparing the formal requirements for a
CONTESTS valid QSO (p3-5) with what actually happens
A typical contest exchange may sound in contests, you may gather that many contest
something like this (where a callsign is printed QSOs are not valid, strictly speaking.
in italics, it should probably be spelled out Although this is less true of VHF/UHF contests
phonetically): than those on HF, it may come as a surprise to
CQ contest, CQ contest from GD4APA contestants and adjudicators alike! Almost all
portable, GD4APA portable, contest, over. contest rules tend to concentrate on the
G4ASR requirements for logging, and forget about the
G4ASR, youre 59672 in IO74PD, QSL? requirements for making valid QSOs in the
Roger, youre 59084 in IO81MX, good first place.
luck. A good contest operator never seems to be
Thanks Dave, and 73. QRZ contest from in a hurry, but is actually making a constant
GD4APA portable, over. stream of split-second decisions. Normally a
Make it easy for the other station to copy good operator can well afford the time to greet
your information onto the contest logsheet. callers and thank them for the contact. (That
Always give it in logsheet order report, serial dont-have-time feeling isnt actually true;
number, locator. The quickest way to get your its merely distracting your attention and
information accurately into his log is to speak spoiling your efficiency as an operator.) Even
clearly at dictation speed no faster than you when the going gets really rough and theres
yourself could write in block capitals. If signals no substitute for 100% efficiency, you can still
are weak, use your best judgment about sound interested and pleased to work some-
repeats and phonetic spelling. one without actually expressing it in flowery
Time is at a premium in contests, so people words. Although contests are competitive,
frequently cut corners in QSO procedures if outright rudeness is fortunately rare it
they can. The trick is to understand where usually means that someones getting rattled
youre cutting corners, and what risks youre because too many things are happening at
taking. Good contest operators are accurate once!
and fast but strictly in that order.
What might get left out of a contest ex-
change? You dont always make a two-way
exchange of both callsigns with the other
station; because he comes back on the right
frequency at all the right times, you may be
forced to assume that hes working you. This
may or may not be true! Similarly, if a contest
station wishes you thanks and 73 and
presses on to the next contact, as GD4APA/P
did in the above example, you can usually
assume that hes copied everything OK from
you. If you let things go at that, hell assume
that youve copied his information as well; so
if theres anything youre not absolutely
certain about, you must stop him and make
sure. Theres no need to read back everything
he said to you, but if you do need to confirm
3 12
AURORA
3 13
OPERATING
3 14
SPORADIC-E & METEOR-SCATTER
3 15
OPERATING
interested stations will arrange in advance the you wish, eg the periods could be reduced to
frequency, timing and duration of the test, as one minute by agreement; but your first
well as the transmission mode and callsigns to priority should be to avoid causing interfer-
be used. Scheduling may be by exchange of ence to local stations who are using the
letters, one station offering a proposal, or by standard periods.
radio via the European VHF net which is The recommended standard period for both
active around 14345MHz and is particularly random and scheduled SSB is one minute.
busy on Saturdays and Sundays before major However, time periods of less than that are
shower periods. As the popularity of meteor- encouraged during major meteor showers.
scatter is increasing, its not uncommon to Quick-break procedure within scheduled
find some activity on this net at many times contacts can be very effective. This could
throughout the week also. Depending on involve taking a break every few seconds,
propagation, 3625MHz and 28385MHz are using a pip-tone (Chapter 12) if so desired.
also used for scheduling of MS tests. Although not defined in the IARU procedures
Random (ie non-scheduled) MS contacts are for random SSB contacts, a growing number of
far more difficult. Because youre starting stations leave a break every quarter-minute in
entirely from scratch, its particularly impor- case the QSO can be completed in one long
tant for both stations to follow the standard burst. If other local stations adopt this proce-
QSO procedures. dure and are able to set their clocks correctly,
Because of the very short burst lengths there should be no problems.
normally encountered via meteor-scatter, a
greater level of station organization and TIMING ACCURACY
operating skill is required than for normal DX Unfortunately a large number of operators
working. In Europe, the standard procedures seem unable to set their watches or clocks
co-ordinated by IARU Region 1 must be accurately. This is surprising since accurate
followed to ensure that a maximum of correct Standard Time (UTC) is freely available
and unmistakable information is passed both anywhere in the world by the magic of radio.
ways. As with operating procedures in general, It is absolutely vital that time-pieces are
the virtues of the MS procedures are mainly checked prior to any MS activity, eg by
that they are standard and are widely under- checking against the time-ticks on standard
stood throughout Europe and beyond. And frequency transmissions, or by using teletext
although the MS procedures are different in or the telephone companys speaking clock.
the USA, its obviously equally important to For MS activity, clocks need to be set with
be using the same procedures as the person better than two seconds of Standard Time.
youre trying to work. Any inaccuracy will result in wasted time,
when both stations are either listening to
TIMING nothing or both transmitting at once, and will
Accurate timing of transmit and receive periods also cause totally unnecessary interference to
is important for two reasons: to maximize the local MS stations. Time it right.
chances of hearing the other station, and to
avoid QRM between local stations. MS takes SCHEDULE DURATION
place in relatively narrow frequency segments Scheduled contacts are usually in the range 1
and it can be very difficult to hear weak DX 2 hours, although during shower periods this
while a local stations keyer is hammering away can be reduced to 30 minutes or less. Start
on a nearby frequency. times should be arranged to be on the hour, eg
The recommended time period for random 0000, 0100, 0200 etc. Every schedule period
CW contacts is 2 minutes to avoid interfer- must be considered as a separate uninter-
ence between local stations. With scheduled rupted test. It is not permissible to break off
contacts you can arrange for any time period and then recommence at some later time.
3 16
METEOR-SCATTER
3 17
OPERATING
3 18
METEOR-SCATTER
Start Start
if transmitting first if receiving first
Transmit calls
for one period
Have I
NO identified
QSO
partner
?
YES
Have I
copied full NO
calls plus Transmit calls Listen for one period
report for one period
?
YES
Have I NO
copied Transmit calls plus R Listen for one period
Rs reports for one period
?
YES
Transmit R strings
for 3-minute periods
Fig. 3.1. Flowchart of meteor-scatter
operating procedure
3 19
OPERATING
ously the 38 part of these example confirma- at Fig. 3.1 which incorporates the same
tion reports would change according to the information as Ive been discussing above [3].
burst length and signal strength, so for the MISSING INFORMATION: If you receive an
rest of this section Ill call it an R** report. R** report, it means that the other station
When either operator receives a confirmation has copied both callsigns and the report from
report, he cant get any further until he himself you, yet you may still need something from
has received everything he needs. When he him. At this stage, you can try to ask for the
has, he confirms with a string of Rs, inserting information you need, by sending one of the
his callsign after every 8th R to avoid confusion following strings:
with any other QSO on an adjacent frequency: MMM my callsign missing
RRRRRRRR G4ASR RRRRRRRR G4ASR YYY your callsign missing
RRRRRRRR G4ASR ... BBB both callsigns missing
When the other operator has received just SSS duration and signal strength report
one of these Rs, the contact is complete. There missing
is some general uncertainty about this, so let OOO information incomplete you seem
me explain in more detail. to have left something out
In our example, if I hear an R** report UUU unreadable keying
from ZB2IQ (implying that hes copied both The other operator should respond by
our callsigns and also my report to him) and if sending only the required information.
I too have copied everything I need, then Ill Although none of the above strings can be
start sending RRRRRRRR. When ZB2IQ confused with an R, even if the characters
hears just one of my Rs, the minimum valid are run together, this technique must be used
QSO will be complete and he can start writing with great caution to prevent confusion.
out a QSL card for me. SSB CONTACTS: Contacts using SSB are
But I have no way of knowing what hes conducted in the same way as for CW. When
heard, or whether the sked has succeeded or attempting random contacts, speak the letters
failed; so to save me the agony of waiting clearly, using phonetics where appropriate. It
until the QSL arrives, ZB2IQ would usually may not be necessary to use phonetics during
send a string of Rs back. In fact wed probably a scheduled contact, but dont gabble.
exchange Rs for two or three further periods if
there was time. Note that this is purely for our LOGGING FOR M-S
own satisfaction the QSO is already com- Logging is an important aspect of meteor-
plete. Alternative (though less positive) scatter work. It provides each operator with a
methods of indicating that a QSO is complete record of the received information which can
are to stop sending, or to start calling CQ be used to establish whether sufficient has
again if appropriate. been received to send a confirmation report or
If you find yourselves running out of signing report. A copy of the logsheet is a nice
scheduled time in an MS test and the QSO courtesy if youre QSLing direct.
seems nearly complete, it is conventional to The example MS logsheet in Fig. 3.2 shows
run on for a period or two into the next sked. the development of a contact with T70A. After
Conversely, its unwise and discourteous to four receive periods, sufficient information
abandon a sked if you dont hear anything had been received to allow a Roger report to
straight away the other station may be be transmitted. Two receive periods later a
finishing off with someone else, his alarm string of Rs was received from T70A and the
clock might be late, he may be making a quick contact was complete. For our own satisfac-
repair, or there may simply have been no tion I transmitted a string of Rs and 73s at
meteors in the first few minutes. 0227, but these were not strictly necessary:
FLOW CHART: If you prefer to visualize the the contact was already complete.
MS QSO procedure as a flow chart, take a look It is customary to record the number of
3 20
METEOR-SCATTER
separate bursts, pings and duration and signal burst which is certain to contain everything
strength of the longest burst received. This is you need, you can of course play it back
not part of the QSO procedure but is generally straight away. Experience will tell you how
reported on QSL cards, and it is also the much information is contained in a specific
reporting format used in specialist newsletters length of burst, and even though you cant
such as DUBUS [5]. fully read the high-speed Morse you can often
recognize the rhythms of callsigns and reports.
RAPID REVIEWING OF As a guide, it takes a little under two seconds to
CW TAPES receive G4ASR SP5EFO 36 36 36 transmitted
To enable a CW contact to proceed smoothly, at 600 letters per minute. At that speed, whole
you need to review the tape recordings of the callsigns sound rather like dashes sent at a
received bursts quickly so that your transmitted much slower speed, and reports or Rs sound
information can be changed if appropriate. like dots, so the message above has the rhythm
When each burst is received, note the reading of the Morse character 7.
on the tape counter and keep track of which
bursts are the longest or loudest. During your INVALID MS CONTACTS
Fig. 3.2.
Example of a transmit period you can then rapidly return to On many occasions, especially during major
meteor- each burst, starting with the most promising, shower periods, operators are heard using the
scatter and play it back slowly. If you get a long, loud wrong procedures and consequently making
logsheet doubtful or completely invalid contacts.
Heres how not to work ZB2. You are
listening on 144400MHz during the Perseid
meteor shower:
You receive CQ ZB2IQ CQ ZB
You transmit G4ASR 38 repeatedly on
144400MHz
You receive ROGER 37 ROG
You transmit ROGER repeatedly, until
You receive ROGER QRZ ZB2
That was not a valid contact! How many
things can you find wrong with it?
For a start, G4ASR didnt give ZB2IQs
callsign when calling with the report. The
chances are that six other people also think
they are working ZB2IQ, but even ZB2IQ cant
be sure. Next, ZB2IQ is heard to be sending
ROGER 37 to someone; but it might not be
G4ASR, because G4ASR never said who he was
calling in the first place. Finally, G4ASR didnt
identify himself while giving the closing
Rogers but its too late by then, because the
whole thing is already a complete mess.
Heres how to do it properly. ZB2IQ selects a
convenient frequency in the SSB sub-band,
lets say 144.403MHz. So:
You receive CQ ZB2IQ CQ ZB for 6
seconds at S6
You transmit ZB2IQ G4ASR 38 repeat-
edly on 144403MHz during your allocated 60-
3 21
OPERATING
second period
until
You receive G4ASR ZB2...ROGER 37 MOONBOUNCE
ROGER 3" during their 60-second period
Right! Youve copied both call signs, 37
and ROGER, so
You transmit ROGER G4ASR repeatedly,
possibly giving breaks every quarter minute
until
You receive QRZ ZB2IQ
and ZB2IQs further QSOs continue on
144403MHz.
This time both operators have copied both By using the moon as a passive reflector, any
callsigns, the report and an unambiguous two stations whose antennas can simultane-
confirmation. Whats more, youve heard ously see the moon can work each other
ZB2IQ move on to the next QSO, so you can maybe. Hundreds of stations are now active
be pretty certain that yours is in the bag. on the 144 and 432MHz bands, and a few
have very large antenna systems which can
OPERATING TACTICS make up for the smaller systems of other
When starting MS, I strongly advise you to operators. As noted in Chapter 4, it is now
listen to other peoples skeds particularly in quite easy to work these mega-stations on CW
order to master the technique of rapid review with quite a small system at your end, espe-
of high-speed CW tapes. When youre ready to cially during moonrise or moonset periods.
try a QSO, its best to make a sked with a DX The best time for EME (earth-moon-earth)
station who youve already been hearing well, contacts is when the moon is nearest to the
rather than plunge in at the deep end on the earth (perigee), when it is overhead at rela-
random calling frequencies. Above all, dont tively high northern latitudes (positive
get into MS by picking up the bad operating declination), and when the moon is nearly
practices which are frequently heard around full (less ionospheric absorption and Faraday
144200MHz during showers. rotation at night see Chapter 2). The one
Although it is feasible to work MS with low weekend of the month which has the best
power, eg 50W to a nine-element Yagi, you combination of all these three factors is
would have much more success by scheduling designated the main activity or sked weekend.
tests rather than trying to work random-style. The dates of these activity periods can be
In fact the same could be still said for stations found in specialist VHF/UHF newsletters
running higher power to larger antennas! available in many countries throughout
Higher power pays off in MS, not so much by Europe and North America [5].
making the signals stronger but more by Schedules are normally arranged via the
making weak bursts longer as they trail away EME nets which meet on 14345MHz every
into the noise and thus increasing the infor- Saturday and Sunday from around 1600 UTC
mation transfer. for stations interested in 432 and 1296MHz,
You might think that contacts via SSB followed by the 144MHz net at about 1700
should be quicker than CW, because of the UTC. As an alternative, 432/1296MHz sched-
shorter time periods. This may be true on uling arrangements can be made via the 432
50MHz where the bursts are long and strong, and Above EME Newsletter published by
but on 144MHz the signal/noise advantage of K2UYH, or by direct mail or telephone using
CW enables contacts to be made more easily the address lists in the newsletters.
than on SSB, especially if you operate outside As the moonbounce procedures are different
major shower periods. on 144MHz and 432MHz, the newcomer may
3 22
MOONBOUNCE
become somewhat confused. However, as minutes (432MHz), and the report is sent
most moonbouncers only operate on one continuously during the last thirty seconds of
band, the procedures for that particular band your time period.
are picked up fairly quickly. CW speed is very important in EME. You
need to send slowly enough for other people
TRANSMIT/RECEIVE PERIODS to copy your Morse comfortably, but not too
Time periods are two minutes on 144MHz and slowly; otherwise you are wasting valuable
2 minutes on 432MHz and above. Its easy sked time and individual Morse characters
enough to keep track of 2-minute periods, but may get broken up by fading, making them
for 144MHz its advisable to use an operating more difficult to read. In fact the optimum
chart, similar in style to the meteor-scatter technique is to send at an overall speed of 50
log. Unless signals are strong enough to allow 75 letters/minute (1015 words/minute) but to
the transmission to be passed back and forth send the letters and numbers a little faster
with a good degree of certainty, strict time with more distinct spaces between them.
periods are always used. Dont change speed, and take care over letter
The station whose callsign appears first in a spacing in particular, make T, M and
sked-list always transmits first, and subse- O sound very different!
quently in the other odd time periods. When both callsigns have been positively
Otherwise, on 144MHz the station located and completely copied, only then may you
furthest east generally transmits first, while on begin to send a report. On 144MHz the
432MHz and above its the station furthest minimum report acceptable for a valid contact
west. is O, whereas on 432MHz the report can
either be an M or O. Reports must of
SIGNAL REPORTING SYSTEMS course be acknowledged by the other station
Signal reports on EME are used differently for the contact to be complete, and this is
from those on MS, and there are also differ- done by adding an R to the report being sent,
ences between 144MHz and 432MHz. eg RO RO .... Note that the criterion for
144MHz Report 432MHz sending reports on EME is different from that
Signal just detectable. T Portions of call copiable. used in meteor-scatter work; on MS you can
Portions of calls M Both calls fully copied, start to send reports at a very early stage, but
copiable, but not weakly. on EME you have to wait until youve copied
complete calls. both callsigns.
Both calls fully copied. O Both calls copied On receiving RO (or RM on 432MHz),
comfortably. you respond by sending RRRRR on its own,
Both calls and O R Both calls and M much as in meteor scatter. At the end of a
report copied. or O report copied. contact, you may complete by exchanging
An RST signal report may be used when 73, GL and other abbreviated pleasantries
signals are loud enough, though the meaning followed by SK to end the contact; since
may not be the same as on terrestrial modes. these are only sent when everything else is all
For example a weak tropo signal that might right, they are generally accepted as a substi-
scrape RST319 would probably get a 449 tute if the closing Rs were lost in a deep fade.
report on EME and be considered very good! A typical 144MHz EME contact might go
something like the log overleaf.
EME contacts are seldom as easy as that!
When first starting a schedule, both callsigns You may not copy the callsigns or report, and
are sent for the whole duration of your time repeats will then be necessary. The lack of a
period. During the middle stages of the report from the other station will tell you that
contact the callsigns are sent continuously he still needs both callsigns, the lack of an R
during the first 1 minutes (144MHz) or two that he still needs a report as well. If abso-
3 23
OPERATING
lutely nothing is heard from the other station, receive GGG, reply with IO82 or wherever
send nothing in the last 30 seconds; this is a you are.
very definite indication to the other station. If Finally, there are a number of EME-only
you can hear the other station, but not well contests notably the annual ARRL event
enough to copy both callsigns then the T which attract a lot of interest and often
report could be used. This may prompt the stimulate DXpeditions and other special
other station to change the polarization of his stations. Contests can be particularly interest-
antenna system, if he can, but it can also be ing to the first-time listener, although it has to
dangerous since TTT could easily be con- be said that the hurly-burly of a contest is
fused with M or O. Sending the callsigns hardly the best way to start your EME trans-
during the period set aside for reports can also mitting career. Operating procedures are often
be very confusing for the other station if shortened between stations with good signals,
signals are weak, especially if fragments of especially in contests.
either callsign could be mistaken for M or If you want to cut corners with EME
O. You have to operate carefully on EME, procedures you need to be an experienced
and its no wonder that quite large propor- moonbouncer and require no lessons from
tions of skeds dont make it. But then again, me. If youre a beginner or you want to be
nobody goes on EME for an easy ride. absolutely sure of making the QSO, stick to
When answering CQ calls on EME, you the procedures.
should send only callsigns but not the report
because you havent yet heard your own
callsign back from the other station. If the
other station cant copy you, he will reply
with ??? or QRZ? and his own callsign for
the whole transmitting period. If he copies his
own callsign (which is much more familiar)
but cannot make out yours, he may send
YYY; reply with only your own callsign for
the whole period.
Moonbounce is a popular mode in ARRL
VHF and UHF contests, so on contest week-
ends EME stations in the USA and Canada
may send GGG if they need your locator
square as a multiplier. Unlike IARU Region 1
contests in which moonbounce is banned
because someone might try too hard, the
ARRL contest exchange is both callsigns plus
the locator square (or grid as they call it, for
example FN32) but no report. So if you
3 24
To monitor for all the propagation modes
3 25
OPERATING
3 26
WORKING MORE DX
gives, or never takes the trouble to pass the following topics may whet your appetite
message on, this is easily noticed and obvious WWV data, MUF, bearings, sunrise and sunset
steps can be taken to cure the problem! calculations, QSL details, and mail to other
DXers.
DX PACKET CLUSTER
Through the use of packet radio, operators can CONTESTS AND DXPEDITIONS
now access a powerful network providing real- Increased operating skills and greater station
time warnings of band openings. Typically up efficiency are the direct result of amateur-band
to 30 DX operators are logged into a regional contesting. If you dont believe this, tune
DX cluster node via packet radio, with user across any VHF/UHF band when a contest is
access normally being on the 70, 144 or in full swing and listen for the most expert
432MHz bands depending on local arrange- operators. The chances are that they are also
ments. These individual clusters, 1015 in the well-known DXers. Contests can be viewed in
UK, are then networked so that potentially 200 several ways: as an end in themselves, as a
or more DXers are connected together. way of measuring your own performance and
Once logged into the cluster, you can make that of your station, as an opportunity to
use of its many facilities. The primary function give a few points away and see what turns
is for spotting DX. Anyone who finds a station up, or as a chance to search the band selec-
worthy of note can announce it to all con- tively for well-equipped DX stations.
nected stations. A typical message might read; The aim of looking for contest stations in
50121MHz VK8ZLX 59 PG66 LISTENING UP specific locations can also be applied to the
All you do then is turn on the 50MHz rig working of DXpedition stations. They will
and work the DX! You dont even have to be generally be operating from rare locator
near the rig, since your computer can be made squares, counties or countries. These stations
to beep every time a DX spot is received. usually operate from good sites with high
Some computer-aided transceivers will even power and large antenna systems, and are
tune your rig to the correct frequency! therefore equipped to contact large numbers
The DX spot is not only broadcast but also of stations with more modest systems. If
stored in the clusters database. So when you youre working towards a certain goal such as
arrive home late from work you can ask the an operating award, you are more likely to
system to list what DX youve missed on any receive a QSL card from DXpedition station
particular band. Fig. 3.4 shows the 144MHz than from a contest group thats only inter-
data received on my local cluster (GB7DXC) ested in winning a trophy.
during an aurora on 13 July 1991. It shows
frequency, callsign, date, time, signal strength, LOGGING
locator and the callsign of the station who During some intense openings, via sporadic-E
supplied the information. for instance, the numbers of stations con-
Other facilities available on the cluster are tacted can be enormous. In one such opening,
too numerous to mention in detail but the I worked 120 stations in 2 hours, during
Fig. 3.4. Extract from a DX cluster database. Obviously there were many more DX stations
active users enter only the rarer ones
3 27
OPERATING
3 28
BACKGROUND KNOWLEDGE
3 29
OPERATING
Fortunately, specialist operating manuals are really arent on the DXers grapevine yet!)
available which catalogue most of the infor- VHF/UHF DXer is a UK-based monthly
mation that you will require [3, 4]. One newsletter covering all aspects of VHF/UHF
problem with these books is that callsign DX, edited by G8ROU.
systems are forever changing throughout the
Six News is the publication of the UK Six
world, so it is impossible to keep right up-to-
Metre Group. This members newsletter is
date. Band plans and specialist activity
edited by G0JJL and G0JHC.
frequencies are other items of information
which seem to change constantly or require The 50MHz DX Bulletin is issued twice
modification. monthly and contains comprehensive details
of world-wide 50MHz DX information.
NEWSLETTERS, CHARTS European Moonbounce Edition is a monthly
AND MAPS newsletter primarily intended for the 144MHz
The only way to keep up-to-date with changes enthusiast, although it does occasionally give
in operational procedures, or to learn about details of 432MHz operation. It is edited by
the latest DXpeditions, is to haunt the bands, HB9CRQ and HB9DBM.
swap news with other DXers and listen a lot. 2-Meter EME News contains all you need to
Monthly newsletters [5] are a good way to know about 144MHz moonbounce operation.
learn about events fairly quickly after they It is edited and published by K0IFL.
have happened, but the deadlines for the VHF
The 2 Metre News Sheet, issued monthly,
columns in monthly magazines are simply too
gives details of forthcoming expeditions and
long for late-breaking news. In other words, to
other DX information. It is edited by SM6EOC
be a successful DXer you need always to use
and SM6AFH.
the grapevine for both technical and operat-
ing information. More than that, you need to 2-Meter EME Bulletin, edited by KB7Q, is
become part of it. Contribute to a specialist obviously for the 144MHz moonbouncer.
monthly VHF/UHF newsletter or write a VHF 432 and Above EME Newsletter is the
column in your radio clubs magazine. The monthly newsletter for the UHF or SHF EME
more information you contribute, the more enthusiast. It is edited by K2UYH and contains
youre likely to receive in return! news, technical notes and the sked-list for the
next months activity weekend.
OSCAR NEWS is the magazine for satellite
REFERENCES
enthusiasts. Edited by G3AAJ, it is available to
members of AMSAT-UK.
[1] Bob Locher W9KNI, The Complete DXer.
CQ-TV is the definitive magazine for those
[2] Juka Heikinheimo OH1BR and Miika
interested in all aspects of ATV, SSTV on the
Heikinheimo OH2BAD Radio Amateurs
VHF, UHF and Microwave bands. Only
Conversation Guide (book and tapes)
available to members of the British Amateur
[3] Ray Eckersley G4FTJ, Amateur Radio Television Club (BATC).
Operating Manual (RSGB)
All Europe VHF, UHF, SHF Contests Calendar,
[4] ARRL Operating Manual (ARRL) a yearly publication detailing most contests
[5] Newsletters are a prolific source of operat- being held throughout Europe. Written by
ing information, news and views DH2NAF.
DUBUS is a widely-circulated quarterly VHF/
UHF/Microwave DX and technical magazine.
Published by DUBUS-Verlag in Germany, it is
available through national distributors. (If you
dont know anyone who takes DUBUS, you
3 30
ASSEMBLING
YOUR
STATION
4 1
ASSEMBLING YOUR STATION
ASSEMBLING YOUR
STATION
by Ian White G3SEK
4 2
RECEIVER SENSITIVITY
spectral bandwidth across which the noise is and the noise-free receiver. The noise power
being collected. The basic formula relating delivered to the receiver now has two compo-
noise power PN (watts) to noise temperature T nents. One is the amplified thermal noise
(K) and bandwidth B (Hz) is simply: from the resistive noise source, and the other
is noise generated in the preamplifier itself.
PN = kTB (1)
The noise contribution from the preamplifier
where k is called Boltzmanns constant and can be expressed simply as a power in watts,
has the value 138 x 1023J/K (joules/kelvin). or more subtly and usefully as the equiva-
For example, the thermal noise power lent noise temperature of the preamplifier.
generated in a resistor at a temperature of Equivalent noise temperature is one of those
290K (around room temperature) is: ideas that are hard to swallow at one gulp, so
lets take a few steps back and look at it. As far
1 38 10 -23 290 B watts
as the receivers concerned, the noise power
ie 400 x 1021W in every hertz of bandwidth that its hearing could be coming either from
B, across the entire electromagnetic spectrum. a noisy preamp, or merely from heating the
The noise power is independent of the ohmic 50 resistor above its actual temperature; the
value of the resistor, and the value calculated receiver has no way of knowing that in fact
in equation (1) is the power which would be its from the preamp. So we can measure the
delivered into a matched load having the noisiness of the preamplifier in terms of the
same impedance as the resistor. increase in the temperature of the resistor that
Another example: if a 50 resistor at 290K would be required to produce the same
was connected to the input terminals of a amount of noise. Thats the meaning of T1 in
noise-free receiver, with a 50 input imped- Fig. 4.2. A perfect preamp would have a noise
ance and a bandwidth of 25kHz (Fig. 4.1), temperature of zero K, ie it would contribute
what would be the noise voltage? The noise no extra noise. But nothing in this universe is
power from (1) is: noise-free; any real device has an equivalent
noise temperature greater than zero K. The
1 38 10 -23 290 2500 = 1 00 10 -17 W
higher its equivalent noise temperature, the
The voltage across 50 is obtained simply more noise the device is making. The word
V2 equivalent is there simply to remind us that
from P = , or in this case V = PR , and is: the temperature of the resistor at the input
R
doesnt physically increase and indeed there
neednt be a resistor there at all.
(1 00 10 -17
)
50 = 22 4nV The practical advantage of thinking in terms
of equivalent noise temperature is that it form
Thats right, nanovolts thats what we a common basis for measuring random noise
mean by small signals in modern VHF and arising from just about anything, from a
UHF receivers! GaAsFET to a galaxy. There are many different
Resistors are not the only electronic devices types of noise in electronic devices, and most
which generate noise. Imagine a preamplifier of them are not truly thermal in origin nor
(Fig. 4.2) connected between the 50 resistor is galactic noise, for that matter. This doesnt
4 3
ASSEMBLING YOUR STATION
matter so long as they can all be expressed as Electron Mobility Transistor. The reason was
equivalent amounts of thermal noise. In that 432MHz was too low a frequency for the
practice, people dont always include the word HEMT device!
equivalent, and after a few more paragraphs Compared with other ways of measuring
Ill probably start to forget it as well. noise, noise temperature has some very useful
Because much electronic noise is strictly properties. It is independent of bandwidth
speaking non-thermal, the equivalent noise (unlike noise power, which is proportional to
temperatures of most electronic devices (apart bandwidth), and its also independent of the
from resistors) are seldom the same as their gain or loss of the device generating the noise.
physical temperatures. For example a modern Most useful of all, the noise temperatures of
MOSFET amplifier at room temperature may different components in a total system can be
have a noise temperature as low as 50K at added directly to give a system noise tempera-
144MHz, and GaAsFETs can go even lower. ture.
On the other hand, a very noisy device can
have a noise temperature well above 1000K SYSTEM NOISE
without glowing white-hot! This means that TEMPERATURE
you cant always cure a noisy device by
cooling it towards absolute zero, because that CASCADES OF AMPLIFIERS
will only reduce the thermal part of its noise. So far weve looked at a simple RF preamplifier
The reason cooling works for parametric fed from a 50 resistor (Fig. 4.2). Weve seen
amplifiers used in radio astronomy is because that the noise temperature of this combina-
most of their noise is thermal. Cooling also tion is the sum of the noise temperature of the
works for GaAsFETs in the right circuit but resistor (TR, equal to its physical temperature
dont get too excited, because well shortly see because it is a resistor) plus the equivalent
that low noise isnt everything as far as VHF noise temperature T1 of the preamplifier. This
and UHF DX is concerned. sum is actually a simple form of system noise
Unlike true thermal noise, which is uniform temperature.
across the whole electromagnetic spectrum, Now lets add to the system another stage
noise from an active device like a transistor (or whose noise temperature is T2 (Fig. 4.3). We
a galaxy) is frequency-dependent. Although can calculate the noise temperature of this
noise temperature may be taken as constant new system by finding out what excess noise
within an amateur band, it will probably be the second stage contributes, in terms of
different on another band. Since there are excess equivalent noise temperature at the
probably several different noise-generating input. The excess noise temperature is simply
mechanisms operating within the same T2/G1, where G1 is the power gain of the first
device, the noise temperature may well stage, so the system noise temperature TSYS is
increase at both the HF and the LF end of its given by:
useful frequency range. At the 432MHz T2
preamplifier noise-measuring session in a TSYS = TR + T1 + (2)
G1
recent Moonbounce Convention, it was
instructive to see a cheap consumer GaAsFET Dividing T2 by G1 is called referring T2 to
almost beating an expensive microwave High the input. Power gain G1 must be a pure ratio
Fig. 4.3. A
simple noise
analysis
4 4
SYSTEM NOISE TEMPERATURE
(eg a power gain of 15 times), and not in dB. The noise temperature T1 of the first stage
We can continue in this way, adding on adds directly to the system total, but the
extra stages (Fig. 4.4A) and referring their situation is different for the later stages. If
noise contributions back to the input, so that: they were completely noiseless, TSYS would
come down to just (TR + T1). And TSYS can be
T2 T3 T4
TSYS = TR + T1 + + + + ... (3) brought close to this lower limit if the last
G1 (G1G2 ) (G1G2G3 )
term in equation (2), namely (T2/G1) can be
The noise temperature T3 of the third stage made small. We already know how to do this
is referred back to the input by dividing it by in practice: by either using a low-noise device,
the cumulative gain up to that point (G1 x G2 which makes T2 small, or by increasing the
of the two previous stages) and so on. Once all gain G1.
the noise temperatures of the later stages have
been referred back to the input in this way, CUMULATIVE GAIN
their contributions to the system noise In a multi-stage system we can minimize the
temperature are simply added together. effect of later stages by using low-noise
There are limits to how low the system circuitry or adding more gain. In practice its
noise temperature can get. Go back to the usually the gain effect which dominates: in
simplest two-stage system (Fig. 4.3) and look equation (3) the cumulative effect of G1, G2 etc
at equation (2) in more detail. For a start, the on the bottom lines of the fractions can
system noise temperature cant be less than TR, quickly make the noise contributions of later
because thats the noise temperature of the stages extremely small. To show this effect in
source outside the receiver where the signal is practice, lets put some typical performance
coming from. This is the most basic limit, figures into the system of Fig. 4.4A, to give
which many people dont appreciate: there is Fig. 4.4B. In a real multi-stage system we can
always some noise coming into the receiver, say that the last stage is the rest of the
along with the signals. It forms part of the system, and give it a noise temperature, so we
overall system noise and, no matter how low- no longer need the fictional noise-free receiver
noise your receiver, that external noise is still which appears in Figs 4.1 4.4A. Then using
there. equation (3) we have:
4 5
ASSEMBLING YOUR STATION
NOISE FROM LOSSY DEVICES So you can see that 09dB of cable loss has
We also need to know how to deal with such changed the noise temperature from 180K at
loss-introducing devices as cables, attenuators the receiver to 288K at the top of the cable.
and filters. These degrade the system noise And 668K of that is due to thermal noise
temperature, mainly by attenuating the generated in the cable itself.
wanted signal. But theres also another, less
obvious source of noise: thermal noise NOISE FACTOR AND NOISE FIGURE
generated in the resistive part of the losses. Noise temperature is the fundamental concept
This second effect is little-known and often underlying every aspect of noise in receiving
neglected, but it can be important in the ultra- systems. So how are the more familiar-
low-noise systems we are dealing with here. sounding terms noise factor and noise
The complete formula for the effect of lossy figure related to noise temperature?
components on system noise temperature is: Noise factor F is related to noise temperature
T by a standard definition [1]:
T = TF + (L - 1) TP (4)
T
where: F = 1+
290
T is the noise temperature at the input of
the lossy component 290K is about 17C or 62F, close to room
L is the loss expressed as a power ratio (ie L temperature, so noise factor is a measure of
= 1 for a lossless system, going up to how the device or system noise compares with
infinity for infinite loss) thermal noise under typical conditions of use.
TF is the cumulative noise temperature of all This definition gives the perfect system a
following stages, referred to the output of noise factor of 1, and all practical noise factors
the lossy device are greater than 1.
TP is the physical temperature of the lossy Even more familiar than noise factor is noise
device. figure, which is simply noise factor expressed
The first term in equation (4) is the attenua- in decibels:
tion term and the second is the term for extra
T
thermal noise, which depends on the physical NF = 10 log10 1+ (5)
290
temperature of the lossy device. The latter
term is small, except in extremely low-noise Since the minimum possible noise factor is
systems. Note that we have been assuming 1, the minimum possible noise figure is 0dB.
that the losses are resistive and occur in a Unfortunately, noise factor and noise figure
matched system. If the losses are solely due to are easily confused, and theres little need to
an impedance mismatch, only the first term become familiar with both. In this book we
4 6
SYSTEM NOISE TEMPERATURE
use either noise temperature (T and always At 144MHz, flat ground is a fairly good
with its units of K) or noise figure (NF and reflector of radiation and a correspondingly
always with its units of dB). poor emitter, ie the ground acts mostly like a
Noise temperature and noise figure each mirror. So a horizontal antenna sees either
have their own distinct advantages. The noise cold sky above the horizon or the sky image
temperatures of cascaded stages add up in a reflected by the ground, and the warm ground
very simple way, as weve seen, whereas the itself makes only a small contribution to the
NFs dont. The handy thing about NF is that antennas equivalent noise temperature.
the attenuation of a lossy component (in dB) Pickup of man-made noise may be much
can be added to the NF of the following stage more important than thermal or galactic
to give a new, larger NF for the two stages noise, but so long as the man-made noise is
combined. To see how, lets go back to the virtually random this too can be expressed as
example I gave a few paragraphs ago, about an equivalent noise temperature. The noise
the effect of a length of cable with 09dB loss. temperature of a 144MHz antenna can
If you convert the initial noise temperature therefore range from less than 200K in
(180K) into NF, and do the same for the final satellite work to well over 1000K in an urban
noise temperature (use the more accurate environment full of man-made noise. For
value of 2882K), youll find that the NF has terrestrial working on 144MHz, 200K is a
increased by precisely 09dB. But theres a reasonable minimum estimate for the antenna
catch: this trick only works accurately if the noise temperature.
lossy component is at the standard physical At higher frequencies, galactic and man-
temperature of 290K. made noise tend to decrease but the ground
Finally, theres one more serious problem and nearby objects become poorer reflectors
about noise figure. It can only deal with noise and better emitters of thermal noise. At 1GHz
in receivers and receivers are not the whole and above, the noise temperature of an
story. antenna pointed at the ground will be quite
close to the physical ground temperature. A
ANTENNA NOISE TEMPERATURES normal horizontal antenna will see approxi-
The receiver is only one part of the receiving mately half warm ground and half cold sky, so
system as a whole. An equally important its noise temperature will be about 150K. For
contributor to the system noise temperature is design purposes, you could also reckon on a
the noise picked up by the antenna, along minimum noise temperature of 150K for
with the signal. This antenna noise can be 432MHz antennas aimed at the horizon. With
brought into the system noise analysis by a little care, you can use ground noise to
giving it an equivalent noise temperature. estimate receiver performance (see box on
The noise temperature of an antenna (TANT) page 4-15).
is determined by the RF environment that it At lower frequencies, man-made and
can see, not by its physical temperature. For atmospheric noise become considerably
example a 144MHz beam pointed at the sky greater than at 144MHz or 432MHz. Although
for satellite work may be picking up only a the pops, crackles and lightning flashes from
very small amount of galactic noise, which many different sources tend to merge together
can be represented by an equivalent noise into something very much like random noise
temperature as low as 150-200K [2]. At the (like hand-clapping applause from a large
other extreme, if the antenna is picking up crowd), a good noise-blanker to remove the
thermal noise from its earthly surroundings at worst excesses can be a distinct advantage at
physical temperatures of 270-300K, that will 70MHz, 50MHz and below. At 50MHz, the
also be its noise temperature. The usual minimum galactic noise-temperature which
situation is somewhere in between, and your antenna sees, either directly or mirrored
depends rather upon the frequency. in the ground, is about 4000K [3], and subur-
4 7
ASSEMBLING YOUR STATION
ban man-made noise on top of this brings a Noise from any practical receiver must degrade
typical 50MHz antenna noise temperature up the signal-to-noise (s/n) ratio of a weak signal
to more like 300,000K [4]. Typical figures for to some extent, so the choice of receiver noise
70MHz would be about 3000K for galactic temperature can be framed in terms of how
noise and 150,000K including man-made much degradation of s/n is allowable.
noise. Lets try a few numbers. Consider a 144MHz
receiver connected to an antenna with a noise
USABLE SENSITIVITY temperature of 200K, the typical minimum
Knowing that system noise temperature can figure suggested earlier. If the receiver noise
be divided into two contributions, from the temperature is also 200K (NF = 228dB), the
antenna and from the receiver, we can now system noise temperature is 400K, and the
tackle the question: noise power is doubled. Thus the signal/noise
How much sensitivity do we really need at ratio of an incoming signal would be degraded
VHF/UHF? by 3dB, compared with the un-achievable
In Days of Old, when RF devices were noisy noiseless receiver. Fig. 4.5 shows the relation-
and the VHF/UHF bands were empty, the ship between the receiver noise temperature
answer was clear: we always needed more and s/n degradation for a range of antenna
sensitivity, simply to improve the chances of noise temperatures. Clearly, by aiming for the
ever hearing a signal at all. Modern low-noise lowest possible receiver noise-temperature, the
devices can give us all the sensitivity we need, s/n degradation can be made very small. But
and more. Yet we can no longer afford to aim these improvements in weak-signal perform-
for the highest possible sensitivity, because ance will also imply higher front-end gain,
that also entails excessive front-end gain which will ruin the strong-signal performance.
which brings overload problems (Chapter 5). Unless you carefully balance sensitivity
So we have to choose what system noise against front-end gain, on a crowded band
temperature we want, and design our receivers you can easily end up hearing less, not more!
to achieve it. If we aim for a receiver noise temperature
Very well, what noise temperature do we which is equal to the minimum probable
want? Even with a perfectly noiseless receiver, antenna noise temperature, the scope for
the system noise temperature cannot possibly further improving the signal/noise ratio is at
be lower than the antenna noise temperature. most 3dB; more often it will be only 1-2dB.
noise 150K
14 200K
degradation.
Curves are for 12 300K
a range of ANTENNA
10 500K NOISE TEMP
antenna noise
temperatures (K)
8
from 50K to 1000K
5000K. 6
4
2 5000K
0
10 100 1000 10000
RX NOISE TEMPERATURE (K)
4 8
SYSTEM NOISE TEMPERATURE
Such a small improvement may scarcely be arent like that, of course. Their skirts are not
noticeable on the air, except at the very limits perfectly steep, and their passbands are round-
of weak-signal working such as moonbounce. shouldered and by no means flat. The noise
In other words, if the receiver noise tempera- bandwidth of such a filter is simply the
ture is as low as the minimum probable noise bandwidth of a brick-wall filter which would
temperature of the antenna, the receiver is let through the same total amount of noise
about as sensitive as it needs to be for terres- power. For most IF filters with reasonably flat
trial work. On the lower VHF bands, man- passbands and steep skirts, the 6dB band-
made noise can exceed galactic and ground width is a fair approximation for the purposes
noise by 10dB or more, so theres absolutely of calculating noise floor.
no need for very low noise figures on 50 or Notice that if we change the bandwidth, the
70MHz. noise floor changes too. Narrower bandwidths
Referring back to the typical minimum result in a lower noise floor, ie a quieter
antenna noise temperatures, the design-target receiving system. This is a very real effect, and
receiver noise temperatures and noise figures is largely responsible for the vast improve-
for ordinary terrestrial DX-chasing would thus ments in weak-signal communication obtain-
be: able by using CW.
In order to talk about s/n ratios in the usual
50MHz 4000K 12dB
units of dB above the noise floor, its conven-
70MHz 3000K 10dB
144MHz 200K 22dB tional to convert the noise floor into dB
432MHz 150K 18dB relative to one watt (dBW) or one milliwatt
(dBm), by the usual formulae:
These are not hard-and-fast rules, merely
design targets in round figures. When you set dBW = 10 log10(watts)
up your receiver, you will have an opportunity dBm = 10 log10(watts) + 30
to balance sensitivity against strong-signal
handling, to suit your own particular noise In our previous example, 69 x 10 18W
environment. Thus you can optimize the corresponds to 1716dBW or 1416dBm.
overall performance of your receiving system. One final point about noise floor, until
Chapter 5, is that its definition can depend on
NOISE FLOOR whether were talking about weak or strong
A concept related to sensitivity is noise floor. signals. When were asking whether a weak
This is simply the power level corresponding signal is audible, the definition of noise floor
to a receiver or system noise temperature, must include the actual antenna noise
calculated by P = kTB. For example, the noise temperature; we calculate noise floor using a
floor of an SSB receiving system meeting our noise temperature of (TRX + TANT). But when
target noise temperature of 200K for 144MHz were thinking about performance with strong
is: signals (Chapter 5), an accurate definition of
1 38 10 -23 200 250 = 6 9 10 -18 watts noise floor is less important; for that applica-
tion the noise floor is frequently calculated
To make this estimate, I had to assume a assuming a nominal antenna noise tempera-
bandwidth of 2500Hz. The noise bandwidth ture of 290K.
used in these calculations is not quite the
same as the more familiar 6dB filter band- POWER, GAINS
width. Imagine a filter with a perfectly flat AND LOSSES
passband and perfectly steep sides, sometimes In this short section well assemble the rest of
called a brick wall filter because thats what the ideas we need in order to think about
its passband would look like: straight up, flat path- loss capability. When it comes to string-
top, and straight down. Practical IF filters ing together a series of power gains and losses,
4 9
ASSEMBLING YOUR STATION
4 10
POWER, GAINS & LOSSES
sensitivity (ERS) is very similar to ERP, in that it of calculating station performance we can
combines the receivers noise-floor power assume that a minimal SSB signal will require
level with the antenna gain. a s/n ratio of at least +3dB in a bandwidth of
25kHz, whereas a minimal CW signal is
{ }
ERS(dBW ) = 10 log10 k( TRX + TANT )B (dBW ) copiable at 0dB in a bandwidth of 100Hz.
- (RX antenna gain, dBi) These figures are highly debatable because
they are crude mathematical representations
(8)
of the abilities of the human brain, so dont
The plus and minus signs require a bit of take them too literally. In particular, the
explanation. Although the kTB term looks bandwidths Ive just quoted are not necessar-
positive, it works out as a large negative ily the physical bandwidths of the IF or AF
number in dBW; you can receive signals way, filters. Although typical IF bandwidths for SSB
way below one watt. The antenna gain allows reception are reasonably close to the band-
you to receive even weaker signals, so it is width of the speech signal itself, the CW
subtracted to give an even more negative receiving bandwidth is more closely con-
number for ERS. As with EIRP, strictly speaking nected with how well the brain can distin-
we should say EIRS to remind ourselves that it guish a single tone in the presence of noise.
was calculated using antenna gain in dBi. This varies from person to person, and also
Another similarity between ERS and ERP is depends on the amount of practice one gets.
the possibility of misunderstanding what the Experience suggests that the above estimates
antenna and feedline are doing. A beam of aural s/n ratio and bandwidth are about
antenna will collect more of the RF energy right for many people, give or take a few dB. If
coming from the wanted direction, and hence you feel your ears dont behave that way, you
will improve the signal/noise ratio, but it can simply plug your own figures into the
wont actually make your receiver any more formulae. But do remember that were talking
sensitive. Although feedline loss doesnt come about copying very weak signals, which is
into the ERS equation (8) because we meas- where CW really shines compared with SSB.
ured TRX at the antenna terminals, dont forget As the signal/noise ratio improves the two
that it can seriously damage your ERS as well modes become equally readable, and it then
as your ERP. becomes more a matter of personal preference.
In order to obtain a desired s/n ratio, the
SIGNAL/NOISE RATIO: gain/loss budget has to look like this
WHAT CAN WE REALLY HEAR? EIRP path loss EIRS = s/n (dB) (9)
The ability to copy a weak signal depends
primarily on the signal/noise ratio. This is Path loss is outside our direct control, but
simply the difference in dB between the each of us can do something about the ERP
received signal level PRX and the noise floor of and ERS of our own station.
the receiving system. Weak-signal working
obviously implies signals right down in the PATH-LOSS CAPABILITY
noise, ie signal/noise ratios around zero dB. The path-loss capability of your station is the
Since the signal is received against a back- sum of its transmitting and receiving perform-
ground of receiver noise plus antenna noise, ance.
both of these contributions to the system Path-loss capability (dB) = EIRP (dBW)
noise temperature must be included when
EIRS (dBW) (10)
calculating the receiving-system noise floor.
A machine can measure signal/noise ratio, In other words, path-loss capability is the
but how well do human beings copy weak path loss that you and an identically equipped
signals? The answer ultimately depends on station could overcome between you. Echo-
our own ears and brains, but for the purposes testing on moonbounce gives a very clear
4 11
ASSEMBLING YOUR STATION
illustration of path-loss capability. You press could work, and the other is to study how you
the key, and about 2 seconds later you hear could improve your station.
the result in your own receiver maybe!
Coming back to earth, lets work out the WORKING RANGE
path-loss capability of a minimal 144MHz DX Forecasting your working range is not an exact
station. Lets give it a single long Yagi, a 100W science, unless youre prepared to work with
amplifier, a sensitive receiver and fairly low long-term statistical trends, because path loss
feedline losses. Inserting some reasonable on any mode of propagation depends on what
figures, the path-loss capability of this station we loosely call conditions and of course
comes out as follows. these can change markedly from day to day,
On the transmit side, we have 100W = or even minute to minute. With these limita-
+20dBW. Assume the feeder is about 12m of tions in mind, how far do you think a path-
URM67 (RG-213U), with a loss of 10dB at loss capability of 217dB will get you? Accord-
144MHz. A medium-length 144MHz Yagi will ing to the so-called standard tropospheric
have a gain of about 15dBi, so the EIRP is propagation curves for 144MHz, you might
+20dBW 10dB + 15dBi = 340dBW expect your SSB signals to disappear into the
noise at about 500km. More to the point, in
On the receive side, assume a receiver NF of average conditions from an average site you
20dB; add the feeder loss of 10dB (it adds could hold an SSB QSO with a similarly-
directly to the NF) and convert to noise equipped station about 400km away at a
temperature. Work this backwards through tolerable s/n ratio of +10dB. Although these
equation (5) in the earlier part of this chapter, figures are pretty vague, they do accord with
and you should get an answer of 289K. To be experience in normal conditions, as distinct
optimistic, assume a rather low antenna noise from an opening or those dead-band condi-
temperature of 200K to give a system noise tions when it feels like a struggle to work
temperature of 489K. Convert this to the across town.
noise-floor power in a 25kHz SSB noise As a flight of fancy, what about moon-
bandwidth using equation (1), and convert to bounce with a path-loss capability of 217dB?
dBW; if youre still with me, the answer is Well, the minimum path loss for 144MHz
close to 168dBW. Subtract the antenna gain EME is 2515dB, so your SSB echoes would
of 15dBi, and the EIRS comes out at 183dBW. come storming in with a s/n ratio of... oh
So the total path-loss capability is: dear, 345dB below the noise. OK, so you
EIRP EIRS = 34dBW (183dBW) = 217dB already knew that 100W and a single Yagi
wont let you chat with the Ws on 144MHz
If you get confused with juggling all the EME, but now you know precisely why not!
plus and minus signs, just remember a few
obvious facts. The overall path-loss capability CHANGING TO CW
should be large and positive. The receive All the previous calculations were for SSB. If
noise-floor should be large and negative, but you rework the receiver sensitivity part with a
the double minus sign turns this into a bandwidth of 100Hz instead of 2500Hz, youll
positive contribution to path-loss capability. see an immediate improvement of
Each of the following also makes a positive
2500
contribution: higher power, lower feedline 10 log10 = +14dB
100
loss, and higher antenna gain.
Try working out the path-loss capability of If you also take account of the fact that
your own station, and list the contributions of weak-signal CW will get a message through at
all the individual parts of your RF system. a much lower s/n ratio than SSB, youve
Having done so, you have two ways of using gained the best part of 20dB in path-loss
this information. One is to ask how far you capability. In average tropo conditions this
4 12
STATION IMPROVEMENTS
might extend your working range by another station performance is your receiver. Chapter
200km. So switching to CW and being 5 is about optimization of receiver perform-
prepared to use weak signals will considerably ance, although as far as sensitivity is con-
increase your potential for working DX, cerned weve already seen that the achievable
especially when you remember that youre improvements will ultimately be limited by
more likely to find the Eastern European DX antenna noise. The best way towards good
on CW than on SSB. receiving performance is not through micron-
Changing to CW has also made consider- sized gallium arsenide devices in the preampli-
able inroads into that 34dB deficit on the fier; its through large-sized aluminium-alloy
moonbounce path. All you need now is a little devices in the sky.
help from a mega-station like W5UN, whose The remaining area for improvement is in
array of 48 long Yagis boasts a gain about transmitter power, but you need to balance
17dB greater than your own 15dBi. Throw in a this against your own receiving capability and
few dB of ground-reflection gain at one that of other stations. Suppose youre testing
stations moonrise or moonset, and youre in your new 400W amplifier in flat band
with a shot at working the States! conditions on 144MHz, and someone 350
The final question is: does it work? Take a 400km comes back to your CQ call. Lets
listen around 144010MHz at moonset, on a imagine that you have a middle-aged com-
weekend when the moons been high in the mercial 144MHz transceiver with a front-end
sky. If you have a single Yagi of reasonable noise figure of perhaps 810dB, while his
size, the chances are that you will hear W5UN receiver has a noise figure of 2dB as I recom-
and other big stations like his. All around the mended earlier. Hell probably be giving you a
world, there are dozens of single-Yagi stations report of 5-and-2 but youll only be copying
running 100W of CW who now have the EME him at about 3-and-1. Your 400W amplifier
QSL cards to prove that path-loss budgets do has turned you into an alligator all mouth
work! and no ears! To make proper use of high
power, you need a good receiving system too.
STATION IMPROVEMENTS Unless you actually enjoy low-power (QRP)
Path-loss budgeting can give you a very clear operating for its own sweet sake, its equally
idea of what you ought to be able to hear and frustrating to be an elephant all ears and no
work. And if the performance of your station mouth in which case Chapter 6 should
is regularly falling short of what it should be, provide some inspiration.
then maybe yours is one of those stations
with equipment which isnt working properly. MEASUREMENTS AND
The key to making your station work properly RECORDS
is to understand how the various parts From the system-engineering point of view,
contribute to the overall station performance. you need to keep performance of your station
And thats the second major role of path-loss in balance so that its about equal in all areas
capability calculations. and without any notable weaknesses. As you
The antenna gain and feeder loss each appear progress, make measurements and record
twice in your stations path-loss capability, them, either in your station logbook or
once on transmit and once on receive. So preferably in a more detailed technical
theres no point in having superb equipment notebook. Making notes and keeping records
in the shack and then connecting it to a dipole will be a recurring theme throughout this
in the roof space or an HB9CV on the chim- book, for two very important reasons
ney. The main improvements that can be made (1) So that you can understand what youre
to any station will always be in the antenna doing without having to keep everything in
and feedline Chapter 7 tells all. your head; and
The next most important aspect of overall (2) To keep track of station performance by
4 13
ASSEMBLING YOUR STATION
repeating the same tests over the months and amplifier (+6dB improvement over 100W).
years. None of these improvements may seem very
Chapter 12 describes a wide variety of worthwhile on its own, considering some of
system tests using home-made instruments. the difficulties involved. But look adding up
Two additional tests for the antenna and all the bits and pieces, youd have improved
receiving system are measurements of ground your path-loss capability by at least 13dB!
noise and sun noise see the box oppposite. Even 1dB can make all the difference to the
Ground noise measures receiver performance, success or failure of a weak-signal QSO, so
and sun noise tests your receiver and antenna every improvement will make a real, positive
together. difference to the DX performance of your
station in the long term.
PLANNING YOUR Finally, remember that lots of other people
IMPROVEMENTS are reading this book too. If both you and
You can plan the best way to upgrade your they upgrade your stations along the lines Ive
station, by estimating how the various suggested, well all be working a lot more DX.
changes would improve your path-loss
capability. By looking at your station from the REFERENCES
system point of view and doing a few simple [1] IEEE Standard Dictionary of Electrical and
calculations, youll soon begin to see whats Electronic Engineering Terms. Wiley-
going on where the weaknesses are, where Interscience.
you can make real improvements, and where [2] R. E. Taylor, 136/400MHz radio-sky maps.
best to spend your hard-earned money. Proc IEEE, April 1973.
The biggest rewards come first, which is
[3] Ray Rector WA4NJP, 6 meter EME. Proc
encouraging if youre a beginner. They come
22nd Conference of the Central States VHF
from simple changes like putting up an
Society. ARRL, 1988. ISBN 0-87259-209-X.
outside beam, improving a deaf receiver, and
increasing power from 310W to around [4] Jim Fisk W1DTY (W1HR), Receiver noise
100W. These improvements will raise a figure, sensitivity and dynamic range what
beginners station to one with moderately the numbers mean. Ham Radio, October 1975.
successful DX performance. After that, further
improvements become more difficult and
expensive; instead of increasing your path loss
capability by leaps and bounds, youre
reduced to picking up a dB here and there. But
youll still find it worthwhile especially if the
bug has bitten!
Bearing in mind the need to keep every area
of station performance in balance, the next
steps onwards from 100W and a small Yagi
should probably come in the following order
(1) Improve your antenna; doubling its size
will gain you about 25dB on both transmit
and receive.
(2) Improve the feeder; every 1dB reduction
in loss is as good as 1dB more antenna gain, or
1dB off your receiver NF.
(3) Finish improving the receiver sensitivity
(not much more to be gained here).
(4) Last of all, upgrade to a 400W power
4 14
STATION IMPROVEMENTS
4 15
ASSEMBLING YOUR STATION
4 16
RECEIVERS
&
LOCAL
OSCILLATORS
5 1
RECEIVERS & LOCAL OSCILLATORS
5 2
RECEIVERS & LOCAL OSCILLATORS
bandwidth of the IF circuits, a modern quite a lot about whats important in front-
transverter is very similar to a transceiver end design. At a deeper level, you can really
front-end for the same band. The advantage of get to grips with the subject by working
transverting is that it provides all the facilities through the example calculations. Calculation
of the HF rig on the VHF or UHF band, and is the key to modern RF design, and a scien-
allows extra bands to be added at modest cost tific calculator or home computer can be just
compared with a range of new single-band as useful as a soldering iron. If you do decide
rigs. The disadvantage is that receiver perform- to delve into the worked examples, dont start
ance may be compromised, especially with using a computer program straight away.
strong signals. Unless you work through the detailed calcula-
The receiver front-end has to do two things. tions yourself, thinking about what youre
One is to amplify and frequency-change the doing, youll never truly understand the
weak signals without introducing too much subject.
extra noise; the other is to cope with any
unwanted off-channel signals, however strong SENSITIVITY AND NOISE
they might be. When all the incoming signals TEMPERATURE
reach the IF filter, only the wanted signals In Chapter 4 we looked at system noise
pass through together with any on-fre- temperature as part of a stations overall path-
quency interference, of course. All other loss capability. Since the lowest possible noise-
unwanted signals should fall outside the filter temperature for the whole receiving system is
passband and cease to bother you. But before limited by the noise seen by the antenna, we
they are filtered out, strong off-frequency concluded that a noise temperature
signals may already have caused trouble by equal to the lowest probable antenna noise-
overloading an earlier stage in the front-end, temperature would result in an acceptably
perhaps generating spurious signals on the small signal/noise degradation. To be on the
wanted frequency. The IF filter cannot remove safe side, we made estimates of antenna noise
these signals, so they can interfere severely temperatures for each band which excluded
with the weak DX that you want to hear. any contribution from man-made noise, and
The problems arise when off-frequency these became our design targets for receiver
strong signals become amplified to such high noise temperature.
levels that they drive one or more stages in 50MHz 4000K 12dB
the front-end into some kind of overload. A 70MHz 3000K 10dB
receiver with good strong-signal performance 144MHz 200K 22dB
432MHz 150K 18dB
needs to have
. But a
certain amount of front-end gain is essential
in a sensitive receiver. We thus have to make a Now lets try a more complete example of
trade-off between sensitivity and strong-signal noise and gain analysis than we managed in
handling a very careful trade-off indeed, Chapter 4. The system we shall consider is
aimed at achieving the optimum balance of G4DDKs Suffolk transverter described in
performance in operating conditions. Chapter 8. The Suffolk is a transverter for the
The optimum balance between weak-signal European 144146MHz band, with a tunable
and strong-signal performance is not a matter IF of 2830MHz.
of luck. It has to be achieved by careful We begin by drawing a block diagram of all
planning and , based on sound under- the separate stages of the front-end, with the
standing, optimized by trial calculations, and signal flowing from left to right (Fig. 5.3).
backed up by performance testing. Dont worry about what kind of circuitry goes
This chapter can be read at two levels. Just inside the blocks; what were concerned with
by skimming over the surface, you can learn at present is simply the performance of each
5 3
RECEIVERS & LOCAL OSCILLATORS
Fig. 5.3. Block diagram of the Suffolk transverter for 144-146MHz (receiver portion only) with
input data
block in terms of noise temperature and The formula for converting noise figure (NF)
power gain. This we write in the space above to noise temperature (T) comes from equation
each individual block. Since the Suffolk is a (5), Chapter 4, and is:
transverter, its block diagram ends at the HF
NF
T = 290 antilog10 1
10
transceiver.
The performance data for the individual
stages of the Suffolk come from such sources as To calculate the system performance, start at
manufacturers data sheets and G4DDKs the IF filter at the right-hand end of the block
measurements on prototypes. As we shall see, diagram, and work backwards stage-by-stage
the HF rig may be a crucial factor, but for the towards the antenna, using equations 15 of
present we only need assume that it has good Chapter 4. To avoid getting confused, work
sensitivity. right through in terms of noise temperature
The block diagram is shown in Fig. 5.3, with alone, and then go back and convert all your
the data for each individual stage above. system results to NF. Similarly, calculate the
Below the block diagram will go the results for cumulative front-end gain first in ratio terms,
the system. If you want to follow though this and then go back and convert all the results to
worked example, draw up a similar worksheet decibels. (A more detailed approach to this kind
for yourself. All the data were originally of worksheet is given in references 14.)
obtained in decibel units of noise figure and The results of the noise/gain analysis are
gain. For our analysis, these will have to be entered below the block diagram as shown in
converted to noise-temperature and gain ratio. Fig. 5.4. If youre following this worked
5 4
RECEIVERS & LOCAL OSCILLATORS
5 5
RECEIVERS & LOCAL OSCILLATORS
STRONG-SIGNAL
PERFORMANCE
5 6
STRONG-SIGNAL PERFORMANCE
monic region and the difference product will the signal frequency to the intermediate
be down at LF. Even a simple filter will frequency. The signal is at f1, f2 is the LO and
separate these widely spaced groups of signals. (f1f2) is the IF; in this particular case the filter
As I mentioned earlier, your receiver relies on is intended to select only the IF. The mixer
second-order intermodulation to convert from stage has to be a non-linear device, or else it
wouldnt work; but its non-linearity can often
let in a host of further intermodulation
problems.
If the two strong signals at frequencies f1
and f2 are increased in strength, another set of
IPs appears. These are the third-order prod-
ucts, so called because they involve mixing
between three signals [5]. These three signals
can be totally independent, or the same effect
can be generated by two parent signals,
counting one of them twice. Since these two
frequencies can either add or subtract, the full
range of possible third-order IPs from two
signals includes:
third harmonics (f1 + f1 + f1) and (f2 + f2 + f2)
sum products (f1 + f1 + f2) and (f1 + f2 + f2)
difference products (f1 + f1 f2) and (f2 + f2 f1).
If both of the parent frequencies f1 and f2 are
Fig. 5.5. Spectrum analyser display of third- close to the wanted frequency, the third-order
order intermodulation between signals on sum products must be somewhere up in the
144240MHz and 144260MHz. The third-harmonic region and wont trouble us
intermodulation products appear on
further. But the two difference products
144220MHz and 144280MHz
containing a minus sign are somewhere close
Fig. 5.6. Higher-order intermodulation to their parent frequencies. For example, if f1 =
products, on an expanded frequency display. 144240MHz and f2 = 144260MHz, their
Note the uniform frequency spacing, equal to third-order IPs fall on 144220MHz (f1 + f1 f2)
that of the parent signals and 144280MHz (f2 + f2 f1). Fig. 5.5 shows
how the two parent signals and their IPs
would appear on a spectrum analyser display.
If the receiver front-end is allowed to
generate these in-band spurious signals, were
in real trouble. No practical kind of RF
filtering can take them out, because theyre
too close to the frequencies we want to listen
to. If one of these IPs just happens to fall on
the frequency of a weak wanted signal, it will
cause QRM. And if that can happen, then sure
enough it will right when you least want it.
In the example above, the two strong
signals were spaced 20kHz apart, and note
how their third-order IPs were evenly spaced
20kHz above and below them (Fig. 5.5). As the
strength of the two equal parent signals
increases, the strengths of their third-order IPs
5 7
RECEIVERS & LOCAL OSCILLATORS
(also equal) come up three times as fast, will wipe out even more of the band. Remem-
increasing by 3dB for every 1dB increase in ber, this is not necessarily the fault of the
the two parent signals. So the onset of inter- transmitters: all your intermod problems can
modulation can seem quite sudden. Further be generated right inside your own receiver.
increases in the strengths of the parent signals With the low end of the band full of strong
will produce new generations of higher-order SSB and CW signals during a contest or an
IPs. Thrashing through the algebra [5] shows opening, intermodulation in the average
that the odd-order (3rd, 5th, 7th, etc) IPs are commercial VHF/UHF receiver front-end can
the ones that cause the trouble, because some be ghastly. Listening on the band is like
of them can fall close to the parent signals. wading waist-deep through a shifting swamp
These higher-order IPs appear above and of inter-modulated mush.
below their parents, at uniform frequency Yet it is possible to build VHF/UHF front-
spacings (Fig. 5.6). Fifth-order IPs require ends that have virtually no noticeable inter-
stronger parent signals than third-order before modulation, even at open contest signal
they rise above the noise, but then they levels. Compared with an ordinary receiver,
increase in strength five times as fast as the the effect is startling. The strong signals are
parent signals; and pro rata for seventh- and just as strong, but there are also spaces
higher-order products. Severe high-order between them where the weak DX comes
intermodulation can thus appear very sud- through in the clear. Also, many local signals
denly indeed, covering the band with spurious that appear broad on an ordinary receiver
signals. High-order intermodulation in turn out to be quite respectable and it
transmitters is the cause of splatter, as becomes equally obvious whose signals really
discussed in Chapter 6. are broad!
Intermodulation requires at least two strong
parent signals. If either one goes off, all their GAIN COMPRESSION
IP offspring will disappear. But relief will be Gain compression occurs when a strong
short-lived, because the remaining strong incoming signal drives some stage in the
signal can intermodulate with any other front-end so hard that it can barely produce
strong signal that comes along, to make a new any more output. The gain of the saturated
set of third-order IPs. And if the first signal stage is thus reduced, taking with it all other
comes back, making three in all, there will signals including the wanted one. In an
now be four (work it out) different sets of IPs. extreme case, gain compression simply makes
So far weve only been talking about single the receiver collapse in embarrassed silence
frequencies, ie steady unmodulated carriers. whenever a strong signal comes on. Gain
On SSB, its far worse than that. An SSB signal compression requires only one strong signal,
itself contains several different frequencies at unlike intermodulation which requires at least
once, all of varying amplitudes. When these two.
intermodulate, the effect is to broaden the Gain compression occurs with any kind of
bandwidth of the signal. If the intermodula- signal provided its strong enough, and is
tion takes place in the transmitter, the particularly easy to recognise when produced
broadening is real and youll have to ask the by a strong carrier. When the carrier comes
other station to fix it (see Chapter 6). But on, the background noise goes down, together
what happens if two perfectly clean SSB with all other signals. If the carrier is keyed,
signals intermodulate in your receiver? As well the background noise appears as negative
as each seeming broader, their in-band third- Morse code, the dits and dahs being the
order IPs together wipe out a further slice of silences in between.
spectrum, totalling typically 20kHz. A more
severe case of intermodulation generating LOCAL-OSCILLATOR NOISE
appreciable levels of higher-order products Reciprocal mixing is a large-signal effect
5 8
STRONG-SIGNAL PERFORMANCE
5 9
RECEIVERS & LOCAL OSCILLATORS
resulting output to a suitable level. Examples spurious mixing products and noise in low-
of transceivers using this technique are the old level circuitry. Or does it?
TS700 and the IC201. Mixer VFOs offer many Phase-locked loop systems certainly can
opportunities for oscillator noise. The mixer give very low noise sidebands if designed and
cannot be run at too high a level without constructed with care. But a bad PLL can be
producing spurious products, and at the very noisy indeed. In addition to all the noise
output of the mixer the carrier level can sink problems that can beset ordinary oscillators,
perilously close to the circuit noise. Any noise can easily be coupled onto the VCO
subsequent amplification of the low-level control line, leading to FM noise sidebands.
carrier also brings up the accompanying noise. Also, some designers have skimped on the
Other transceivers of similar vintage, such normal precautions for generating a good-
as the FT221 and FT225, use similar frequen- quality VCO signal, relying instead on the
cies for the HF VFO and the VHF crystal phase-locked feedback loop to clean up the
oscillator, but add them using a phase-locked noisy, hum-laden, drifting signal for them. It
loop (PLL). A PLL is essentially a feedback seems to work, if you look only at the carrier,
control system. In Fig. 5.8B the output signal but theres a penalty in the form of high levels
is generated by a voltage-controlled oscillator of noise sidebands. Its a shame when just a
(VCO), whose frequency is also heterodyned little more care in design and layout could
down to HF using a VHF crystal-controlled have produced a very quiet LO.
oscillator. The phase of this HF signal is More recent VHF and UHF LOs are usually
compared with that of the VFO, and any error of the digitally-synthesized PLL variety. These
voltage is fed back to control the VCO. Tuning too are capable of good noise performance
the HF VFO thus makes the VHF VCO follow but digital synthesis offers the careless de-
in step, at a constant frequency difference set signer even more opportunities to get things
by the crystal oscillator. That sounds fine: it wrong! The digital parts of these synthesizers
produces the same output frequency as if the generate high-speed, high-level switching
VFO and the crystal oscillator had been mixed transients which can stray onto the VCO
directly, and the PLL avoids the problems of control line unless careful precautions are
5 10
STRONG-SIGNAL PERFORMANCE
taken. The frequencies fed to the phase offs between various aspects of performance,
detector in digital loops are quite low, and any plus extreme care in the detailed design and
leak-through to either the supply or control physical layout [6].
lines of the VCO can cause severe modulation A good local oscillator needs to have a noise
sidebands. These digital modulation sidebands level of 100dBc/Hz at 1kHz offset, rolling
may sound like distinct carriers or like noise, away to 140dBc/Hz at 10kHz and ultimately
depending on the method of synthesis and to better than 160dBc/Hz at offsets of 100kHz
the source of the leakage. If unwanted LF or more. Well return later to the significance
signals get onto the VCO control line, its very of these figures.
difficult to get rid of them because filtering To give synthesized local oscillators their
affects the ability of the loop to lock quickly due, they have brought us the same standards
and stay locked as you tune or switch fre- of frequency stability and readout accuracy at
quency. The design of low-noise PLL synthe- VHF/UHF as we enjoy at HF. And although 99
sizers thus requires some very careful trade- digital memories are something of a mixed
blessing (you still have to remember whats
stored where!), a synthesizer with dual VFO
Fig. 5.9. Mixing processes capabilities and intelligent programming of
A: Normal mixing: wanted signal + LO = IF signal the RIT and split-frequency facilities can be a
B: Reciprocal mixing: strong off-frequency signal +
LO noise = IF noise real boon for cracking pile-ups and keeping
skeds. Yes, we have seen some progress during
the past decade. Its just a pity that much of
the progress as far as oscillator noise is
concerned has been backwards.
So much for our digression into oscillator
noise. Now then, what about reciprocal
mixing?
RECIPROCAL MIXING
Normally, the mixer is fed with a strong signal
from the LO, and frequency-shifts a fairly
weak wanted signal into the passband of the
selective IF filter. But the mixer will do the
same for any pair of signals whose frequencies
differ by the IF. Reciprocal mixing gets its
name because the roles of LO and signal are
reversed: the strong signal is an unwanted one
off-frequency, and the weaker signal is part of
the LO noise sidebands.
For example, if youre listening to
144250MHz with an IF filter centred on
10700MHz, your LO is tuned to 133550MHz
(Fig. 5.9A). The intention is that any signal
appearing on 144250MHz is mixed with the
LO and delivered into the IF passband.
Reciprocal mixing occurs when a strong
incoming carrier mixes with LO noise side-
bands, delivering that noise into the 107MHz
IF and raising the background noise level. Fig.
5.9B shows the example of a carrier on
5 11
RECEIVERS & LOCAL OSCILLATORS
144240MHz which mixes with the LO noise Our true objective is not to cure overload,
sidebands around 133540MHz, so that the but to make sure that it hardly ever happens,
reciprocally-mixed LO noise interferes with even at the strongest amateur-band signal
the wanted signal being received on levels. For all practical purposes intermodula-
144250MHz. tion is non-existent if the products never rise
The offset between the wanted frequency significantly above the background level of
and the strong unwanted signal 10kHz in system noise. Similarly, the receiver can be
our example is the same as the offset considered immune to gain compression or
between the LO carrier frequency and the reciprocal mixing if these effects never
interfering part of its noise sidebands. This significantly change the noise level. In other
means that the severity of reciprocal mixing words, what you dont hear wont trouble
will have the same frequency dependence as you.
the LO noise spectrum. Thus the main This leads directly to the idea of spurious-free
problems usually occur at close-in frequency dynamic range (SFDR). The bottom end of all
offsets corresponding to the LO noise pedestal the dynamic ranges is the noise-floor power
(Fig. 5.7) although, at very close frequency level of the receiver (Chapter 4). The top end
spacings, reciprocal mixing is often masked by of each individual dynamic range is the power
leakage of the unwanted signal through the IF level of the strongest off-frequency signal that
filter. A particularly bad LO with a high the receiver can tolerate without the particular
broadband noise-floor may cause reciprocal overload effect becoming noticeable. SFDR is
mixing problems across the entire amateur simply the difference in decibels between
band. these two power levels. The overall SFDR of
On the air, the effect of reciprocal mixing is the receiver is the smallest of its various
opposite to that of gain compression. When a dynamic ranges.
strong carrier appears, the background noise Although spurious-free dynamic range
goes up. In a transceiver, the LO is used for relates directly to the things you hear when
both receive and transmit functions: thus, if using a receiver, it needs some further defini-
the receiver suffers from reciprocal mixing, tion to turn it into something measurable. Ive
the transmitter will also radiate noise already defined noise floor in Chapter 4.
sidebands. It can be very hard to distinguish Another name for this power level is minimum
reciprocal mixing in your own receiver from discernible signal (MDS), though this is an HF-
the corresponding problem in someone elses inspired term VHF/UHF DXers have very
transmitter. The only way to be sure is to test different ideas about copying really weak
your own receiver independently, using a low- signals! Formal definitions for the points at
noise signal source. which the various overload effects become
significant have been proposed by Wes
DYNAMIC RANGES Hayward, W7ZOI, and these lead directly to
Weve seen in this Chapter that there are methods of measuring the various types of
three main problems of front-end overload. SFDR [8]. The W7ZOI dynamic range tests are
They are all different, so each problem starts based firmly on reality what you measure is
to be noticeable at a different level of strong also what you hear on the air so its not
unwanted signal. Two strong signals are surprising that they have become the standard
needed to cause intermodulation, but only for HF equipment reviews in QST and Radio
one to cause gain compression or reciprocal Communication. The test methods described by
mixing. Because intermodulation and gain Hayward were developed for HF receivers, and
compression are both due to some stage in the have needed further development for use with
front-end being driven beyond its linear very sensitive receivers at VHF and UHF; we
range, both problems can be cured by keeping will leave that topic to Chapter 12.
front-end gain as low as possible. Another aspect of dynamic range which is
5 12
STRONG-SIGNAL PERFORMANCE
5 13
RECEIVERS & LOCAL OSCILLATORS
INTERMODULATION
ANALYSIS
e
ord
ig
3rd
re
Pa
5 14
INTERMODULATION ANALYSIS
If we speak in terms of output power, were well as amplifying them it will produce yet
talking about the third-order output intercept more IPs of its own making. The two sets of
(IPOUT). The two are equivalent, but not equal: IPs will be on the same frequencies, and at
they differ simply by the small-signal power- small frequency separations or in wideband
gain of the stage in question. If an amplifier systems they will usually add in phase at the
with a power gain of 19dB has an output output of the second stage. If the input
intercept of 30dBm (30dB over one milliwatt, intermodulation intercepts of stages 1 and 2
ie 1W), its input intercept is (30 19)dBm, ie in Fig. 5.11A are IP1 and IP2, the system input
11dBm. On the other hand, if a diode-ring intercept is given by:
mixer with a loss of 6dB has an output 1 1 1
= +
intercept of 10dBm, its input intercept is 6dB IPSYS IP1 IP2
greater, ie 16dBm. Since amplifiers have gain, G1
their output intercepts always look bigger Lets look at this formula more closely. It
than their input intercepts, merely because may seem vaguely familiar, because its rather
theyre higher. Always check which one like two formulae youve seen somewhere
youre being quoted, especially if someones before. One is the formula for adding noise
trying to sell you something! temperatures in cascaded stages:
The third-order intermodulation intercepts
T2
of individual amplifiers, mixers etc can usually TSYS = T1+
G1
be estimated either from manufacturers data
sheets, or failing that by some simple rules-of- That was equation (2) from Chapter 4,
thumb which Ill describe in the Appendix to where I mentioned that the action of dividing
this chapter. Given the necessary information, T2 by G1 was called referring T2 to the
you can predict the intermodulation perform- input, so that it could be added directly to
ance of a multi-stage front-end, in much the the noise temperature T1 of the first stage. In
same way as we predicted its small-signal the intermodulation equation above, were
performance earlier in this chapter. referring IP2 to the input in just the same way
To predict intermodulation in a complete when we divide it by G1.
front-end, we need to know how IPs build up The other familiar aspect of the intermodu-
in cascaded stages. If two equally strong lation equation is that after youve referred all
signals are fed into the input of a two-stage the individual intermodulation intercepts to
system (Fig. 5.11A), the first stage with gain the input, you can then combine them like
will produce an amplified output, and also resistors in parallel. As you know, the resist-
some internally generated IPs. The next stage ance of a parallel combination is always less
will accept all these signals as inputs, and as than the lowest of the individual resistances.
5 15
RECEIVERS & LOCAL OSCILLATORS
Its exactly the same with intermodulation 10dBm limit imposed by the first stage.
intercepts: once youve referred everything to Conversely, no matter how robust the first
the input, the input intercept of the whole stage, IPSYS cannot exceed (2015) = 5dBm.
system is less than the lowest of the individual This is because of intermodulation in stage 2,
stage intercepts. Unlike the chain in the which is made worse by the 15dB gain ahead
proverb, this chain of stages is weaker than its of it in stage 1. Time after time, we keep
weakest link. coming back to the need to keep the system
Time to try some numbers. Fig. 5.11B puts gain down!
some values into the general system of Fig.
5.11A. Using the equation above: WORKSHEET METHODS
1 1 1 Earlier we used a worksheet (Fig. 5.4) to
= + analyse the noise temperatures, noise figures
IPSYS (10dBm) (20dBm - 15dB)
and gain distribution of the Suffolk 144MHz
This equation cannot be evaluated as it transverter front-end. Now lets use the same
stands. The intercept of the second stage can worksheet to analyse its third-order inter-
be referred to the system input simply by modulation performance. Fig. 5.12 shows the
subtracting the 15dB gain of the first stage, details, with typical intermodulation inter-
but power levels in dBm (dB relative to 1mW) cepts for each stage. The stage gains are
must be converted to milliwatts before they exactly the same as in the earlier noise
can be combined. So, more correctly: analysis, of course.
1 1 1 Since the Suffolk is a transverter, you cant
= + usefully analyse its intermodulation perform-
IPSYS (10mW ) (3 16mW )
ance in isolation from the receiver that
and: follows it. Well go further into this topic
IPSYS = 2 40mW or 3 80dBm towards the end of this chapter, and also in
Chapter 8 where G4DDK describes the Suffolk
Note that the result, 380dBm, is much in more detail. For now, lets assume youre
lower than either of the individual stage using one of the modern generation of HF
intercepts of 10 and 20dBm. transceivers with a 28MHz IPIN of +20dBm.
What happens if one stage has a much Having written down the input intercept for
higher intercept than the other (after theyve every stage, including the HF receiver, the
both been referred to the system input)? If the next step is to refer each intercept to the
second stage is the more robust, the system 144MHz input. Do this by subtracting the
input intercept still cannot be greater then the cumulative gain up to that point, remember-
ing that here were working in decibels: check
Fig. 5.12. To combine the stage intercepts into
Fig. 5.12. Intermodulation analysis of the
a system value, convert everything into
Suffolk transverter
5 16
INTERMODULATION ANALYSIS
milliwatts, and add all the values as if they experience on the bands. Assuming a decently
were resistors in parallel yes, it is tedious, calibrated S-meter (a big assumption!), with
but try doing it at least once with a calculator, 6dB per S-point and S0 at the noise floor,
just for the experience. absolutely no third-order intermodulation
The third-order input intermodulation should be noticeable until signals reach
intercept of the Suffolk front-end is thus almost S9+40dB. That is the standard of
52dBm. What else does the analysis tell us? intermod performance youre entitled to
As soon as youve referred each of the stage demand from a modern front-end.
intercepts to the input, you can see which are
the vulnerable stages. The mixer is the single RECIPROCAL MIXING
weakest link, but note that we have assumed As I remarked earlier, a receiver actually has
an HF receiver with an excellent IP3 of several dynamic ranges. What are the maxi-
+20dBm, which represents reasonable per- mum tolerable levels of local-oscillator noise
formance in the 1990s. Many older or budget before the overall spurious-free dynamic range
HF rigs have poorer intermodulation perform- becomes limited by reciprocal mixing rather
ance, and such receivers would limit the than intermodulation?
overall performance of the system. Therefore When used with a good HF transceiver, the
G4DDK has tried to protect the HF receiver by Suffolk has a predicted intermodulation-free
keeping the overall gain of the Suffolk to the dynamic range of 91dB in a 25kHz band-
minimum consistent with meeting the target width. So we want to ensure that no discern-
noise temperature. And although he hasnt ible reciprocal-mixing noise will be produced
spent an excessive amount on the mixer, hes unless the carrier responsible is at least 91dB
left the option of upgrading various parts of above the noise floor. So the LO noise level
the system as described in Chapter 8. must itself be 91dBc in a bandwidth of
2500Hz. What does that mean in terms of
PREDICTING dBc/Hz? The equation relating noise levels, in
DYNAMIC RANGES dB terms, for various bandwidths is:
B
INTERMODULATION N1 N2 = 10 log10 1 dB
B2
Another thing that the intermodulation/gain So 91dBc in 2500Hz corresponds to
analysis can predict is the dynamic range for 125dBc in a bandwidth of 1Hz.
third-order intermodulation. Recalling from An LO noise level of 125dBc/Hz would
Fig. 5.4 that the receiver noise temperature actually raise the noise floor level by 3dB at
was 196K, its noise floor in a typical SSB the signal levels were assuming, which would
bandwidth of 25kHz is 1417dBm (from the be totally unacceptable for weak-signal
kTB equation in Chapter 4, which should be working. So we really need more like 135dBc/
getting pretty familiar by now). The spurious- Hz. At close frequency separations, thats
free dynamic range (SFDR), also known in this asking quite a lot, though it should be per-
context as the intermodulation-free dynamic fectly well achievable at offsets beyond a few
range or the two-tone dynamic range, is given kHz. And were still entitled to expect that LO
by: noise levels should roll away towards
2 160dBc/Hz and better at wider offsets.
SFDR = (input intercept - noise floor )
3
everything being measured in dB, so:
2
( )
SFDR = -5 2dBm - {-141 7dBm} = 91dB
3
Measurements on the Suffolk front-end
agree quite well with predictions, and so does
5 17
RECEIVERS & LOCAL OSCILLATORS
FRONT-END ANALYSIS
USING TCALC
Up to now, Ive taken you through the data you want to leave unchanged, and this
calculations of front-end noise/gain and can be recalled by a single keystroke.
intermodulation performance by hand,
because thats the only way to explain it TYPES OF STAGES
properly. Now is the time to introduce an TCALC requires you to label each stage, to
easier way of doing the same things. TCALC is help you remember which is which, and so
a microcomputer program which simply that TCALC knows what data to ask for. If you
automates the calculations Ive just been go back through Chapter 4, youll find that
describing. It is freely available as a BASIC there are really only three types of stage which
program listing, and also on disk for the BBC need to be considered.
Micro and IBM PC compatibles [9]. (1) Amplifiers have gain and noise tempera-
Noise/gain and intermodulation analyses tures, and the calculation for cascaded stages
using TCALC work exactly as I have already uses equation 2 from Chapter 4. Since this
described. You divide the system into a block analysis takes no account of input or output
diagram and type in the performance data for frequency, mixers can be treated exactly the
each block. TCALC then calculates the same as amplifiers. Diode mixers with nega-
cumulative and system performance, taking tive gain are no problem in this regard.
care of all the tedious details like conversion TCALC permits the labels AMPlifier or MIXer
between noise temperature and NF, or be- for this type of stage.
tween decibels and ratios. (2) Attenuators include any type of passive
By making your input as simple and device which has an essentially resistive
convenient as possible, TCALC frees your insertion loss. The calculation uses equation
attention to think about the real problem (4) from Chapter 4, and TCALC assumes a
the design of a receiver front-end. For exam- physical temperature of 290K.
ple, when TCALC asks you for the noise TCALC permits the label ATTenuator or
temperature or noise figure of an amplifier, it CABle for this type of stage. When entering
makes intelligent decisions about which one data, passive filters also use the ATT label.
you mean. Enter a number less than 25, and (3) Antenna denotes the end of the block
TCALC will assume you mean noise figure; a diagram. On encountering the label ANT,
larger number is taken to mean noise tem- TCALC asks for the antenna noise-tempera-
perature. Having made its decision, TCALC ture, adds it to the receiver noise-temperature
confirms what it thinks you meant and then and displays the system results. When analys-
continues by asking you for the gain. ing a receiver in isolation from its antenna,
When you have completed the calculation, the label END serves the same purpose.
you can then go through it again to see the
effects of making changes. To save you the NOISE/GAIN ANALYSIS
trouble of retyping, TCALC remembers all the The easiest way to describe how TCALC
5 18
FRONT-END ANALYSIS
Fig. 5.13.
STAGE PARAMETERS SYSTEM RESULTS
Noise/gain
T(K) NF(dB) G(dB)
analysis for
Stage 1 (rest of system)
the Suffolk
NF = 10dB 2610 10 0
transverter
using Stage 2 Type: ATT
TCALC. Loss (dB) : 1 3360.9 11 1
Compare with Stage 3 Type: AMP
Fig. 5.4 NF = 6dB Gain (dB) : 8 1397.2 7.65 7
Stage 4 Type: MIX
NF = 6dB Gain (dB) : 5.5 5821.9 13.24 1.5
Stage 5 Type: ATT
Loss (dB) : 6 24042 19.24 4.5
Stage 6 Type: ATT
Loss (dB) : 3 48258.8 22.24 7.5
Stage 7 Type: AMP
NF = 1dB Gain (dB) : 26 196.3 2.25 18.5
Stage 8 Type: END
performs a noise/gain analysis is to show you identity and gain of each stage is known to
the output for the Suffolk. Compare the screen the program, so it lists all the stages and asks
output in the panel above (Fig. 5.13) with the you for an intermodulation intercept where
worksheet of Fig. 5.4. appropriate, ie for AMP/MIX stages only. If
Notice that TCALC only produces a gain you wish to disregard intermodulation in that
analysis working towards the antenna, rather stage, you just press Enter.
than a display of relative signal levels working Having collected all its information, TCALC
forwards from the input as shown in Fig. 5.4. now produces a table of signal levels, the
This is because all the necessary information intermod intercept of each stage referred to
for a signal-level presentation of gain is not the input, and the system results including
available until the details of the last stage have noise floor and dynamic range. Fig. 5.14
been entered. However, a relative signal level shows the results for the Suffolk, which can be
display is available as part of the intermodula- compared directly with Fig. 5.12 (below).
tion analysis. Whenever it completes an analysis, TCALC
offers you the opportunity to go round again,
INTERMODULATION ANALYSIS switch to the other type of analysis, or quit.
Having made at least one pass through the
noise/gain analysis, TCALC offers the option OTHER POSSIBILITIES
of intermodulation analysis. By now, the TCALC is a simple program written in BASIC
Fig. 5.14.
Intermodulation STAGE PARAMETERS SYSTEM RESULTS
gain analysis T(K) NF(dB) Level(dB) IPi3(dBm)
for the Suffolk Rest of system IP3(dBm): 20 2610 10 18.5 1.5
transverter Stage 2 (ATT) 3360.9 11 19.5
using TCALC. Stage 3 (AMP) IP3(dBm): 18 1397.2 7.65 11.5 6.5
Compare with Stage 4 (MIX) IP3(dBm): 15 5821.9 13.24 17 2
Fig. 5.12 Stage 5 (ATT) 24042 19.24 23
Stage 6 (ATT) 48258.8 22.24 26
Stage 7 (AMP) IP3(dBm): 1 196.3 2.25 0 1
Whole receiver: 5.2
Noise floor (2.5kHz) = 141.7dBm
Dynamic range = 91dB
5 19
RECEIVERS & LOCAL OSCILLATORS
5 20
PRACTICAL FRONT-END
SYSTEMS
5 21
RECEIVERS & LOCAL OSCILLATORS
5 22
PRACTICAL FRONT-END SYSTEMS
worsened, reciprocal mixing is becoming the mixing performance, and then use your power
limiting factor in strong-signal performance. of choice. If it isnt good, dont buy it.
In the past few years there have been
marked differences in the ways that various TRANSVERTERS REVISITED
manufacturers and importers have tackled this If you already have an HF rig, the alternative
problem. Some have steadily improved their to buying a complete VHF or UHF transceiver
products. But others have done nothing, or is to buy a transverter or to build one from
have even brought out new equipment whose this book. VHF and UHF SSB got started in the
LO noise is worse than in older models! The sixties when HF SSB enthusiasts added
RSGB VHF Contest Code of Practice sets a transverters to their HF transceivers. This
target of 90dBc for all wideband spurious opened up the VHF bands to the advantages
products from the transmitter, and if applied of transceive operation, VFO control, stable
to LO noise this would stand our receivers in and selective receivers, and of course to the
very good stead too. That target was set at the advantages of SSB as a DX mode. But by the
end of the crystal-controlled AM era when late seventies transverters had acquired a bad
some particularly nasty SSB transceivers were reputation for receiver overload, and on
making their first appearance; the figure of 144MHz they have been almost entirely
90dBc was chosen because it was being superseded by single-band, single-conversion
routinely achieved by the older AM equip- transceivers for serious DX and contest work.
ment, mainly thanks to crystal oscillators and However, transverters have always been more
high-level Class-C stages. Its a lot more widely used than single-band rigs on the other
difficult to achieve that order of noise per- VHF/UHF/microwave bands, and now theyve
formance along with the kind of frequency even made a comeback on 144MHz. So what
mobility we require for todays operating went wrong with transverters in the seven-
conditions, but the technology does exist and ties, and whats changed since then?
the design principles are well established [5, 6, A transverter followed by an HF transceiver
7]. is a multiple-conversion system. The receive
So why cant we buy transceivers with truly signal path in the transverter is very like the
excellent VHF and UHF front-ends from the front-end of a normal single-conversion
Big Three manufacturers? In their HF models transceiver, except for one thing. The LO in a
they have recognised our requirements for transverter is crystal-controlled on a fixed
strong-signal performance as well as sensitiv- frequency, and the HF transceiver does the
ity, and high dynamic range is now a strong tuning. That means you cant put a
selling point. But at VHF and UHF all the narrowband IF filter directly after the trans-
recent design effort seems to have gone into verter front-end, so the HF transceiver front-
digital features like scanning, memories, voice end is vulnerable to overload from the
synthesizers and built-in packet TNCs. These amplified wideband output of the transverter.
are all very well in their way, but what about Transverters are worth a second look today
the radio? because the front-ends of commercial HF
As purchasers and users of amateur radio transceivers have improved a lot in the past
equipment, we need to resist the notion that few years, far more than those of VHF/UHF
other people arent entitled to put strong transceivers. Also the IF performance of HF
signals into our receivers; or that we shouldnt transceivers has always been a few steps
expect equipment manufacturers to produce ahead; few VHF/UHF transceivers even
clean local oscillators; or that any rig is nowadays can offer adjustable noise blankers,
acceptable merely because the importers have switchable filters with really good stopband
sold a lot of them. So read the equipment rejection, variable bandwidth, IF shift or
reviews carefully, look out for reports of poor notch filtering. All these are very handy on
sensitivity, dynamic range or reciprocal todays crowded VHF and UHF bands.
5 23
RECEIVERS & LOCAL OSCILLATORS
Lets use our techniques for analysing front- ADJUSTING FRONT-END AND
end performance to see what went wrong TRANSVERTER GAIN
with transverters in the seventies. At that There is no point in using more front-end
time, transistorized HF transceivers were just gain than is necessary to achieve reasonable
coming in and we all thought they were sensitivity in your particular RF noise environ-
wonderful. Actually, their strong-signal ment. Every decibel of unnecessary front-end
handling was dreadful: their third-order gain knocks 1dB off your input intermodula-
intermodulation intercepts were 30dBm or tion intercept. Considering how difficult it is
worse. A 1970s-style transverter had about to get IPIN into the region above 0dBm, its
25dB of gain, which dropped the system input surprising how many people then waste their
intercept at 144MHz to 55dBm or lower. efforts by failing to set up the whole receiving
With a system noise floor of 140dBm the system on the air.
dynamic range would be less than 60dB (Fig.
5.16). In practical terms, signals over about S9 Fig. 5.17. Fitting an adjustable attenuator to
anywhere on the band would cause havoc! the output of a preamplifier or transverter.
Now do the same analysis for a more (A) 15-30dB constant impedance design;
modern system. A typical NF for a modern (B) Wide-range non-constant impedance.
RV1 is a miniature cermet or carbon trimpot,
high-quality HF transceiver at 28MHz would and all other resistors are miniature carbon
be 10dB, with a third-order input intercept of film or metal-oxide.
+20dBm. Using a transverter like the Suffolk,
we can achieve a creditable 5dBm system
input intercept and dynamic range of 90dB,
along with the required noise figure of around
2dB. Thats not quite as good as a single-
conversion front-end designed with equal
care, but its a sight better than an ordinary
144MHz transceiver plus preamp. And unlike
a single-band 144MHz transceiver, the HF-
plus-transverter approach equips you for the
HF bands and allows you to add further
transverters for 50MHz, 70MHz and 432MHz.
Thats why we chose the transverter approach
for the designs in this book.
Turning to the bands other than 144MHz, a
target noise temperature of 150K at 432MHz
and above could be achieved using a trans-
5 24
PRACTICAL FRONT-END SYSTEMS
Any system you buy or build will probably at the beam heading that gives the lowest
start out with too much gain. Commercial background noise.
preamps need enough gain to produce some Signal/noise ratio is not easy to judge by ear,
improvement on the deafest transceiver. The so the easiest way to do it is to take advantage
same applies to transverters: even the ones in of the threshold effect of an FM detector. As
this book are designed with enough gain to youve probably noticed, FM signals tend to
suit a transceiver with a fairly high noise- be either virtually noise-free or very poor, and
figure at 28MHz. Your transceiver may be the change takes place over a narrow range of
more sensitive, so you can get rid of the excess incoming signal strength, or rather signal/
gain in the preamplifier or transverter. noise ratio.
Preamplifiers and transverters generally Setting-up is a simple three-step process.
have sufficient gain to perform well in a low-
noise environment. The target noise tempera- 1 Tune in a weak and slightly noisy FM signal.
tures given for each band in Chapter 4 refer to If you dont have FM capability, use SSB,
the lowest probable levels of antenna noise. preferably with the AGC off.
Your own RF environment may be signifi- 2 Reduce the front-end gain until the audio
cantly noisier, and this too will allow you to starts to become noisier.
eliminate some excess front-end gain. 3 Then increase the gain very slightly until
Some preamps already include a variable you no longer notice the deterioration.
attenuator at the output, so thats a good start.
If yours does not, you can easily fit one Thats all it takes to optimize your front-end
following the circuits of Fig. 5.17. The attenu- gain. If your system still gets crunched by
ator must not carry RF or DC power, so take strong signals, at least you know youve done
care if the preamp includes RF switching or your best!
has a DC power feed up the coax; if in doubt,
check with the technical guru at your local
radio club. Some transverters are likewise
equipped with a variable attenuator at the
28MHz output. The ones in this book happen
not to be, but each has a 50 attenuator in
the signal path which can be used for the
same purpose. More details are given in the
appropriate chapters.
To adjust your preamp or transverter gain
down to the minimum your system needs,
you simply turn the gain down until the
signal/noise ratio on a weak signal just begins
to deteriorate. By using a real signal on the
band, you are automatically taking account of
your own particular RF noise environment,
including galactic noise and the high man-
made noise levels typically encountered in
city environments and practically everywhere
on 50 and 70MHz.
To avoid the risk of reducing the sensitivity
too much, leaving your system unnecessarily
deaf when the band is quiet, you should do
the following test at the quietest possible
time, eg the small hours of Sunday morning,
5 25
RECEIVERS & LOCAL OSCILLATORS
5 26
APPENDIX
5 27
RECEIVERS & LOCAL OSCILLATORS
5 28
T R A N S M I T T E R S,
POWER
AMPLIFIERS
& EMC
6 1
T R A N S M I T T E R S, POWER AMPS & EMC
TRANSMITTERS, POWER
AMPLIFIERS AND EMC
by John Nelson GW4FRX
Hokey religions and ancient weapons are no match for a good blaster
(Han Solo in Star Wars)
6 2
CLEAN SIGNALS
INTERMODULATION
DISTORTION
In real life, however, examination of the
output of an SSB transmitter fed with two
audio-frequency tones in the above manner
would show many other frequencies as well;
indeed there would be a veritable family of
them (Fig. 6.2). Even very good SSB transmit-
ters will produce some power at spurious
frequencies, but only at low levels which
decrease rapidly as you tune away from the
main transmission certainly not enough to
upset other band users. On the other hand,
very bad transmitters will produce enough Fig. 6.2. Intermodulation products from a
real-life SSB transmitter grossly overdriven by
spurious outputs to cause interference across
the same tones as in Fig. 6.1. IPs beyond 9th-
the entire SSB area of the band for tens of order will undoubtedly be present too,
miles around. This is the sort of transmission further increasing the signal bandwidth
6 3
T R A N S M I T T E R S, POWER AMPS & EMC
between input and output. Youll recall from constantly varying interplay between the
Chapter 5 that if two signals at frequencies f1 range of frequencies which combine to make
and f2 are fed into an amplifier and increased a human voice. Low-order distortion is not a
in strength sufficiently to drive the amplifier major problem; for example third-order
beyond its linear range, a set of third-order intermodulation between low-frequency audio
intermodulation products appears. Their components gives rise to products within the
frequencies are (2f1 f2) and (2f2 f1) and wanted upper sideband, which are heard as
theyre troublesome because they lie close on distortion. They also fall within the area
either side of the parent frequencies f1 and f2. where the lower sideband would be. So an SSB
In our example, they will appear on transmitter with poor third-order intermodu-
1442497MHz and 1442527MHz and its lation performance but no other defects
important to note that their level will increase would produce very distorted speech with a
by 3dB for every 1dB increase in the two characteristically hard quality, but the
parent signals. Further increases in the transmission would not be particularly wide
strengths of the two tone signals will produce because the third-order intermod products do
fifth-order intermod products of (3f1 2f2) not fall very far from the frequencies which
and (3f2 2f1) on 1442487 and 1442537MHz, produce them in the first place.
increasing at an even worse 5dB/dB, followed Higher-order intermodulation is a different
by 7th-order products which increase even story. If the tones producing intermodulation
more rapidly, and so on as shown in Fig. 6.2 up to 9th order in Fig. 6.2 are replaced by
for odd-order products up to the 9th. The speech under the same conditions, the width
overall bandwidth of the signal on the of the transmission would be about seven
spectrum analysers screen has increased to times the original audio bandwidth which is
the original tone spacing of an utter disaster for the poor souls who
1000Hz. To make matters worse, there are happen to want to be using the band at the
bound to be even higher-order products off same time. When you hear the 15kHz-wide
the edges of the display! special-event station set up by the local club
Suppose we replace the two tones by speech one fine weekend, with Simons rig driving
whose bandwidth is carefully controlled to be Freds amplifier off Robs power supply into
200-2400Hz. In the absence of any intermodu- Bills antenna and with Walter (who else?)
lation, the output spectrum would simply bellowing into the microphone, youre
consist of a group of radio frequencies span- hearing significant levels of intermodulation
ning some 22kHz, the bandwidth of the products up to about thirteenth-order. Not
audio frequencies fed to it. Nothing else not good, but quite typical of what can happen
a scrap. In other words, with a suppressed when a station isnt put together properly as a
carrier frequency of 144250MHz all the RF . Whats even more irritating is that the
output would lie between 1442502 and wider the signal, the more splattery it sounds.
1442524MHz. This is because the levels of high-order
As you might be beginning to suspect by intermod products are extremely sensitive to
now, intermodulation broadens the band- the level of drive, eg 21st-order intermods
width of the SSB signal which is why its Bad (which are spreading something like 25kHz)
News for other band users. Bear in mind, by increase by 21dB for only a 12% voltage
the way, that the intermodulation process change in the AF signal. So the high-order
usually takes place in the driver and PA stage products snap on and off very suddenly, with
of the equipment, so the SSB filter bandwidth that characteristic and tiresome sound of
has very little effect in the overall bandwidth buckshot hitting a barn door. Few things in
of the eventual signal. Bandwidth is deter- life are as tedious as trying to winkle out some
mined by the various kinds of non-linearity weak DX when a station 30kHz up the band is
exhibited by the amplifier, together with the splattering away on top of you.
6 4
CLEAN SIGNALS
6 5
T R A N S M I T T E R S, POWER AMPS & EMC
peaceful coexistence. So the design target for radio amateurs using solid-state amplifiers
all serious VHF/UHF DXers should be some- dont know how to use them properly. Thats
thing like 80dB for intermodulation products one reason why the 144MHz band often
of seventh-order, and even greater suppression sounds more like the unacceptable face of
for higher orders. That isnt easy; but it is very 27MHz. The solid-state add-on PA is the worst
necessary, for the sake of other users of the thing to happen to amateur radio since...
band and for the sake of your own reputation. (insert whatever you think marked the end of
If youre planning to run the full legal limit, the Good Old Days).
these requirements add up to a powerful It seems to me that the reviewers of commer-
argument for a grid-driven tetrode linear cial linear amplifiers could do much more to
amplifier. Such an amplifier requires very little raise standards, by presenting more relevant
drive, allowing the solid-state exciter to operate test results and comment. The current practice
well within its own linear range: small clean is to present the results of static two-tone tests
signal in, big clean signal out. Mind you, even of low-order intermodulation performance;
with valve power amplifiers, a high standard of typical figures might be 30dB for third-order
intermodulation performance can only be products and 40dB below the wanted output
achieved by using high-quality power supplies for the fifths. Considered in isolation, these
and taking considerable care with the drive levels are not seriously anti-social because they
levels and loading. Commercial solid-state are close to the main signal. But it would be far
transceivers and power amplifiers are only more relevant to measure the higher-order
likely to reach levels of 80dB for intermodula- products, up to the nineteenth-order or
tion products higher than seventh-order if their thereabouts, to exercise the dynamic regulation
output power level is reduced considerably of power supplies using real speech signals, and
from the manufacturers specification (see also to comment on high-order intermodula-
panel opposite this page). Those containing tion performance under circumstances such as
contemporary hybrid PA modules wont even overdrive, mistuning or operating into badly
reach this level, as weve seen elsewhere in this matched loads. Dare I suggest that the majority
book, although hopefully the situation will of reviewers arent very clear about their criteria
change at some point as manufacturers make in this area?
better components available.
In making these criticisms of solid-state OUT-OF-BAND SIGNAL
amplifiers, its important to draw a distinction SUPPRESSION AND
between the inherent intermodulation OTHER REQUIREMENTS
performance of the linear amplifier and what Every transmitter ever made produces harmon-
happens when its maladjusted. The best valve ics and other out-of-band spurious signals, at
linear amplifier ever made will produce least to some degree. Although our power
splattery signals if its too lightly loaded, as levels arent in the same league as broadcast
will the best transistor amplifier if it sees a stations, we commonly use higher effective
high output VSWR. And any linear amplifier radiated power than the military and profes-
will splatter if its overdriven or has an sional services which also operate between our
inadequate power supply. Even so, some amateur bands. So we definitely do need to
awkward facts remain. Most commercial solid- take some precautions. The suppression of
state amplifiers for the amateur market are harmonics and other out-of-band spurious
extremely non-linear at the manufacturers signals can only be checked on a spectrum
rated power levels, and their performance can analyser; if you cant gain access to one, its
vary greatly between different samples of the best to assume that bandpass and low-pass
same model. Worst of all, they are chronically filtering is required as a matter of course.
overdriven in normal use. It may sound The place to put bandpass filters is in the
arrogant to say it, but the vast majority of exciter, and further details are given in
6 6
CLEAN SIGNALS
6 7
T R A N S M I T T E R S, POWER AMPS & EMC
Chapters 810 and 12. The only spurious phone gain turned to zero. Ideally the output
signals which power amplifiers can generate should be zero at all frequencies, but you may
(apart from intermodulation products, which well be able to hear synthesizer noise close to
are too close to filter out anyway) are harmon- the nominal output frequency. If you own
ics. So theres no point in placing a bandpass such a rig, not only will your near neighbours
filter after your linear amplifier; whats needed be suffering whenever you transmit but you
there is either lowpass or harmonic notch yourself will also be suffering from reciprocal
filtering, or perhaps both. The practice of mixing when your neighbours are transmit-
putting an extremely high-Q bandpass filter or ting and youre trying to receive weak signals
high-Q break at the output of a linear (Chapter 5). Unfortunately there is usually
amplifier is particularly dangerous, although little that can be done to improve transceivers
several authors who should know better have which suffer from this problem. The best plan
recommended such practices. The slightest is not to buy equipment which is known to
mistuning can result in an extremely high have a propensity to produce wideband noise.
output VSWR, which will certainly wreck the Many earlier Yaesu equipments such as the
intermodulation performance and can easily FT480R and some samples of the FT726 and
destroy solid-state output devices. If you need FT767 are prone to it, for example, and should
a filter after your power amplifier and you be avoided for serious VHF/UHF work. Icom
probably do always make it a lowpass filter. equipment is generally much better, and
Harmonic filtering is particularly important Kenwood somewhere in between but it can
for 50MHz transmitters since the second vary between different models from the same
harmonic falls in the FM broadcast Band II. maker, or even different samples of the same
Note also that the second harmonic of a rig.
144MHz transmitter falls in a military VHF
allocation used principally by aircraft; and CW TRANSMISSIONS
higher harmonics fall in the UHF TV Bands IV Any investigation of problems involving SSB
and V, as does the second harmonic of transmissions should begin by examining the
432MHz. And of course the third harmonics of CW signal before adding the complications of
144MHz and 432MHz fall right into our own modulation and intermodulation. Keyed
amateur bands. In all cases, its a whole lot wideband noise (otherwise known as DIY
better to be safe than sorry. Youll find designs aurora) is sometimes heard on a CW trans-
for lowpass filters in Chapters 810 and 12. mission; this is usually because of deficiencies
Turning now to other requirements of the or a fault in the prime movers synthesizer.
SSB transmitter, most of them can be dealt However, noisy screen supplies in tetrode
with at considerably less length. Frequency amplifiers have also been known to cause this
stability and resettability are clearly impor- effect: it can be recognized by a somewhat
tant, especially if youre proposing to make phasey quality and the fact that it is usually
successful MS and EME schedules; most gone within about plus and minus 20kHz of
modern commercial equipments are just the signal. In severe cases the close-in keyed
about adequate in this respect but only if the noise is only some 2025dB weaker than the
various crystal oscillators in the synthesizer main CW note! All newly-built power supplies
and counter have been adjusted accurately to need to be checked with an oscilloscope to
frequency within the last few months. Check ensure that they are clean and completely free
your own frequency errors before making from spurious noise or modulation, and then
skeds. checked again when theyre actually powering
Some synthesized equipments produce an amplifier and therefore under the influence
rather large amounts of wideband white noise. of RF fields.
Bad cases are easy to detect on another The keying characteristics of a CW signal
receiver just key the PTT with the micro- must be beyond reproach, since key-clicks
6 8
CLEAN SIGNALS
make a large part of the band unusable by sideband suppression are adequate; however,
neighbouring stations and remember that it is applicable only to MS operation in which
the UK licence also has something to say youve chosen your schedule frequencies and
about the quality of a CW signal. The effects transmission periods so that no-one locally
of key-click suppression measures can be needs to operate close to your signal. Keyed
evaluated before making on-air tests with the tone is definitely not recommended for other
locals, simply by connecting a low-frequency types of DX CW operation.
oscilloscope to the output of the diode However you do it, the keying waveshape
detector in the station VSWR meter and will be preserved through a linear amplifier
sticking the latter in the antenna feed. A but sharpened by an amplifier in Class C.
string of dots from the el-bug will display the Since UK licensing conditions relate to power
keying waveshape, and spikes and sharp edges supplied to the antenna, the lower efficiency
can easily be observed and corrected. Another in Class AB doesnt matter, and the extra on
useful test is to connect a variable resistor the electricity bill is a small price to pay for
across the key socket, and listen on another helping to keep the bands clean. I would
receiver for spurious signals at levels between strongly recommend forgetting about Class C
zero and full power. The keyed stages in some forever unless you need to use the amplifier
rigs (eg the FT221 and FT225) can become for occasional high-power FM working.
unstable during turn-on and turn-off, produc- Transverters have been discussed in a
ing transient sweepers which sound like receiving context in Chapter 5, and opting to
virulent key-clicks even though the use an HF transceiver as the prime mover is
waveshape looks perfect on the scope. likely to be a Good Move from the CW point
Unfortunately, many VHF/UHF commercial of view. Firstly, the keying characteristics of
equipments perhaps the majority have most HF equipments are considerably better
poor keying characteristics. The Yaesu FT225 than those of most of the radios mentioned
which with a muTek front end is amongst the above; the designers of HF rigs treat CW as an
very best 144MHz transceivers is notorious important mode which should be properly
for its very hard and clicky CW, caused by a catered for, whereas CW usually seems to be
combination of poor waveshape and the added to VHF and UHF multimode rigs as
transient instability noted above. The FT221 something of an afterthought. Secondly, most
and FT726 are also deficient in this area. The HF transceivers incorporate proper CW filters
Icom IC211/251/271 family has keying which, in the IF stages, which can be worth their
although hard, is acceptable when the trans- weight in gold when theres a good opening
ceivers are used by themselves. Some to Eastern Europe or during a good aurora.
Kenwood products, such as the venerable Most VHF/UHF multimodes just use the USB
TS700 series, have excellent keying character- or LSB filter for CW, which can cause you big
istics; later rigs like the TS711/811, TR9130 problems when the band is busy. Incidentally,
and TR751E are usually worse. Although a potential drawback of using transverters is
many modifications to keying circuitry have that the 28MHz low-level outputs of some HF
been published, they appear to have very transceivers arent as clean as they might be,
variable effects on individual equipments. so Chapter 12 includes a design for a suitable
High-speed keying for meteor-scatter is filter.
particularly tricky because rise and fall times Having looked briefly at all the requirements
have to be short, perhaps only a millisecond for transmitters, how well does commercial
or two. Good results may be achieved by equipment measure up? The short answer is
generating a keyed audio tone whose rise and probably reasonably well, especially if you
fall times are carefully controlled, and inject- use it sensibly and are prepared to make some
ing it into the mic socket. This technique is modifications. But in some areas (in the words
satisfactory provided the carrier and opposite of the archetypal school report) there is still
6 9
T R A N S M I T T E R S, POWER AMPS & EMC
room for decided improvement; manufactur- by trying to whistle at a constant level for
ers definitely could try harder. minutes on end, by the way; take a look at
Chapter 12 and save your breath.
BAREFOOT SSB Set the mic gain to maximum and apply the
Many of the problems with SSB transmissions tone or whistle. Note the power meter reading
arise from improper use of external linear and then back off the mic gain control until
amplifiers, but even with a basic barefoot rig power level reading reduces by about 10% (see
there is plenty of scope for poor-quality panel on page 6-7). You can then forget all
transmissions. As with add-on amplifiers, about this control, since its only function is to
most of these problems are caused by the establish the level at which your audio drives
operator yes, you trying to make the rig the transmitter. The microphone gain should
deliver more RF output than its capable of. not be used as a general-purpose RF power
You get home, unpack the box and plug it control; later in this chapter Ill describe some
all in; its sitting there in the shack all shiny proper ways of winding the power down.
and new and exuding a faint smell of silicone. So whats next? It might be worth just
You have a tune around yes, the beacon is checking the VSWR of the antenna before
S9 and there are some stations on the band. proceeding further. The designer of the solid-
Do you pick up the mic and give them a call state PA in the rig will have worked on the
to tell them about your new pride and joy? basis that it will always see a load impedance
No, not yet. You have some things to do first. close to 50. Things like the output filters and
First you need to establish the right micro- most importantly the ALC system will
phone gain setting for your rig, your micro- only work properly if the load impedance is
phone and above all your voice. reasonably close to what the designer in-
To do this you need a dummy load and a tended. A VSWR of 15 or less should be good
power meter. If you havent got a dummy enough for this purpose.
load capable of handling the peak power Nearly there now lets just have a quick
output of your station, youd better change think about the power supply. If its internal to
your name to Walter; I bet youre also one of the rig it should be all right, but how about
those people who whistle and go fitting a mains plug with integral overvoltage
Waaaaahlo on the calling frequency. Come spike protection? The microprocessors in some
on if you can afford to buy a VHF rig, or if rigs take offence at spiky mains supplies and
youre skilled enough to build an amplifier, are liable to crash, and although this is no
you have no reason not to possess a dummy problem (just switch off, wait a few seconds
load. Buy one, read Chapter 12 and make one, and switch on again) it can be tiresome and
borrow one for now but for the sake of the would be disastrous if it happened in the
rest of us whod like to use the band occasion- middle of a short Es opening to SV9. You
ally without the station were trying to work might also consider some filtering in the main
being obliterated by splatter, get hold of a 240V AC line to the shack. Some of the 1015
dummy load somehow and use it. amp filters obtainable cheaply at rallies seem to
Connect the rig to the dummy load via have excellent HF and VHF attenuation.
some form of power meter. It doesnt have to If your rig or more particularly an add-on
be a super-accurate laboratory instrument solid-state amplifier requires an external
anything will do at this stage because youre power supply, you do need to give the subject
mainly interested in a relative measurement. some careful thought before firing up. The
Chapter 12 contains plenty of ideas for power supply will need to be in the region of 12
meters. You should also have some form of 138V and it must have excellent regulation
single-tone or two-tone source in order to set right up to the maximum current drawn on
your drive levels using a constant audio speech peaks, which is where the regulation
signal. Theres no need to collapse your lungs matters the most. Assuming youve bought or
6 10
CLEAN SIGNALS
built such a supply, make sure that the leads Having checked all those things, now is the
which connect it to the rig are hefty enough time to go on the air and ask the locals who
not to drop any volts when high currents are know your voice what they think of it. Your
drawn. Unless your power supply is really best plan is to ask someone to note your signal
rock-steady, you and your solid-state PA will level and then to measure the width of your
find yourselves in deep trouble with other signal (see panel above). At the same time, ask
users of the band. your friend to measure your carrier suppression
6 11
T R A N S M I T T E R S, POWER AMPS & EMC
if he can. Having got your answers, make a VHF and UHF operators seem to manage quite
note in your log. This means that from time to well with power levels somewhere between
time you can ask that same station for a 25 and 25 watts, and when the band is in
routine test (which you are required by your good enough shape they can work many
licence to do, by the way) and you can also hundreds of kilometres. Equally, Chapter 4
check the effects of any changes you might placed a high-power amplifier last in the list
make, such as adding a power amplifier. of ways to improve path-loss capability. So
If youve done everything mentioned above, why go to all that trouble and expense?
your transmission shouldnt be over-wide; There are various levels of answer to this
most modern commercial rigs have reasonable question. From the operating point of view, as
intermodulation performance unless some- we pointed out in Chapter 3, high power is no
thing is amiss. If a new rig is seriously bad, substitute for good operating but sometimes
take it back and ask for another. If its a you need both. If you have to join a pile-up,
second-hand rig, try checking the driver and its better to get through decisively and be
PA standing currents you can usually do this done, rather than sit there contributing to the
by removing a link and putting a meter across QRM until your luck turns. From the con-
the pins or connectors. Other than that, one structors point of view, a well-made valve
possible cause of peculiar-sounding transmis- amplifier is a particularly satisfying project
sions is RF feedback see later. because there is no other adjunct to the
Just to recap, what points do you need to station which is so easy to make whilst adding
pay attention to when running your prime so much to the stations potency. A home-
mover on its own? built power amplifier represents a very good
The match to the antenna (the VSWR) so return on the investment in time and money
that the final stage, ALC and output filters can required to make it.
work properly. There are also serious technical reasons for
The mic gain too much and the transmis- wishing to run high power at VHF and UHF.
sion will sound nasty, although it may not First of all, some modes of propagation at these
actually spread. Dont try using it as a power frequencies need high power in order to be
or drive control. exploited at all. More accurately, whats needed
The power supply, especially if its external is high effective radiated power, ie both large
to the rig. antennas and high power. The challenge of
moonbounce is that it is only just possible,
HOW MUCH POWER DO even for amateurs who are prepared to do
DXERS REALLY NEED? everything thats necessary to bridge the half-
So far Ive mentioned power amplifiers more million-mile path which includes running
or less in passing, but now the time has come the UK legal limit of power, or more if you can
for us to do some serious thinking in depth obtain a Special Research Permit for this kind
and detail about the various factors associated of work. Taking another example, there is still a
with these devices. First of all, what do we great deal of work to be done with ionoscatter
mean by an amplifier in this context? and troposcatter on the VHF/UHF bands, and
Basically, one or more interconnected units those modes need lots of power to exploit
whose function is to take the transmitted them. Another interesting point is that high-
signal from the prime mover and deliver a ERP stations seem to do much better than
larger and more powerful replica of that signal lower-powered stations during auroral open-
to the antenna. At present in the UK we may ings, in a way which is not wholly consistent
not use a PEP greater than 400W, as measured with what would be predicted by reference to
at the antennas driven element. theory and textbooks.
It is pertinent to ask why anyone might Or maybe its simply because high-ERP
need that much power. After all, over 90% of operators try harder? In Chapter 1 we made
6 12
CLEAN SIGNALS
6 13
T R A N S M I T T E R S, POWER AMPS & EMC
WHAT KIND
OF AMPLIFIER?
6 14
WHAT KIND OF AMPLIFIER?
so the manufacturers can get away without ers might eventually produce a range of
any reference to the amount of intermodula- replacement RF power modules with im-
tion distortion produced by their so-called proved intermodulation performance.
linears at their rated output powers. To Although the 10W PEP level may be
make matters worse, some manufacturers and adequate for a beginner, a dedicated QRP
their dealers completely abandon any claim to operator or a casual DXer, you and I probably
professional respectability by specifying their regard 10W transmitters merely as drivers for
amplifiers in terms of input power the volts something much bigger. Again, let me remind
and amps taken from the supply rather than you that if you want to go after the real DX, as
RF output power, presumably because it makes distinct from merely accepting whatever DX
the product look more powerful and hence comes your way, you need more than 10W at
more attractive. Given that truly linear your command. Indeed, more than 100W. At
amplifiers cannot be much more than 45% these power levels, you may as well forget
efficient, its obvious that a so-called 500- solid-state and start thinking about valves or
watt (input) amplifier may well produce little vacuum tubes as they say in the United
more than 200W of RF flat-out, and probably States.
less than 200W PEP at acceptable intermod At this point, especially if you cut your teeth
levels. on solid-state components, you may experi-
Youve already seen my opinions on ence a sharp stab of pain; your instinctive
acceptable intermod performance, and how response may be that those weird thermionic
test results should be reported. This is an area things are emphatically not for you and that
where someone preferably national societies financial outlay is called for. Before you raid
like RSGB, ARRL and DARC should lay down the coffers, I ought to point out that there are
some standards for reviewers and advertisers, very few commercial valve power amplifiers for
for two reasons. One is that claims made by 144MHz or 432MHz which will reach the UK
some advertisers suggest that either they legal power limit with good intermodulation
should be gently led away by nice men in performance. The only ones Id currently
white coats or their advertisements should recommend are the American Tempo range
pointedly not be accepted by respectable which now use 3CX800A7 triodes. Commercial
magazines. The other is that there are maga- amplifiers such as the Dressler, Nag, Heatherlite
zine reviewers who themselves seem none too etc which use a single ceramic tetrode of the
clear about what to expect from an RF power 4CX250/350 variety will not repeat not
amplifier. produce the UK legal limit at anything like
acceptable intermodulation levels, for reasons
GOING FOR THE LIMIT which will become blindingly obvious in the
Chapters 810 include designs for transistor rest of this chapter. And theres no point in
amplifiers up to about 10W PEP output but anyone, especially equipment reviewers,
as G3XBY makes very clear in Chapter 9 pretending that they can. Single-4CX amplifi-
more powerful solid-state linear amplifiers of ers can perform tolerably well at lower power
high quality are strictly for the dedicated levels (although some still possess remarkable
experimenter who is prepared to spend some deficiencies which lead you to believe that
serious money. The 10W PEP amplifiers their designers never so much as glanced at the
featured in Chapters 810 are based on no- valve data sheet) but legal-limit they definitely
tune RF power modules (except on 70MHz for are not.
which no modules are available). In spite of This means that for socially acceptable legal-
the mediocre intermodulation performance of limit power output you only have two choices:
these things, such an approach at least offers buy a Tempo, or build your own. Im sorry to
less prospect of really appalling signals. have to spell it out again, but whatever some of
Theres also some hope that the manufactur- these companies proclaim in their advertise-
6 15
T R A N S M I T T E R S, POWER AMPS & EMC
ments, and despite the best efforts of some an overview of the available triodes and
reviewers to collude with them, the single tetrodes for legal-limit VHF and UHF working.
4CX250 or 4CX350 simply will not do the Ill list some of their advantages and disadvan-
business. tages, and discuss in detail how to use them.
Valve amplifiers for the UK legal limit are
not especially difficult to build. There are TRANSMITTING VALVES
several reputable designs for 144MHz and AND BASES
432MHz which usually present little difficulty Well take the ceramic tetrodes first, since the
from the RF point of view. However, to obtain majority of both home-brew and commercial
the performance of which the valves are amplifiers capable of something approaching
capable, some care needs to be taken with the legal limit of power use them. The first so-
various aspects of the power supply; Ill be called external-anode tetrode, given the type
delving into that side of things in Chapter 11. number 4X150A, was developed by the
Valve amplifiers for lower powers are also not American Eimac company in 1947. As with
difficult, although it is tempting to suggest Eimac products to the present day the initial 4
that if youre prepared to go to the trouble of implies a tetrode and the 150 gives the anode
getting a largish valve linear together, you dissipation in watts. X indicates an external
may as well build one thats capable of the full anode, meaning that the majority of the anode
legal limit anyway. surface is external to the valve so that it can be
The vast majority of modern designs for cooled by a stream of air blown over it. The
high-power VHF and UHF amplifiers use letter A simply means that it is the first
forced-air-cooled ceramic triodes and tetrodes production version of that particular valve.
of American origin or design. Since many More recent valves have a four-digit EIA
younger amateurs brought up on solid-state coding as well as Eimacs own 4CX number.
have no background in thermionic devices at The 4X150A founded a dynasty of external-
all (and the oldsters knowledge about receiv- anode devices. The 4X250B was introduced in
ing and small transmitting valves is not all 1953, and was similar to the 150A except that
that relevant) lets start at the beginning with it featured ceramic anode/screen insulation
6 16
WHAT KIND OF AMPLIFIER?
instead of glass, permitting higher operating to inadequate design or bad practice by the end
temperatures and hence the higher anode users. What is beyond doubt is that these
dissipation. In the following year the glass valves are considerably more expensive than
base insulation was changed to ceramic too, 4CX tetrodes to replace, and a 20,000-hour life
and the device became the familiar 4CX250B, is by no means unknown for a 250B run at full
with the C implying all-ceramic construc- ratings.
tion. And with only slight internal modifica- The 3CX series is numbered according to
tions, largely for production reasons, this anode dissipation, eg the 3CX100A5 has a
valve is still very much with us today. It dissipation of 100W. The 3CX100 family
remains an industry standard and users such stems from the glass-insulated 2C39 and is
as the Ministry of Defence, NATO, the Civil occasionally found in amplifiers for 432MHz,
Aviation Authority, the Home Office, the but even used in twos or threes the 3CX100 is
Foreign Office and the broadcasters all testify rather small for our purposes. The more recent
to the long life and ruggedness of the 4CX in 3CX800A7 is becoming popular in amateur
full-time professional service. service; several amplifier designs have ap-
Essentially the 4CX250B is a tetrode with an peared in print, and the 3CX800A7 has
anode dissipation of 250W and an upper superseded the pair of 8874s in the current
frequency limit of 500MHz. In general terms it range of Tempo VHF/UHF amplifiers. Rated
is delightfully easy to work with, and a pair to 300MHz, this valve will perform well at
of 4CXs is an excellent choice for a full-legal 144MHz and even at 432MHz with some
144MHz or 432MHz amplifier. careful design and attention to detail. For UHF
service the coaxial-based 3CX800U7 would be
88 & 3CX SERIES TRIODES better still but at a price, alas.
Eimac produces VHF/UHF triodes as well as
tetrodes, and these are numbered either in the OTHER VALVES
3CX series (rather like the 4CX family) and/or Other transmitting triodes and tetrodes appear
given a 4-digit code. on the surplus market from time to time, and
The 8873, 8874 and 8875 are high-mu power are at least worth a mention here if only to
triodes intended for use in zero-bias Class-B prevent you from repeating other peoples bad
amplifiers; they differ only in their anode experiences.
construction and hence their method of Starting with the tetrodes, and working
cooling and anode dissipation. The 8874 is like upwards in anode dissipation, glass-envelope
a 4CX tetrode in that it requires axial-flow double tetrodes such as the QQV06-40A (aka
forced-air cooling; it has an anode dissipation 5894 or CV2797) and QQV07-50 are far too
of 400W. The 8874 requires the same order of small for serious power amplifiers. They were
anode voltage and current as a 4CX250B and never intended for Class-AB1 use and in
can produce rather more output power. Several consequence their intermodulation perform-
American commercial amplifiers such as the ance is not good; they can also be unstable at
earlier Tempo models have used 8874s and the some drive levels, especially if incorrectly
like, often in pairs, but there have been mounted through the chassis. This valve
relatively few VHF/UHF designs for home family dates back to 1953, and most of the
construction. It is interesting to note that these ones you hear on 144MHz sound like original
valves appear to be seldom used in professional examples! The 8122 family of ceramic tetrodes
equipment, and it has been suggested that the was introduced by RCA in the fifties as a rival
reason is their rather short life. Users of 8874s to the Eimac products. A few commercial HF
have often complained of useful lives of less amplifiers used them and they sometimes
than 1000 hours, with a long period of poor crop up at rallies for reasonable prices. On
performance because of low cathode emission. paper they look an attractive alternative to
It is not clear whether these problems are due 4CX250Bs but in practice they seem to have
6 17
T R A N S M I T T E R S, POWER AMPS & EMC
6 18
WHAT KIND OF AMPLIFIER?
6 19
T R A N S M I T T E R S, POWER AMPS & EMC
6 20
WHAT KIND OF AMPLIFIER?
6 21
T R A N S M I T T E R S, POWER AMPS & EMC
1
drop across the resistor represents the back
pressure. The sum of the back pressures in the
0.8 80
WATTS
6 22
WHAT KIND OF AMPLIFIER?
figures in the data sheet suggest that a ing 100 cubic feet per minute, whereas we said
4CX250B presents about 082" of back pres- that we needed 128. Thats always the way of
sure to an airflow rate of 64 cubic feet per it: any blower capable of generating enough
minute. For a two-valve amplifier you require pressure for the 4CX series of valves will
double the airflow rate but the back pressure deliver far more than the minimum airflow
stays the same. requirements, even for two valves. In any case,
The problem with small blowers is that they plenty of cooling air is a good thing. Profes-
arent very good at generating the pressure sional transmitter manufacturers tend to allow
required to deliver enough air to cool a 4CX, something like a factor of 3 in this area to
which is where so many people come unstuck. cater for things like low mains, minor block-
Going back to the electrical analogy, a blower ages in the anode cooler and so on; in this
is a bit like a power supply with a high output particular application about 40cfm would be a
impedance: the output voltage looks fine reasonable minimum figure to shoot for.
without any load but it sags as soon as you ask However, valve life depends very much on the
it to deliver some current. Conversely, operating temperature, so using a VBL5/3
choosing a blower on the basis of the gale it blower with a pair of 4CX250Bs would be a
creates when discharging into free air is like very good move from that point of view.
choosing a power supply on the basis of its When youre doing initial tests into a dummy
short-circuit current; not very useful. Unless load you may well want to run the valves at
the blower is designed to generate sufficient more than 250W anode dissipation for a
air pressure it wont cool your 4CXs. So you period, and they can cope with this if you
need to design your cooling system. blow them hard. Basically, you want to blow
As a first step, look at the specifications of any ceramic triode or tetrode as hard as you
the blowers you might consider using. What can, provided your headphones and micro-
you want to know is how much airflow the phone will keep the noise out and the valves
blower delivers at a given back pressure. dont get blown out of their sockets!
Blower manufacturers supply this information The two major sources of blowers in this
in the form of a graph by way of an example country are Airflow Developments Ltd of High
take a look at Fig. 6.7, which shows the output Wycombe and Air Control Installations Ltd of
characteristic of the Air Control Installations Chard, Somerset. Both will be delighted to
model VBL5/3. Airflow rate in cubic feet per supply you with catalogues and data sheets,
minute (cfm) is plotted against fan static and youll probably discover that the blower
pressure basically the air pressure the fan you need to do the job properly is much larger
can produce. Could we use this blower to keep than the one on Freds linear. For instance, a
a pair of 4CX250Bs cool? Well, first of all, lets common blower found at rallies is the Air
examine the requirements again. The specifi- Control Installations VBL4/3 one down from
cation says that at 250W anode dissipation the 5/3 we discussed a moment ago. You
the valve needs 64cfm through it, and at that might think from looking at it that it was just
flow rate the anode cooler produces 082" of the job, but it cant get anywhere near 1"
back pressure. Thats just the valve, though, static pressure and sure enough, if you try
and a good rule of thumb for the 250 series one of these on a 4CX250 linear youll find
based on experience and a lot of messing that almost no air comes out of the anodes. A
about with pressure gauges is to assume that new VBL5/3 cost about 47 in 1992, and an
the valve, base, chimney and other impedi- Airflow Developments Type 45CTL (also good
menta in the grid compartment (Fig. 6.5) all for about 100cfm at 1" static pressure) was
add up to 1" of back pressure. For a pair of about 40. Isnt it worth that much to know
250Bs that means we need 128cfm at 1". that youve done the cooling properly?
Taking the VBL5/3 graph again, youll see If youre absolutely determined to go to a
that at 1" static pressure the blower is produc- rally and pick up a cheap blower, heres what
6 23
T R A N S M I T T E R S, POWER AMPS & EMC
to look for. First of all, axial Muffin-type fans and up a chimney. Part of the airflow is bled
are no use whatsoever; they are designed only downwards through the valve socket and out
to move air and generate hardly any pressure. via a restriction in the grid compartment, and
Youre looking for a centrifugal snail blower this is where the weakness lies: there is no air
with a wheel of at least 10cm (4") internal blast directly on to the base seals. The RIW
diameter. Expect to see a hefty induction supporters reply that because the airflow paths
motor, and avoid anything rated below up through the anode and down through the
2000rpm. Pressure-generating capability base are in parallel rather than in series, this
depends on the square of the speed of the reduces the back pressure and allows the
blower blades, ie on (diameter x rpm)2, and is blower to deliver more air to cool the valve as
severely affected by a shortfall in either the a whole. Maybe so... its your decision. By
diameter or the rpm. 1992 Eimac had rather confused the issue by
When you see the size of a proper blower for admitting that they saw nothing wrong in
a pair of 4CXs, youll realise why all commer- principle with the K2RIW method of cooling,
cial amateur-band 4CX amplifiers get hot to which I would reply that Ive personally
theyre all grossly undercooled. One manufac- wrecked three 4CX250Rs by using it.
turer of a single-4CX amplifier fits a Muffin- Blowers can fail, and their power leads can
type axial fan, claiming that centrifugal drop off or break. Maybe Ive been unlucky,
blowers are ...too noisy for consumer use. A but in the last eighteen years I have had one
4CX250B in that amplifier seems to have a life blower which seized solid, another in which
of somewhere around 100200 hours, which the fan became uncoupled from the motor
must be something of a record low. Like it or shaft and a third which had a fatigue fracture
not, part of the price of having a 250W anode of an internal wire to the motor and all
dissipation tetrode (or triode) which can work failed when the associated amplifier was in
efficiently at 500MHz is that you need a large use. You might think it would be impossible
blower to keep it cool. Moving several tens of to miss the deafening silence when the blower
cubic feet of air a minute through an inch of stops. But if youre wearing headphones and
back pressure is an inherently noisy process. youre out on a contest, with the wind
Put it this way if the blower isnt noisy, it howling and rain lashing against the side of
probably isnt big enough! the tent, its perfectly possible to miss the fact
About a third of the cooling air requirement that the blower has stopped and the amplifier
of a ceramic tetrode is to cool the grid spigot is quietly doing a China Syndrome. Its easy
and seals on the valves base, leaving the other enough to provide some form of airflow
two-thirds to cool the anode. Certain ampli- switch or sensor which will shut down the
fier designers seem to assume that they can whole amplifier if the air ceases to flow. Try
cut down on the blower requirement by either the local industrial central-heating boiler
blowing air backwards through the valve and spares and repair depot for adjustable low-
thereby avoiding all that clutter at the bot- pressure switches, or make your own by
tom, or alternatively blowing the air the right adding a vane to a low-torque microswitch.
way but ducting through the anode only. One more thing: before you commit
According to Eimacs original data sheet, you yourself to expensive metalwork, spend a
must blow a 4CX only in the base-to-anode pleasant evening building a cardboard mock-
direction, with air blowing on to the base seals up of the proposed airflow system. It doesnt
as well as through the anode. Having said have to look pretty, but it should include all
that, there is plenty of favourable amateur the main features of the airflow path: the
experience with the K2RIW airflow system valves in their bases, the mesh used for RF
(Chapter 10) which blows air into the output screening, and any lengths of air ducting.
compartment, cooling the tuned circuit before Power up the blower and check that the water
most of it passes out through the valve anodes gauge shows sufficient air pressure.
6 24
WHAT KIND OF AMPLIFIER?
6 25
T R A N S M I T T E R S, POWER AMPS & EMC
offer them as an extra-cost option in an need to look at the most peculiar feature of the
amplifier designed (if thats the right word) for 4CX family the dreaded secondary emission.
a 4CX250B and with no other changes to suit This is what makes screen current meters do
the 350. The 4CX350 was designed for linear funny things (ending in an almighty bang if
service in Class AB1, and is capable of ex- youre not careful), and makes the design of
tremely good high-order intermod perform- the screen supply about fifty times more
ance, but only if the power supply is good complicated than most designers appreciate.
enough to bring out that performance. Whats supposed to happen in a tetrode is
These power output figures mean that you that electrons emitted from the cathode are
need to use a pair of valves to achieve 500W controlled by the electrostatic field of grid 1
PEP output at the intermod performance and are whipped away towards the anode by
which weve specified as the minimum the positive voltage on grid 2, the screen grid.
desirable and please note that its a pretty Most then pass through grid 2 to be collected
modest minimum, not some high-grade on the anode proper. The balance between
unattainable pie-in-the-sky figure. If you have electrons intercepted by grid 2 and by the
the choice and the money, go for 4CX350s at anode depends on the design and internal
144MHz. The 4CX350FJ is the best, closely geometry of the valve, and also on the relative
followed by the F or the A. However, do bear voltages of the screen and anode. When the
in mind that in order to get the performance anode is very positive, either because the DC
of which these valves are capable, youll need voltage is high or because were at the positive
to build a very high-grade supply for the peak of the RF cycle, the screen grid intercepts
screen and control grid, and keep the anode very few electrons. On the other hand, if the
voltage up around 25kV with good regula- anode voltage is not much greater than the
tion. If you can manage all this (and it isnt screen voltage, a considerable proportion of
too difficult see Chapter 11) you can achieve the electrons emitted from the cathode will be
better than 80dB for all intermods beyond collected by the screen grid.
seventh-order; if you cant, theres no point in An electron emitted from a cathode at
bothering with 4CX350s. Also remember that 4CX250B-type temperatures and attracted
350s cannot be used in Class C (which is no towards the anode and screen by a positive
loss) and they really dont want to know voltage on both is doing about 50,000 miles a
about the 432MHz band. second by the time it hits the anode. When a
Overall, the best buy for either 144MHz or stream of electrons moving at this speed
432MHz is probably a pair of 4CX250Rs. Next smashes into the anode, it dislodges electrons
best would be a pair of 4CX250BMs if you from the anode material itself; this phenom-
could find them; they were a special version enon is called secondary emission and it results
characterized for low secondary emission and in a negative contribution to the net current.
they sometimes show up new or slightly used. In a tetrode there is also secondary emission
After that its a pair of 4CX250Bs. For 144MHz from the screen and the fate of these second-
only, a pair of 4CX350s or 350FJs will do a ary electrons depends on the instantaneous
superb job if the power supply is up to it. anode and screen-grid potentials.
It is extremely difficult to design and
QUIRKS OF THE 4CX FAMILY manufacture a tetrode which is physically
A tetrode such as the 4CX250B contains four small enough to work at VHF/UHF, has low
electrodes: a heated cathode, a control grid, a inter-electrode capacitances, is rugged and
screen grid and an anode. In all the amplifiers reliable, and yet is capable of extremely high
Im going to describe, the cathode is simply power for its size. The trade-off for these good
connected to DC and RF ground but the other features of the 4CX series is that the overall
three electrodes need some kind of DC voltage screen-grid current can be positive at some
applied. Before we go further into that, we drive levels and negative at others, and is
6 26
WHAT KIND OF AMPLIFIER?
6 27
T R A N S M I T T E R S, POWER AMPS & EMC
COMMISSIONING
A NEW AMPLIFIER
6 28
COMMISSIONING A NEW AMPLIFIER
twist up a couple of inches of wire to achieve power. This means that we need at least that
the same effect. If theyre too long, remove a amount of isolation between the grid and
plate from each rotor of the split-stator the anode sides of the amplifier. Lacking
capacitor. If the anode circuit wont resonate, this, the amplifier can turn into a very
however, youve got a bit of a problem and effective and powerful TPTG (tuned-plate
youll need to do some fairly major surgery tuned-grid) oscillator. Actually, Great Brains
somewhere but not before re-checking all who know all about things like Bode plots and
the dimensions against the drawings! Nyquist criteria will tell you that you need at
Having checked roughly that the tuned least 3dB more isolation than the power-gain
circuits will resonate, connect a VSWR meter in figure, to meet the proper criteria for stability.
series with the amplifiers input. With the All in all, lets say that your amplifier requires
heaters active (dont forget the blower) and a at least 30dB isolation between its input and
little grid bias, apply a whiff of RF drive and output circuits.
spend a happy half-hour adjusting the grid From the constructional point of view its
tuning and loading controls (and if necessary easy to achieve 30dB isolation provided you
the position of the input coupling loop) to get take a little care solid metalwork with lots of
a VSWR something like 1:1. If the amplifier is screws in the right places and plenty of
accurately built from a published design you screening and decoupling. Isolation really
should have no difficulty in finding the right shouldnt be a problem unless you do a
settings. If it doesnt want to know, check your Walter and run the grid leads through the
dimensions and wiring. Its amazingly easy to anode compartment, or run an unbypassed
omit a crucial grounding strap or assume that a HV lead alongside the grid-tuning capacitor.
ground return exists when in fact it doesnt. Unfortunately, no matter how well you did
the metalwork, as soon as you plug the valves
NEUTRALIZING into their sockets the whole thing turns to
Eventually you should end up with a reason- worms. The reason is quite simple its
able VSWR on the input and a grid circuit that quoted in the data sheet as CAG, the capaci-
seems to be tuning up. Whats next? Probably tance between the control grid and anode of
neutralizing, if its a push-pull amplifier. the valve, and it represents a feedback path
between the input and output circuits.
Because a tetrode has a second grid which
For some reason the prospect of neutralizing screens the control grid from the anode, CAG is
frightens amplifier builders to death; as soon pretty small for a 4CX250B its 004pF but
as someone mentions the word, they start when multiplied by something known as the
gibbering and muttering and finding excuses Miller effect, it provides enough feedback to
not to play with the new amplifier this turn your amplifier into an oscillator. Its this
evening. Actually, neutralizing a well-built feedback path that youre trying to neutralize
amplifier is dead easy. You dont need any test so that, as far as the amplifier is concerned, it
gear whatsoever; all you need is a source of RF isnt there any more. For the purists, we ought
and the existing PSU metering and if youve to add that in theory any common screen or
never done it before a bit of patience. cathode inductance needs neutralizing out as
Maybe itll help if we work through what well, but this shouldnt enter into it if youre
were trying to do and why. Think of it like using proper bases and solidly grounding each
this. A 4CX tetrode amplifier at VHF or UHF cathode pin directly to the metal body of the
has a fair bit of gain: a two-valve grid-driven base.
144MHz machine with reasonably high-Q You might think that a feedback capacitance
input and output circuits should have a power of 004pF isnt a lot, however much it might be
gain in the region of 26dB, implying that if we multiplied by Mr Miller. In point of fact, most
put in 1W of drive well get 400W of output single-valve amplifiers using a 4CX250B can
6 29
T R A N S M I T T E R S, POWER AMPS & EMC
6 30
COMMISSIONING A NEW AMPLIFIER
6 31
T R A N S M I T T E R S, POWER AMPS & EMC
no amount of messing about with the anode and no cooling air, so do take a little time to
tuning and loading controls affects the grid work out how to keep the air going through it.
current in the slightest. Having checked the heater voltage, switch
everything off and remove the valves from the
READY TO RUN amplifier. Now you can replace the fuses and
Were getting closer. Put the screen changeo- restore the other supplies to working order,
ver relay back in its socket. Then take out but DONT connect the HV supply up yet. The
some strategic fuses in the power supply so next check is that the screen voltage is
that the only voltage applied to the valves switched to its Class-AB1 value and that the
when you switch it on will be to the heaters. voltage on pin 1 of each valve base is zero on
Double-check that the HV supply is un- receive and the proper screen voltage on
plugged and the HV connector is removed transmit. When youve done that, switch to a
from the amplifier. Then switch on and resistance range and check that pin 1 is solidly
measure the heater voltage, using a good grounded on receive. Then check that the grid
DVM if at all possible. The figure you want to voltage switches nicely between the proper
see depends on a number of factors including transmit and receive values when the PTT is
what frequency the amplifier is to be used for operated. If youre using 4CX250Bs, set the
it might be worth having a quick look at the grid voltage on transmit to approximately
section dealing with heater supplies in 65V. Switch off, replace the valves and put the
Chapter 11 to remind yourself of the require- cover back on the anode compartment with
ments before you power up. In simple terms, all the requisite screws this is essential, both
if its anything over 600V you ought to be for safety and also because if you dont you
pondering the wisdom of knocking it back a can get some very odd effects (not to mention
bit before you go any further. Bear in mind massive feedback) due to distortion in the RF
that when an air-cooled triode or tetrode is in circulating-current pattern. Finally, take away
its socket, its physically impossible to meas- any little bodges you may have put in to fool
ure the heater voltage (or any other voltage, the power-supply interlocking for the tests
come to that) and at the same time keep the youve been doing.
proper amount of cooling air flowing through Before you do anything else, you MUST now
the valve, so dont hang around when making connect and double-check your safety ground
measurements. braid between the amplifier chassis, the main
You should be able to slide the bottom power supply and the HV supply as discussed
cover of the amplifier back over the grid in Chapter 11. Use an ohmmeter to check for
compartment, leaving a gap for the meter lead continuity between the amplifier chassis, the
to come out; seal this with a bit of Blu-Tak or chassis of the PSCU and HV supply, the prime
similar. Another way to achieve the same end mover and ground. When youre happy, make
is to cover the entire grid compartment with a absolutely sure that the HV supply is not
sheet of clingfilm, secured to the sides of the plugged into the mains and then re-connect
amplifier with Sellotape or insulating tape. the HV cable to the amplifier. Connect the
You can then gently push the meter probe station dummy load to the amplifier output
(you only need one since the other is an earth via a power meter set to a low range such as
return and can go anywhere on the metal- 10W. Connect the prime mover to the
work) through the clingfilm and clip it on to amplifier but make sure that (a) its in receive
the heater pin. The film will balloon like and (b) the output power is backed right off to
mad when you fire the blower up, but itll do zero. Set the frequency to somewhere in the
an excellent job of forcing some air to flow middle of the SSB sub-band.
where itll do most good. From bitter experi-
ence, its fatally easy to crack the ceramic at STATIC TESTS
the bottom of a 4CX if you apply heater volts Now then youre shortly going to run the
6 32
COMMISSIONING A NEW AMPLIFIER
amplifier for the first time and see the culmi- scheme outlined in Chapter 11, and the
nation of all your hard work. Double-check anode-current meter will also show some-
that the dummy load is connected and that thing. If youve set the grid bias correctly it
the power meter is set to a low range. Switch should show about 100mA if the amplifier has
on the blower, heater and all supplies but at two valves and about 50mA if it has one.
this stage make sure that the HV cannot come Anywhere in that region is near enough for
on until youre good and ready for it to do so. now. Having established that all these things
Allow ten minutes or so for everything to happen and that there are no untoward events
warm up and settle down then power up the or clouds of noxious black smoke, let the PTT
HV supply and see what happens. Actually, go and switch back to receive. Switch between
nothing whatsoever should happen and the transmit and receive a few times to get used to
only meter needle which ought to move is the how placidly the amplifier behaves.
one on the HV PSU indicating anode volts. If Ive just described what should happen.
there are any pyrotechnics or if fuses blow, What could happen when you switch to
shut down and find out what went wrong transmit is that the grid meters and the
its probably something quite simple like anode-current meter go hard over, the HV
inadequate clearance somewhere (look for transformer emits a deep hum, the shack
carbonising across a gap) or a decoupling lights go dim and the RF output-power meter
capacitor which didnt like 2kV very much. If shows about thirty trillion watts going into
you cant see anything, take the valves out the dummy load. If so, the amplifier has
and carefully inspect both them and the become bored with merely amplifying and has
inside of the bases for any evidence of decided to make a career move into high-
flashover. If you still cant see any reason for power oscillation. No problem; switch every-
the pops and bangs, leave the valves out, thing off, read the section on neutralizing
replace the anode-compartment lid and again and this time do it properly! Actually,
disconnect everything except the HV supply this little nasty usually strikes when youre
and its ground return. Turn the HV on again. doing your third or fourth amplifier and have
If there are now no flashes and bangs, change become over-confident. First-time builders do
the valves; one of them obviously has an their neutralization with care and dedication
internal fault. It might be a good move to so, when they switch to transmit, the new
check that nothing in the PSCU has taken Thunderbox just sits there quietly. Hardened
umbrage, incidentally, before you power up amplifier-bashers dont, so it doesnt.
again make sure you physically unplug the Assuming everythings looking good, go to
HV supply and then check as before that the transmit again and lock the PTT. Wave your
screen and grid supplies are doing their thing. hand over the valves air outlet; the exhaust air
Assuming that youve now got the amplifier should be pleasantly warm and should be at
happily sitting there in the receive condition about the same temperature for both valves.
and with HV on, what next? Check again that You can then adjust the standing current to the
the drive power is still set to zero and then requisite value for low intermods, which for
(great moment, this) switch to transmit! Class AB1 is 100mA for a single valve and
Several things will happen when you do therefore 200mA for a pair. Those who have
this, and its good to have a clear idea what to owned commercial amplifiers might think this
expect. Assorted relays will go over, for a start sounds a bit high compared with the figures
the antenna relays, the screen relay and any quoted in their manuals; you may draw your
other PTT relays you may have and the own conclusions. Switch back to receive and
sound of these all operating at once can really check that the standing current falls to zero.
startle you if the adrenalins flowing a bit. The Go back to transmit and check that the anode
screen-current meters will swing up to a current rises smoothly to the value you just set.
centre-zero position if youre using the Stay in transmit and check that no power is
6 33
T R A N S M I T T E R S, POWER AMPS & EMC
Fig. 6.10. Meter readings for a two-4CX250B amplifier, correctly set-up for standing current
(zero RF output power) and at 100W RF output power. The RF output and anode current
readings are for both valves combined, while the G1 and G2 currents are for each valve
being indicated on the output wattmeter. anode current is somewhat higher than the
There shouldnt be so much as a microwatt if standing-current figure, but dont worry too
the neutralizing was done properly. much about its exact value for the moment.
Youre much more interested in the screen-
RF TESTS current meters, which should be reading
Now for some RF. Incidentally, youll need to negative (Fig. 6.10). Adjust the output loading
note some meter readings shortly, so have a capacitor so that each screen meter shows
writing implement and some paper some- about 2mA; the reading should be fairly
where in the vicinity. First of all, set the equal but dont worry if one meter says 1 and
output loading capacitor to maximum loading the other says 3mA for example. Re-peak the
thats fully unmeshed for the W1SL design grid and anode tuning for maximum RF
with loop-coupled output (Chapter 8), or with output power. Now turn the power down and
the string slack in the K2RIW (Chapter 10). back up to 100W or so a few times; make sure
Switch the output power meter to a slightly that all the meter needles move smoothly and
higher range, eg 50 or 100 watts. that the output power goes up and down
Switch to transmit and lock the PTT. Gently without any funny little jerks and jumps.
begin to feed in some RF drive from the prime If you see any weird meter readings or the
mover while watching the output meter. As needles start twitching about, find out whats
soon as you see any indicated power on the happening before you go any further. Youre
meter, tune the grid and anode tuning unlikely to have a problem with the amplifier
controls to peak it up. Dont adjust the loading itself if youve come this far successfully,
control at this stage. When grid and anode though, and about the only thing which
tuning are both peaked, increase drive to could be causing trouble is RF feedback into
bring the output level up to about 100W and the prime mover (see later). If youre using a
note the readings on the anode and screen proper dummy load, that wont happen; if
current meters. You should find that the youre doing a Walter and trying to run up a
6 34
COMMISSIONING A NEW AMPLIFIER
Fig. 6.11. Meter readings for a two-4CX250B amplifier, correctly set-up for 300W and 500W
RF output power. The RF output and anode current readings are for both valves combined,
while the G1 and G2 currents are for each valve
brand-new amplifier into your antenna, dont control. You will totally wreck and ruin the
expect any sympathy from me. amplifiers intermod performance if you do.
Now change the power meters range to 500 The golden rule is to think of the loading
or 1000 watts. Go to transmit and establish control in a tetrode amplifier only as some-
the 100W power level again; peak the grid and thing to adjust to set the screen current to the
anode tuning for the highest RF output. Then figure you want, and only ever adjust it when
slowly bring the power up to about 300W, youre looking at the screen-current meters. So
keeping a wary eye on the screen-current never tweak the loading while gazing at the RF
meters as you do. If they begin to move in the output-power meter. They say it sends you
positive direction at any point, give the blind.
loading control a tweak in the direction that
takes the screen currents negative again, so FULL POWER
that at the 300W mark theyre both around Good, isnt it? Now go for broke and wind the
5mA. If you have less than about 1800V on drive up to the point at which 500mA is
the anode, or the anode-supply regulation indicated on the anode current meter. You
isnt all that hot, you might not be able to get should see about 450-500W RF output and the
the screen current quite so negative. If thats screen current will have moved in the positive
the case, simply adjust the loading control to direction to +2mA or thereabouts (Fig. 6.11). If
make the screen currents as negative as the screen current is somewhat more positive
possible at the 300W output point as shown than that, try and get it back down by adjust-
in Fig. 6.11. ing the loading. If you cant because the
Still with 300W indicated, have a quick go loading is at maximum and the screen current
at the anode and grid tuning to peak the RF is still much above a few mA you havent got
output; you may be able to get a little more. enough loading range. Make a mental note
Whatever you do, please dont try peaking that youll have to increase the output
the RF output by twiddling the loading coupling by bringing the coupling loop closer
6 35
T R A N S M I T T E R S, POWER AMPS & EMC
to the anode line, or by fitting a bigger flapper milliamps before you hit the legal limit.
if your amplifier uses capacitive loading. If the For some 4CXs in Class AB1 the grid-current
screen current is a lot higher than it should meter(s) may move backwards a small
be, like 10 or 15mA, shut down and do this amount normally only a matter of
before you go any further. The coupling is microamps on speech peaks, just before the
much too light and youre in danger of a onset of normal grid current. This indicates
flashover. that you wont be able to wind the drive up
Now take the input drive down to zero and very much more before departing from Class
back up to 500mA indicated anode current a AB1. Not all valves show reverse grid current,
few times. Again, all meters should move and some only start to do it after a few tens of
smoothly; in particular each screen-current hours use. Reverse grid current is the sum of
meter should go from zero to about 5mA and gas current, leakage current and control-grid
then back towards zero, finally ending up at emission current (bet you wish youd never
about +2 mA at full output. For VHF and UHF asked) and it will generally increase somewhat
use, in which heavy loading is good both for with life. If you see much more reverse grid
valve life and intermod performance, that current than usual, the odds are that the valve
order of screen current is what you want to see is running hot; check the blower and any
in each 4CX250B. If youre using 4CX250Rs, ducts, filters, etc for blockages.
note that the intermod performance of this After a few years hard use you may also
valve is very good if the single-tone screen begin to see signs of positive grid current at
current is around zero but it deteriorates your accustomed full power level. This
rapidly if you use lighter loading and let the suggests that the valve is getting a bit past its
screen current go very positive. Keep the prime, and youll have to be a shade less
screen current of 250Rs around zero at full exuberant with the drive control. Even when
anode current for best results. For the the valve is on its last legs, you still should
4CX350A and FJ you should aim for no more never run a Class-AB1 linear amplifier into
than zero screen current (ie never positive) at positive grid current.
250mA anode current per valve.
Incidentally, all these screen-current figures
are given for 4CXs with 350V on the screen
and something like 2kV on the anode. If the
anode voltage is lower than that, the screen
currents will be more positive. Im also
assuming that the valves are in reasonable
condition. Remember that if you have
clapped-out valves in a new amplifier youre
quite likely to see lots of negative screen
current and so you cant possibly know
whether your loading is correct or not.
By this stage you might be wondering why I
havent mentioned the grid-current meters
shouldnt they be showing something? In a
word, no. If you see any indication at all on
the grid meters at 500W out, youve either got
very low anode voltage or some rather tired
valves. In Class AB1 you can forget all about
the grid meters; they should only indicate
anything when you switch to Class C, and if
the valves are any good itll only be a few
6 36
GOING ON THE AIR
6 37
T R A N S M I T T E R S, POWER AMPS & EMC
In spite of everything Ive said about the The next question was: on SSB, how do you
virtues of under-running the prime mover to keep your PEP output within the licence
obtain a clean drive signal, you may have no limits? Answers involving esoteric mental
options. If the available power output is vastly disciplines or speaking in square waves will
more that your amplifier needs, your best have the inspector reaching for his notebook
approach overall is to run the prime mover at again. Instead, how about a decent speech
one-third to half its rated power and dispose processor? You wouldnt find a professional
of the rest of your excess drive in an attenua- SSB transmitting system that didnt have some
tor connected permanently into the transmit form of peak limiting its regarded as vital
path (see Chapter 12 for design information). for commercial SSB users not to overdrive and
thus cause adjacent-channel interference.
STAYING LEGAL Also, its well known that the human voice
The next issue in day-to-day operation is how has a fairly poor ratio of peak-to-mean energy,
to measure the output power of your station, so some compression of the dynamic range of
and how to set up your routine drive levels as speech is a Good Move to increase the intelli-
opposed to those used for the initial tests. The gibility or talk power of your signal.
terms of the UK licence restrict you to Speech processing of some sort is included
+26dBW PEP at the antenna, which is 400W in almost all modern HF rigs, and increasingly
PEP. When officialdom appears on the in VHF/UHF rigs as well. Although clipping
doorstep one day for a station inspection, the increases talk power, its still basically a form
odds are that theyre going to take one look at of distortion and will generate a wide spec-
your gleaming Master Blaster and ask you to trum of intermodulation products. Any
demonstrate how you measure its peak clipper must be followed by a filter to restore
envelope output power and how you keep it the clipped signal to its original bandwidth,
within the licence limits. but the presence of distortion products within
The only way to measure VHF/UHF peak the bandwidth of a heavily clipped signal will
envelope power thats at all practical in an limit the attainable improvement in intelligi-
amateur station is with a good peak-reading bility. Modern rigs usually use RF (strictly IF)
wattmeter. Sure, the Bird 43 will measure clipping, and pass the peak-limited SSB signal
carrier power to within 5% but it wont through a second IF filter to clean it up. The
measure the peaks of the RF envelope, and virtue of RF clipping is that most of the
neither will any other instrument based intermods fall far away from the wanted
simply on an RF detector and an ordinary signal, permitting a heavier degree of clipping
moving-coil DC voltmeter. If you tell the before the audible distortion becomes too
inspector that you set up a carrier level thats noticeable. In contrast, many of the distortion
legal and then, er, assume that speech peaks products of an AF clipper fall within the
equal carrier level, hell probably reach for his wanted audio bandwidth and sound exceed-
notebook and ask you to repeat that for the ingly unpleasant.
court case. Your amplifier probably gives So the best option for a loud, clean SSB
about 10% more PEP than it does sustained signal is an RF clipper, preferably integrated
carrier power, because the HV doesnt sag on into the rig. Next best is an external RF clipper
speech peaks in the same way that it does between the microphone and the rig. This
with a constant carrier or slow Morse. So build device converts your audio signal to SSB, clips
or buy a PEP-reading wattmeter, or modify it, cleans it up and finally converts it back to
your existing wattmeter using the module in audio. The Datong devices are highly recom-
Chapter 12. Commercial PEP wattmeters have mended, especially the ASP Automatic Speech
been reviewed by a number of authors. Try Processor which includes automatic level
and borrow Angus McKenzies Buyers Guide to control of the voice signal presented to the
Amateur Radio; alas its now out of print. peak limiter and also a tone generator to set
6 38
GOING ON THE AIR
the carrier level you need. Having set the speech clipper, turn down the DRIVE LEVEL
levels using the tone, your PEP on speech until its actually narrower than before
cannot overshoot. Even the simple AF clipper because the clipper is limiting your speech
in Chapter 12 is far, far better than no peak peaks below the level that overdrives the rest
limiting at all. of your transmitter. Note the setting of the
DRIVE LEVEL control and dont touch it
SETTING UP A again.
SPEECH PROCESSOR
All forms of speech processor need to be set up 8 Any improvement in talk power will be
with great care. Otherwise you see no im- most noticeable when your signal is weak, so
provement in talk power, or more likely the check the effectiveness of the clipper with DX
processor ensures that you overdrive more stations, not the locals. Adjust the CLIPPING
consistently than before! Here is a general LEVEL as necessary.
procedure for setting up a speech processor.
The locals should appreciate your reduced
1 Connect the rig to a dummy load. Please bandwidth, but dont expect them to give
dont connect it to an antenna because youre glowing reports on your audio quality. Your
going to make some foul noises before you get signal will have more background noise than
everything behaving properly. before and wont sound as smooth, so you
may need to compromise on the CLIPPING
2 Identify a variable gain control ahead of LEVEL. This applies particularly to contest
the clipper input. Regardless of what the stations: although clipping is good for raising
manufacturer called it, were going to call this the DX, you need the points from the locals
the CLIPPING LEVEL control. too, and they wont bother to call you if your
audio sounds unpleasant. As a matter of fact,
3 Identify another gain control after the some people say they cant abide speech
clipper output. Well call this the DRIVE clippers. Maybe thats because theyve never
LEVEL control for the rest of the rig. heard one thats been set up properly or
more likely they have, but havent realised it
4 Turn CLIPPING LEVEL to maximum and because a properly-processed signal just
emit a Waaaaahlo into the microphone. sounds loud.
Adjust DRIVE LEVEL until the RF output If Ive managed to persuade your conscience
reaches the maximum achievable. that some form of PEP limiting is necessary
but you still wont use a speech clipper, your
5 Then back off the DRIVE LEVEL until the only other option is to use ALC. This involves
RF output drops by 10% (see panel on page 6- generating a negative-going signal from an RF
7). This step is the one which gives you a detector at the output of the linear amplifier
clean signal, and its vital. and feeding it right back into the prime
mover, thus enclosing the entire transmitting
6 Listen to your signal on another receiver system within the level-control loop. This is
use headphones to avoid feedback. Advance emphatically not simple. The ALC signal must
the CLIPPING LEVEL control from zero until have the correct characteristics of voltage
your voice sounds loud and crisp but not swing, attack time and decay time to suit not
distorted. Do not touch the DRIVE LEVEL. If only your particular model of transceiver but
the background noise in your shack is very also the power gain within the extended ALC
noticeable, back off the CLIPPING LEVEL. loop. The transverters in Chapters 810 have
internal ALC loops for protection against
7 Check the width of the signal with a local overdrive, and may provide some initial
station. If its broader than it was without the circuit ideas. But it isnt possible to be specific
6 39
T R A N S M I T T E R S, POWER AMPS & EMC
about the ALC requirements of your particular probably wont like it either.
setup; youd have to experiment. Now have I Flashovers are not something you want
persuaded you to try a speech clipper? every day of the week, but theyre not the
catastrophes some people seem to think. As
FLASH-BANG-WALLOP! long as you take a few precautions, like having
So at last youre on the air with the atom- some suitable resistance in the output of the
smasher and it all seems to be working well. HV supply to limit the prospective short-
When you transferred the output from the circuit current and using voltage-dependent
dummy load to the antenna, the amplifier resistors (VDRs) to limit the voltage surge on
tuned and loaded a little differently, but thats the screen supply rail (Chapter 11), a flashover
only to be expected because antennas dont wont necessarily harm anything although
look exactly like 50 resistive. Still, theres itll usually wake you up.
plenty of range in the loading control to The other celebrated cause of flashovers is
match anything reasonable. This, of course, is letting the screen or control grid float relative
one of the beauties of a valve amplifier as to the cathode. The type of flashover doesnt
opposed to solid-state amplifiers which have involve RF at all, but the result is equally
no user-settable loading at all. But never try to unpleasant. A floating screen grid takes about
run a big valve amplifier without a load of two-fifths of a nanosecond to assume the full
some sort. If you do, there will be the father- anode voltage, so about 2kV or whatever you
and-mother of a flashover as the peak voltage happen to have on the anode appears across
in the tank circuit winds itself up to a hun- the 1kV-working screen-bypass capacitor in
dred kilovolts or more. the valve base. This type of flashover isnt
Where does a hundred kV come from, when even big and entertaining; its usually a sharp
the HV is only 2kV? The basic reason is the ffft! followed by a small sizzle, and you need a
difference between the loaded Q and un- new base. To prevent this happening to you,
loaded Q of the RF tank circuit. In normal first of all dont adopt a suggestion in the
operation the peak RF voltage in this circuit German UHF Compendium to wit, sticking a
will be the DC anode voltage multiplied by 100k resistor between the screen of a 4CX
the Q-factor, which is fairly low because of the and deck in the fond hope that itll ...prevent
loading effect of the antenna so typically static damage. You might as well hang some
youd expect peak RF voltages of 46kV garlic in the grid compartment itd probably
depending on the specific design. However, be more effective. A better defence against the
the unloaded Q of the tank circuit is much floating-grid flashover syndrome involves
higher than its loaded Q, by at least a factor of mounting a suitably rated voltage-dependent
10 and in many designs much more. So if you resistor (VDR) directly on the valve base in
disconnect or lose the load from your addition to about 33k of bleeder resistors,
Thunderbox and go Waaaaahlo, the RF with back-up VDRs in the PSU. Even in the
voltages developed in the tank circuit will be best-run households, wires to the screen can
ferocious at least 4060kV, and probably a simply drop off; if a new base costs nearly
lot more. Something has to give, and what 100 and a VDR costs 70p, how can anyone
usually breaks down is the external path justify not fitting one or more? Try the
across the ceramic anode/screen insulator of V250LA40 for a 300V screen supply, or a
the valve. The flashover from the anode has to V275LA40 for 350V.
go somewhere, and its usually straight down
to the metal screen-grid ring. This is why you RF FEEDBACK
need to take precautions, since the 1kV- If youve got RF feedback, the usual symptoms
working capacitor in the valve base wont are that people complain about squawks and
handle a huge overvoltage, and the compo- serious distortion on your speech peaks or
nents associated with the screen supply when you stop talking, the transmitter
6 40
GOING ON THE AIR
6 41
T R A N S M I T T E R S, POWER AMPS & EMC
EMC
6 42
EMC
about you is defective; you are an inadequate rely on the other person because frequently
and pathetic human being. Hence the youre not dealing with the real rational
irrational reactions, especially since weve all human being at all, but with a rigid and
been pretty systematically misled to suspect unconscious set of behaviour patterns.
that the accusation might be true. Have Grasping that simple fact and working with it
another look at the responses above and youll is the first step in doing something about
notice that most of them can be translated as EMC problems.
Even if what you say about my transmission How do the people factors interact with
is true, I dont know what to do about it. the rest of the domestic EMC situation? Well,
More reinforcement of those lurking feelings you certainly dont need me to tell you that in
of inadequacy. In other words, what they 99% of cases its the neighbours equipment
really need is a bit of help starting right here thats defective it doesnt possess adequate
in this chapter. immunity to RF signals. But if you tell your
Its interesting to note that some of the neighbours straight out that their beloved hi-
typical responses are also extremely defensive fis, TVs, videos or telephones are defective,
(Well, youre 50 kHz wide yourself, old theyll all arrive in a deputation headed by
man). Maybe our transmitters or guns, cars, Matey from three doors away! On some
steam locomotives, football teams or whatever primitive level we all like to think that the
serve as symbols for an unconscious part of things we choose to buy and own are perfect,
ourselves, and when that symbol is criticized and discovering that they are not is painful.
its as though we ourselves are under attack, Its the same pain that we felt as small chil-
and we automatically start to fight back. dren when we discovered that our parents
Worse still, an unreasonably defensive werent perfect either, as a matter of fact.
reaction can send the whole situation to pot The staff of the Radio Investigation Service
by provoking the Walter in the person are no longer able to play as effective a
myself, for example who was originally just mediating role as they could when they
trying to be rational and helpful but just worked for the old Post Office. That isnt
CANT STAND UNREASONABLE PEOPLE!!! necessarily to their personal liking, so dont
(Sorry. Deep breath. Carry on...) blame them for it. It just means that youll
By extension theres a Walter in each of have to do the mediating yourself, as well as
your neighbours. If you start breaking through investigating and solving the technical
into someones hi-fi or television, youre problems. Its quite hard to convince people
invading their privacy which is a civilized that because the problem only starts when
term for that very primitive, un-evolved and youre transmitting, that doesnt make it the
unconscious part of ourselves which is fault of the transmitter to the non-technical
concerned with safety and security. If you that takes quite a leap of logic. Try explana-
threaten that, dont for one moment expect a tion by analogy if the builder leaves a slate
rational adult response. Youll get a response off your roof youll get water in, but only
which is probably thirty or forty years out of when it rains. The solution to the problem is
date; itll be exactly the same response as the not to complain about the rain, but to get the
individual made when they were very small builder back to fix the slate.
and subjected to a similar invasive stress. This In the UK, direct legal responsibility to do
isnt a psychological textbook and Im not something about deficiencies in your neigh-
Melanie Klein. Just dont be surprised when bours equipment rests with the supplier of
rational reasoning doesnt work and is met by the goods, especially if theyre new and always
anger, rage and other manifestations of if theyre rented. Unfortunately, for much the
hostility; and dont get drawn into it. same irrational reasons as described above
Whatever happens, its up to you to remain plus ignorance and sheer laziness in some
rational, calm and positive; you cant always cases many dealers are quite prepared to lie
6 43
T R A N S M I T T E R S, POWER AMPS & EMC
6 44
CHAPTER 7
BEAM ANTENNAS
& FEEDLINES
7 1
BEAM ANTENNAS & FEEDLINES
BEAM ANTENNAS
AND FEEDLINES
7 2
BEAM ANTENNAS & FEEDLINES
gain over a half-wave dipole, or dBi for gain plane. The familiar butterfly patterns (eg Fig.
referenced to the isotropic radiator. Even in 7.1) are generated on a test range by plotting
scientific publications, one is often forced to relative power on a circular chart which is
search the text for a clue to the reference used. rotated synchronously with the antenna. The
In professional RF engineering there is a advantage of this circular or polar diagram is
convention that antenna gains below 1GHz that its very easy to visualize what happens
are given in dBd, and above 1GHz in dBi, so when you turn your antenna, because thats
in this book well generally be working in dBd exactly how the measurement was made. An
unless theres a good reason to use dBi. Either alternative method is to record the power
way, well always say which were using. level on a linear strip-chart as the antenna is
rotated; although this shows up the low-level
PATTERN details more clearly, it makes the real-world
The spatial radiation pattern of an antenna behaviour of the antenna harder to visualize.
describes the distribution of power passing To compare antenna patterns, they have to
through the surface of an imaginary sphere be drawn to the same scale. Scientific and
surrounding it. This idea is difficult both to serious amateur publications almost always
picture and to measure, so we normally plot use charts with equally-spaced rings in dB
the power in the horizontal and the vertical steps, the outermost ring being the 0dB or
3dB
beamwidth
First minimum First sidelobe
Main lobe
2nd sidelobe
E-plane null
Rear lobe
Fig. 7.1. Polar pattern of a long Yagi antenna showing the main features (E-plane)
7 3
BEAM ANTENNAS & FEEDLINES
7 4
BEAM ANTENNAS & FEEDLINES
The pattern of a horizontal dipole, meas- with the Kraus formula is to work it back-
ured in the horizontal plane, is called the E- wards, to check what beamwidths you can
plane pattern because it contains the electrical reasonably expect from a high-gain array; Ill
vector of the electromagnetic field. The say more about this later.
pattern perpendicular to the dipole element(s)
is called the H-plane pattern because H is the FEED IMPEDANCE
symbol for the magnetic vector. Except for As seen from the feedpoint, an antenna
very complicated antenna arrays, just these presents a complex impedance with resistive
two plots will tell us all we need to know and reactive components. The resistive compo-
about the radiation behaviour. From the E- nent is composed of the radiation resistance,
and H-plane plots the overall spatial distribu- which consumes power by radiating it,
tion of radiated energy can be estimated, a together with the loss resistance which really
procedure called pattern integration. Measuring does consume power by heating up the
the radiation patterns can therefore be an antenna. The reactive part is caused by the
approximate method of measuring the gain. antenna being made up of elements which
Strictly speaking, pattern integration doesnt behave like tuned circuits. Both the resistive
tell you the gain it tells you the directivity, and reactive components of the antenna
ignoring any non-radiative power losses. If feedpoint impedance will vary with frequency,
you measure the E-plane and H-plane the reactive component usually more rapidly.
beamwidths E and H, you can estimate the An antenna should of course be impedance-
directivity using a very simple formula derived matched to the source of RF power usually
by John Kraus W8JK: the feedline. If it doesnt already match the
feedline, you can obviously use some imped-
42000
DI = 10 log10 dBi ance-transforming device. But remember that
( f E f H )
transformers are not lossless, and loss always
As I mentioned earlier, a dipole has some rises with the transformation ratio; so an
directivity and hence some gain over the antenna design which wants to match the
imaginary isotropic radiator; 215 dB in fact. feedline is always desirable. The best gain and
So we can amend the Kraus formula to give a pattern do not always come at the frequency of
result in dBd, thus: a natural impedance match, so its always
good to know what is happening to the feed
42000
DD = 10 log10 - 2 15dBd impedance as the frequency is swept across the
( f E f H )
amateur band and possibly beyond. Some
Having said all that, you must be very preamps and transmitter final amplifiers have
careful with the Kraus formula it involves a high gain outside the amateur bands, and may
lot of assumptions which may not be correct. break into parasitic oscillation at frequencies
Firstly, if youre going to equate directivity where they see a particularly bad mismatch.
with gain, you must obviously assume zero
losses. This simplified formula also assumes a BANDWIDTH
rather narrow front lobe ( less than 40), and The operating frequency range of an antenna is
negligible back and side lobes (at least 20dB). determined by two factors: gain bandwidth
If there is significant energy loss to these and impedance-matching (VSWR) bandwidth.
minor lobes, the directivity and gain are These two factors are quite independent of
reduced accordingly which may mean each other, but they both limit the frequency
severely! Although it is possible to refine the range where operation is possible without
formula to take some account of sidelobe retuning of some sort. For beam antennas, the
levels, you are just piling one assumption on most practical measure of gain bandwidth is
another, and your final estimate of gain wont probably the frequency span between the 1dB
be worth much. What you can usefully do points. For Yagi antennas the gain drops much
7 5
BEAM ANTENNAS & FEEDLINES
more rapidly on the HF side of the peak than of the physical element cross-sections see
on the LF side, so it isnt always wise to design the panel about Antenna Areas.
for maximum gain in the middle of the There is a complication with scatter modes
operating band. such as aurora and troposcatter, because the
VSWR bandwidth is generally defined as the incoming wavefront is not coherent (Chapter
frequency range within which the VSWR is less 2). This prevents the antenna from developing
than 2. Different types of antennas behave very its full gain on receive, so the effective capture
differently in this respect, depending largely on area is reduced. The shortfall in gain is called
the Q of their matching networks. the aperture to medium coupling loss.
To sum up: capture area is not directly
CAPTURE AREA linked with physical antenna dimensions.
As I mentioned earlier, antenna gain is Although it is sometimes useful to think of
actually a misnomer. In the receiving case, an the capture area as an actual surface with
antennas ability to gather energy out of the definite boundaries, dont take this too
propagation medium is much better described literally its real shape is not always obvious,
by the term capture area. At a given field and the edges are fuzzy. So capture area is
strength, the total power collected by the really just an alternative to talking about gain.
receiving antenna corresponds to a surface
area which can be measured in square wave- ANTENNA NOISE
lengths. This is most obvious for a horn Chapter 4 has a plenty to say about antenna
antenna, which really does capture almost all
the radiation entering its open end. It is also
easy to see how a large parabolic dish captures
the energy striking the surface and focuses it ANTENNA AREAS
onto a smaller antenna at the feedpoint.
There are several different types of area associated with
Wire and rod antennas have next to no antennas, so lets be clear what we mean in this chapter.
surface area, yet they still capture RF energy. It
doesnt really matter how they do it, but the Capture area is a measure of the antennas ability to
amount of energy corresponds to some collect incoming energy from free space. It is not a true
geometrical quantity which you could measure with a
equivalent capture area which often has no ruler; in fact its often very different from any area you could
obvious physical extent. We can discover the see when looking at the antenna.
capture area of a wire antenna from its gain.
Aperture area is a term I would only apply to antennas
Capture area is related to gain by a simple which possess a genuine, solid, physical radiation-collect-
formula: ing aperture, ie dishes and horns. It is the geometrical
frontal area you could see when the antenna is beaming
GI straight at you, and could measure with a ruler. Capture
A=
4p area is always smaller than aperture area; the ratio of the
two is the aperture efficiency.
where A is the capture area (in square wave-
lengths) and GI is the gain over isotropic (as a Frontal area is the term Id use for other antennas apart
from single Yagis, eg broadside arrays, or arrays of Yagis. Its
ratio, not in dB). Thus a half-wave dipole with
the area enclosed by lines drawn around the outermost
its gain of 215dB over isotropic has an elements of the array, as viewed from the front. Capture
effective capture area of 013 square wave- area is usually somewhat greater than frontal area.
lengths. A Yagi antenna with a gain of 13dBi
Physical area is what you (or the wind) would see when
(a factor of 20) has an capture area of about the antenna is beamed straight at you. For an aperture
16 square wavelengths which could be antenna, physical area and aperture area are generally
visualized as a circle of about 14 in diameter. almost identical. For antennas like Yagis and colinear
Although you cant see this shape, or guess its stacks, physical area is generally much smaller than either
frontal area or capture area.
size without a little calculation, the capture
area is obviously much greater than the sum
7 6
BEAM ANTENNAS & FEEDLINES
7 7
BEAM ANTENNAS & FEEDLINES
materials. Ill return to computer optimization additional feedlines, so youd be doing well to
later in this chapter. achieve 25dB net increase. Both theory and
Before we leave the general topic of effi- experiment show that DOPT remains the same
ciency, Ill just mention a different kind of for more than two antennas stacked in a row.
efficiency which relates to horns and para- Usually the beamwidths E and H in the E-
bolic dishes. This is the aperture efficiency: the and H-planes are different, so the optimum
ratio of the effective capture area to the stacking distance must also be different in the
physical aperture area (see box on Areas). two planes. A typical short Yagi antenna
Aperture efficiency is always less than 1 not (about five elements) might have beamwidths
because of resistive losses, but because of such of 50 in the E-plane and perhaps 70 in the
things as incorrect shape of a reflector, poor H-plane. For a box configuration of four
focusing, loss of energy by scattering and antennas with horizontal polarization, this
spillover or failure to capture all the energy would mean an optimum separation of 118
because the holes in a mesh dish are too large. horizontally and 087 in the vertical plane
A warped and battered dish with a poor choice (centre-to-centre). If the four antennas were
of feed antenna will have low aperture simply stacked vertically up the mast, a total
efficiency and if its covered in rusty mesh it height of 3 times 087 would be required.
will have high resistive losses too. These values are theoretical and may need
some correction, especially if there is evidence
STACKING OF ANTENNAS of severe interaction between antennas as
It is common practice to combine two or more indicated perhaps by a bad impedance
directive antennas into groups, for additional mismatch.
gain. This immediately raises the question of Slightly closer stacking is often recom-
optimum spacing. mended as a precaution against excessive
What is optimum, anyway? It is obvious sidelobes in the final array. Sidelobes arising
that two receiving antennas cannot pick up from stacking rather than from the individual
more than twice the power of one, so the limit antennas are sometimes called stacking
of stacking gain is 3dB. To achieve this, the sidelobes or grating sidelobes. Two antennas
antennas must be separated far enough that separated by DOPT will have a first stacking
their capture areas do not overlap, ie they dont sidelobe at 11 to 12dB in the stacking plane,
compete for the available RF energy. Greater and this can be used as a test for correct
separation would not increase the gain but spacing.
would give rise to increased sidelobe levels and Reducing D to 80% of DOPT costs about
unnecessary mechanical problems. So the 05dB in gain but reduces these sidelobes to
optimum separation is the minimum spacing about 20dB; this is valid for stacks of two
which achieves the full stacking gain. Experi- antennas only! If four antennas are stacked
ments and calculations have shown that the side by side or one above another, only the
optimum spacing (DOPT) is a function of the third stacking sidelobe is affected by varia-
half-power beamwidth () in the stacking tions of spacing; and closer stacking reduces
plane. the gain without lowering the overall sidelobe
l level significantly.
DOPT =
f The above rules apply to any kind of
2 sin
2 antenna, so groups of groups (eg 4 x 4) may be
where DOPT is the spacing (in wavelengths) at designed by using the same basic formula. All
which the stacking gain is maximum and is you need to know are the half-power
in degrees. In practice the maximum attain- beamwidths in both planes. Finally, the most
able stacking gain for two antennas is about important generalization of all about stacking
29dB. From this maximum you must subtract is this: antennas which have clean individual
the losses of the power divider and the patterns will also behave the best in stacks.
7 8
TYPES OF
DIRECTIVE ANTENNAS
7 9
BEAM ANTENNAS & FEEDLINES
7 10
TYPES OF DIRECTIVE ANTENNAS
below, broadside arrays are best regarded as a ments. The LP principle calls for a constant
relatively easy way to reach moderate amount ratio of length and spacing of successive
of gain but little more than that. However, elements. The operating frequency range is
where the objective is a broad, flat beam (eg determined by the longest and shortest
for beacon transmitters) they may still be the elements, and the gain by the taper rate. Only
best choice. those elements within about 10% of half-
The situation for 432MHz is essentially the wave resonance draw sufficient current from
same. The mechanical problems of a colinear the feedline to be involved in the radiating
broadside array are not quite as great because process. So an LP antenna designed to work
of the smaller dimensions, but more gain is over a wide frequency range is actually a
required for effective DX operation than on succession of several limited-band antennas
144MHz so youre quickly into astronomical on one boom. Usually only three or four
numbers of elements and feedpoints. Once elements are operative at the same time. If the
again, the Yagi is a clear winner. taper is very gradual, many more elements can
become operative and considerable gain can
ENDFIRE ARRAYS be developed, but only over a narrow fre-
Any departure from equal-phase excitation of quency range. Although theory predicts a very
a row of parallel dipoles will make the pattern high gain per unit length for antennas of this
squint. And if the phase difference between type, they are hardly ever used in practice a
successive elements in a long row becomes Yagi is simpler.
large enough, both main lobes will shift from The main use of the log-periodic antenna in
broadside to endfire, along the direction of VHF/UHF DX-chasing is to listen for propaga-
the row. Unlike the broadside array, the tion outside the amateur bands. One manu-
pattern of an endfire array is basically unidi- facturer (KLM Electronics) has taken advan-
rectional, because both broadside beams are tage of its broad VSWR bandwidth by using a
shifted to the same side by the de-phasing. small log-periodic cell instead of a dipole
The proper phase relationship between the driven element in wideband Yagis.
elements can be generated in two ways: by
driving each element via a feedline of appro- APERTURE ANTENNAS
priate length or by radiation coupling. In the Instead of building up large radiating surfaces
latter case the non-driven elements are called from a multitude of elements, solid reflectors
parasitic since they feed off the radiated can be used to shape a beam and achieve gain.
energy, and re-radiate it with appropriate If the reflector dimensions are large with
phase and amplitude to create the desired respect to wavelength, the laws of geometrical
pattern. This is the Yagi principle and will be optics will apply and parabolic or other shapes
treated in a separate section. of reflector can be used to concentrate the
Practically the only application of an all- radiation from a dipole or some other small
driven endfire array is the logarithmic-periodic feed antenna. Nearly 100% aperture efficiency
(LP) antenna. Fig. 7.4 shows two of the several can be reached if the radiation comes from
possible ways of providing the required phase within the antenna, as in the case of a horn.
shift (close to 180) between adjacent ele- One heroic radio amateur has actually built a
huge steerable horn for 432MHz and above,
but they are pretty impractical for the rest of
us.
To function properly, the diameter of a
parabolic dish must be at least five or six times
the wavelength and is better at ten times so
a 6m dish for 432MHz is electrically quite
Fig. 7.4. Feed arrangements for log-periodics small! Even for a dish of this size, the
7 11
BEAM ANTENNAS & FEEDLINES
aperture efficiency will probably not exceed circular polarization. The sense of the polari-
50%. The problem is to illuminate the dish zation (left-hand or right-hand circular)
evenly from the primary feed antenna at the follows the thread of the spiral. Gain is lower
focus. For maximum gain the primary feed than with Yagis of the same length, although
has to radiate a sharp cone of RF into the dish. their simplicity makes helicals very useful for
This is not feasible at lower frequencies since satellite work. However, several studies have
the dimensions of the feed itself must be small shown the inferiority of circular compared
to avoid blocking the beam. So a compromise with horizontal polarization in tropospheric
must be made between spillover (radiation propagation. The 3dB loss associated with
past the reflectors rim) or poor illumination contacts to ordinary plane-polarized antennas
of the outer regions. A dish with spillover on further reduces the merit of helical antennas
transmit will also pick up noise from the in all-round use.
ground on receive, especially when elevated, Backfire antennas gained a lot of
so small parabolic antennas are not very quiet; publicity in the seventies but never achieved
50K is about the achievable minimum an- widespread use apart from a few military
tenna noise temperature. applications. There are actually two types:
Despite these problems not to mention short-backfire (SBF) antennas and long-
the weight and windload a number of backfire (LBF) antennas. The SBF type consists
amateurs are using parabolic dishes, mostly of a 2 diameter low-rimmed open cavity, a
for EME work on 432MHz and above. In this dipole in front of it, and a 1 reflector plate
high-gain bracket there are several reasons for behind the dipole. This arrangement exhibits
choosing a reflector antenna. Firstly, it can be about 13dBd gain over a relatively wide
used on more than one band with a simple frequency range. The LBF type has some
change of feed, or even a multi-band feed. features of a Yagi antenna. A rod waveguide
And however high you go in size and gain structure extends from a large reflector plate
there is still only one feedpoint, allowing a or cavity towards a smaller auxiliary reflector.
simple connection to the low-noise preamp With growing length of the waveguide
without lossy line and transformer sections. section, the main reflector must become
Also, the polarization can be switched, or even larger. Functionally it is analogous to a Fabry-
continuously rotated, very simply at the feed. Prot resonator (better known today as a laser
The 432MHz band is about the lower cavity) in which a narrow beam is formed by
frequency limit for amateur-sized dishes, multiple reflections. The Yagi rod structure is
though larger professional dishes around the used to hold the energy together between the
world do appear occasionally on the moon- reflectors. The smaller disk corresponds to the
bounce scene including sometimes the 300m semi-permeable laser mirror which allows
monster at Arecibo, Puerto Rico, which I have radiation to leak out. Unfortunately the main
to admit is big enough even for 144MHz! reflector must be made rather large for high
gain, and for best results must approach a
SPECIAL ANTENNA TYPES parabolic profile in quarter-wave steps. LBF
There are a few more types of antennas antennas with large reflectors have the same
occasionally used in VHF DX work which I gain as equally large parabolic antennas, so
should at least mention. there seems to be an advantage only in the
Helical antennas consist of a corkscrew- transition region below about 5 diameter
like spiral conductor extending perpendicu- where true reflector antennas cease to operate
larly from a reflector plate. The pitch of the effectively. Little successful amateur use has
helix is around /4 and the circumference been reported to date.
near to 1, and the gain increases with length Rhombic antennas with high gain are
as youd expect. A minimum of about five often used in short-wave point-to-point work.
turns are necessary for reasonably round So why not use them on VHF? Well, remem-
7 12
TYPES OF DIRECTIVE ANTENNAS
ber that ground reflection plays an important be made to convey energy from a source to a
part in the build-up of gain from a rhombic, terminated load with low loss and virtually no
so rhombics always squint upwards. A single radiation. These properties are determined by
rhombic in free space would have a vertically the lengths and spacings of the rod elements.
split main lobe with a null at 0. Stacking When used as an antenna, the rod waveguide
rhombics may lower the wave angle but wont structure itself may radiate very little energy;
bring it down to zero, and theres always a almost all the radiation comes from the
problem with ground-noise pickup. A com- impedance discontinuities at the two untermi-
plete redesign and a stack of four would be nated ends. And thats basically how short
necessary to get the same gain from a ground- Yagis work their patterns are formed essen-
less rhombic array high enough for grazing- tially by two sources of energy at opposite
angle radiation. One entertaining feature of a ends of the boom.
very long rhombic is that you dont need a If the Yagi structure is long enough several
terminating resistor to make it unidirectional; wavelengths the impedance discontinuity
by the time the RF arrives at the far end, most can be distributed to make the whole structure
of it has been radiated and the rest has leak radiation, rather like a tubular waveguide
forgotten the way back! antenna with multiple slots in the wall. The
Only one successful VHF DX application of lengths and spacings of the rods can be altered
rhombics is known, which is 50MHz and to cause local changes of phase velocity and
144MHz EME. VK5MC and a few other big- characteristic impedance, resulting in radia-
time antenna farmers use extremely long tion. Tapering towards the open end makes the
stacked rhombics on 144MHz for the few structure increasingly leaky, and at the same
minutes each month when the moon passes time raises the characteristic impedance,
through the narrow fixed beam. In fact the thereby improving the match to free space;
first 144MHz EME QSO from the UK between only a very small part of the energy is reflected
strictly amateur stations was accomplished at the last element. So, unlike the electrically
with a rhombic 100 long (yes, 200 metres short HF versions, the VHF/UHF long-Yagi is a
600ft!), erected in a farmers field of course. travelling-wave antenna.
Beam formation by a distributed radiation
YAGI ANTENNAS source is immensely complicated, but fortu-
Although the Yagi is really just a special type nately the Yagi itself isnt aware of this fact.
of endfire array, its importance has become so Its really very easy to build a Yagi that works
immense that Yagis deserve a special section after a fashion. But it has taken decades of
of their own. The antenna itself was more or research to explore the full potential of long
less a by-product of VHF propagation studies Yagi antennas. All of this work was done by
and its interesting to note that the first experiment, with great time and effort; and
description in a 1928 IRE paper by Yagi and much of it was done by radio amateurs, Im
Uda referred to a long Yagi array, very similar proud to say.
to what we use today. In that paper, Yagi and Until relatively recently, the mathematical
Uda described their brainchild as a disrupted- treatment of realistic Yagi arrays lagged far
rod waveguide structure and that viewpoint behind experiment. After all, a large parasitic
explains some features not easily described by (ie radiation-coupled) array is quite a problem
the usual element-by-element method. because each element influences all the others,
A row of rods, electrically shorter than and all the others affect it too. The accuracy of
wavelength and also spaced closer than results depends largely on the mathematical
wavelength, will conduct RF energy just like a representation of the conducting elements,
tubular waveguide. Like any other waveguide, and also on the numerical methods chosen to
it has a characteristic impedance and a phase solve the complex integral equations. So the
velocity, and a passband within which it can early theoretical work was limited to small,
7 13
BEAM ANTENNAS & FEEDLINES
GROUND REFLECTIONS
Ground reflections can play just as important a role higher angles. At the best angles, you can enjoy an
at VHF and UHF as they do at HF. Although our increase in gain of 6dB from ground reflections; at
antennas are at least a few wavelengths above worst, you get severe attenuation. The effect of
ground, that does not automatically guarantee imperfect ground is to reduce the maximum in-
low-angle radiation. crease and to fill in the nulls.
For any horizontally-polarized antenna above These maxima and minima explain why an an-
ground, there is a null in the radiation pattern at tenna without elevation capability can produce
zero wave-angle (horizontal takeoff), and the main some puzzling sun-noise measurements, and why
lobe is tilted upwards. This is equally true of single you cannot always rely on ground gain around
antennas or arrays, and still applies even if you tilt moonrise and moonset. They also show that a
the antennas themselves. small, low-mounted antenna can sometimes be
The bottom curve in the graph below shows the better for high-angle short-skip propagation than
wave angle of the main lobe produced by an the monster at the top of the tower!
antenna and its reflection in a flat, perfectly-reflect- On 50 and 70MHz you can lower the wave angle
ing groundplane. Even a height of 10, the main by mounting the antenna on a long slope, or at the
lobe is tilted upwards by a degree or so. But top of a suitable hill [4, 22]. The antenna height h
antennas for 50MHz and 70MHz are usually much required to produce zero wave angle on a downslope
lower in terms of wavelengths, so the upward tilt of angle to the horizontal is given by:
can be quite marked.
Above the main lobe there is a pattern of alternat-
l
h=
ing minima and maxima in the vertical radiation (2 sin 2F )
pattern. The wave angles for the 1st minimum, 2nd
maximum, 2nd minimum and 3rd maximum are Note that the slope must be many wavelengths
also shown on the graph, and there are still more at long in order to produce the desired effect [4].
10 3rd max
.
2nd min
.
2nd max
.
1st min
.
1st max
.
1
1 2 3 4 5 6 7 8 9 10
HEIGHT
simplified arrays and a few special cases of It takes just as much know-how and
larger arrays, while amateurs and professional common sense to get sensible results out of a
engineers forged ahead developing practical mathematical model of a Yagi as it does to
long Yagis. Although modern computers and develop and measure the real thing. Unless
numerical modelling methods have now you thoroughly understand what Yagi anten-
caught up with experiments, very little has nas are about, either approach leaves plenty of
changed as a result because the experimenters room for mistakes. The first breakthrough of
had already almost arrived at fully-optimized computer modelling into amateur Yagi design
designs. was the publication of a classic series of articles
7 14
TYPES OF DIRECTIVE ANTENNAS
by the late Jim Lawson (W2PV) in 1980 issues will depend in phase and amplitude on the
of Ham Radio, later published as a book [1]. relative tuning of these elements. The element
Unfortunately Lawson was followed by others is an electrical half-wavelength long when it
who did not share his depth of understanding resonates, which creates the maximum
coupled with practical experience, and a coupling with the electromagnetic field
number of perfectly good VHF antennas were surrounding it. Longer elements behave
sawn up and sacrificed as a result of blind inductively the phases of the currents lag
computer-worship. behind the exciting field. Elements shorter
The next step forward was the availability of than the resonant length show capacitive
the MININEC program, a simplified version of behaviour the currents lead the field. Tuning
the well-established Numerical Electro- also depends on element diameter; thicker
magnetics Code (NEC) for mainframe com- elements resonate at lower frequencies than
puters. The arrival of MININEC on the thinner ones. Within reasonable limits,
amateur radio scene coincided with rapid elements exhibiting the same reactance are
growth in the power of personal computers, interchangeable.
so MININEC was soon running on a variety of An inductive (longer than /2) element
PCs and minicomputers. Although the W2PV placed close to a dipole radiator builds up
method is fast and surprisingly accurate, it currents tending to offset the field in its
doesnt predict the fine sidelobe structure of direction, so it is said to act as a reflector. A
Yagis to the same accuracy as good-quality capacitive (shorter) element enhances the field
range measurements or MININEC [2]. More and thus the energy flow in its direction,
recently it has become possible to check the and is therefore called a director.
MININEC calculations on long Yagis using There are mountains of literature on the
NEC itself, and although MININEC is a optimum dimensions of Yagis with small
definite improvement over the W2PV pro- numbers of elements, for the reasons Ive
gram, some puzzling discrepancies remain [3]. already mentioned. Except perhaps for use on
To sum up the present status of computa- 50MHz and 70MHz, such short arrays are of
tional methods: in skilled hands, a program little use to the VHF DXer because of their
such as NEC can give results which are as relatively low gain. If youre not bothered
reliable in general terms as good-quality about a few tenths of a dB, scaling down a
range measurements. But neither computation proven HF Yagi should serve the purpose if a
nor measurement can be trusted blindly, so small antenna is all you desire. Although no
the way forward lies through a combination more than about 8dBd can be obtained from a
of computation and practical experiment. full-size 4-element Yagi, this gain is a bargain
Steve Powlishen K1FO and Rainer Bertelsmeier compared to the small area required. For
DJ9BV have been particularly active in short-range ionospheric and meteor-scatter
developing computer-optimized Yagis for working, where the wave arrives at a signifi-
432MHz and then testing them successfully cant angle above the horizon, up to 6dB more
on the EME path. gain can be added by ground reflection if the
antenna is at the right height above an
SHORT YAGIS unobstructed foreground [4] see the panel
Yagis with small numbers of elements are now on Ground Reflections opposite. So short
extremely well understood. Lawsons articles Yagis can be useful for medium-range DXing
and book [1] give a very thorough explanation on 50MHz in particular, for exactly the same
of what we VHFers would call electrically reasons that similar Yagi antennas are widely
short Yagis for HF and 50/70MHz, so I will used for HF communications.
just recall a few basic facts. But most VHF/UHF propagation modes
Currents induced by an electromagnetic require high gain and low-angle radiation.
field in wire or rod elements close to /2 long Short Yagis have very broad vertical (H-plane)
7 15
BEAM ANTENNAS & FEEDLINES
patterns which mean a lot of useless high- waveguide. As the frequency increases, the
angle radiation when polarized horizontally. directors become electrically longer and
To overcome this problem, and to fulfil our eventually become resonant half-waves. When
ever-increasing demands for gain, two solu- this happens, or when the element spacing
tions have emerged: vertical stacking of simple becomes greater than a half-wavelength, the
antennas, which Ive already begun to discuss, structure will no longer carry a travelling
and the development of very long Yagi arrays. wave; the velocity of propagation becomes
zero. That is why the performance of a long
LONG YAGIS Yagi falls catastrophically above its high-
The history of long-Yagi development is a frequency cutoff point. Conversely, making
fascinating tale, marked equally by experi- the directors shorter increases the velocity of
mentation on a heroic scale and by a cata- the travelling wave towards the velocity of
logue of unnoticed mistakes, proliferating radio waves in free space.
errors and almost total lack of co-ordination. The same wave velocities, and hence the
Some of the best work of the sixties remained same overall antenna performance, can be
unpublished for twenty years, by which time achieved by an infinite variety of element
it was already becoming obsolete; I refer of lengths and spacings. But instead of experi-
course to the NBS Yagis [5]. menting with an infinite pile of aluminium
It is only a slight over-simplification to say rods (or these days, an infinite number of
that the breakthrough for very long Yagi computer optimization loops) its much
antennas was reached by returning to Yagi simpler to look at the velocity profile of the
and Udas original concept looking at a Yagi travelling wave. By proper distribution of the
as a waveguide structure. As I said earlier, a wave velocity along the directors, you can
chain of director elements forms a kind of optimize the gain, sidelobe attenuation and
bandwidth all at the same time, and almost
without compromise. The proper velocity
20 distribution for this purpose is close to
logarithmic changing by the same ratio in
18 each unit of boom length and is achieved in
GAIN OVER DIPOLE, dBd
7 16
TYPES OF DIRECTIVE ANTENNAS
times, or 168dBd, so doubling the length has bandwidth indicates that the antenna is not
increased the gain by only 24dB rather than optimized.
3dB as we might have hoped. All these performance features are readily
Fig. 7.5 shows that the gain relationship obtainable with modern long Yagis. Do not
works over a very wide range of antenna settle for anything less!
lengths, from about 2 to over 50! Even so, it Although many existing Yagis fall short of
is not a rigid limit, and some experimenters what can be achieved today, there are plenty
claim to have surpassed it by a few tenths of a of other reasons why the Yagi has become
dB. Unfortunately the high gain-per-length almost the universal VHF antenna. Construc-
values of Yagis shorter than 2 are not to tion is relatively simple; electrical connections
anyones present knowledge obtainable with are restricted to the driven element; and the
long antennas, and there also seem to be good main extension is horizontal rather than
theoretical reasons not to expect that particu- vertical, making Yagis easy to mount and
lar kind of miracle [4]. rotate. Recognition of all these advantages by
The difference between E-plane and H-plane industry has paved the way to mass-produc-
patterns of short Yagis diminishes with tion of Yagis for TV reception. Many TV
increasing length, and beyond about 5 boom antenna manufacturers have produced
length the beam cross-section approaches a amateur Yagis as well, and of course some still
circle. Even so, the first H-plane sidelobes do.
remain consistently around 2dB stronger than
those of the E-plane. The first sidelobes should STACKING YAGI ANTENNAS
always be attenuated to 15dB or better, and Like any other antennas, Yagis can be stacked
all higher sidelobes should be at least 20dB to produce more gain. However, unlike some
down. other antenna types, these stacked arrays
Front/back ratio should exceed gain over a make use of all three dimensions depth as
dipole by at least 3dB; in other words, what- well as height and width to produce a
ever the forward gain, the rear lobe should mechanically balanced structure.
register 3dBd or lower. The attainable front/ Among the most frequently asked questions
back ratio appears to depend on the overall about Yagi antennas is that of correct stacking
boom length in a periodic fashion, and certain spacing. Although Yagi antennas obey the
lengths seem naturally better than others. general rules given in the earlier section on
Even so, a particularly good front/back ratio Stacking, a few additional remarks are in
only represents a dimple in a cone of energy order. Optimum stacking rules are based on
radiated in the general backwards direction, the assumption of negligible mutual influ-
and the total amount of backward radiation is ence, ie no detuning of one antenna by
more important than the spot figure at 180. another. It is hard to predict which antennas
Multiple reflectors can reduce the frequency- are critical in this respect, though we do have
consciousness of the front/back ratio, particu- a few clues. Practice has shown that antennas
larly if they approximate a solid reflecting with low overall sidelobe levels are less
screen, but it also seems that a well-optimized sensitive to proximity effects than those with
Yagi leaves less scope for improvement by a more ragged pattern. Antennas operating at
using exotic reflector structures [6]. the high-frequency limit of their bandwidth
The final aspect of performance in a good are more likely to be pushed over the edge
long Yagi is that the gain and VSWR band- this can usually be recognized from the
widths should both be at least 5% of the pattern by the absence of a pronounced null
operating frequency. This can be achieved between the main lobe and the first sidelobe.
with a perfectly ordinary dipole feed. In other The decision whether to stack a few very
words, a well-optimized long Yagi is also very long Yagis or a greater number of shorter ones
frequency-tolerant, and an excessively narrow depends to some extent on the desired shape
7 17
BEAM ANTENNAS & FEEDLINES
of the overall pattern see the following exaggeration to say that you can ignore their
section on the choice of antennas for various presence completely. The performance of a
propagation modes. From a standpoint of highly-optimized 144MHz antenna is likely to
volume economy, the shape of the array be noticeably affected by 432MHz antennas
should approach a cube, which would mean within its capture area.
arranging four medium-length antennas in a A few other rules of caution will also apply,
box configuration, or 16 very long Yagis in a which could be summed up like this:
44 stack. Anything more ambitious than that
looks like a case for a parabolic dish. Keep anything which might resonate out of
Another problem for the all-round VHF/ the high field-strength regions.
UHF DXer is how to stack antennas for The higher-frequency antenna will rarely
different bands. This is an area where compu- do much harm, but cables and mounting
ter studies can cast some light, and the general frames may.
conclusions are just what youd expect The sensitive space around a Yagi is
anyway. Basically, the antenna for the lower something like a tube along its length,
band doesnt mind how close you put the spreading out into a cone beyond the end; so
higher-frequency antenna. From the view- watch out for very long lower-frequency Yagis
point of a 144MHz Yagi, a 432MHz antenna is protruding forwards into that cone, regardless
only an electrically small object, even though of the vertical separation.
its inside the capture area of the 144MHz
antenna. But from the viewpoint of the
432MHz array, any part of the 144MHz Yagi
within its capture area will cause a major
disturbance. So the minimum stacking
distance should be the radius of the higher-
frequency antennas capture area, ie somewhat
more than half the recommended stacking
distance for a pair of those antennas.
Stacking single antennas for several bands is
therefore pretty simple. You start the Christ-
mas tree with the largest and lowest-fre-
quency antenna at the bottom and work
upwards with gradually decreasing vertical
spacing, until you reach the highest band of
interest or until the whole thing gets blown
over. In principle it would improve the low-
angle radiation on the lower VHF bands if you
put the longer-wavelength antennas at the top
and work downwards, but practical mechani-
cal considerations usually forbid this.
Problems may arise when stacking pairs of
long Yagis for different bands such as 144MHz
and 432MHz, or nesting boxes of four Yagis,
because the total separation between the
144MHz Yagis may not be sufficient to fit in
the 432MHz Yagis without a clash of capture
areas. Also, the antennas for the higher band
are still within the capture area of those for
the lower band, and its something of an
7 18
CHOICE OF ANTENNAS
7 19
BEAM ANTENNAS & FEEDLINES
7 20
CHOICE OF ANTENNAS
short. High power and high-gain antennas are the polarizations circularity. Close to the
mandatory. But once again, the narrow pattern horizon, linear polarization is often advanta-
of high-gain antennas will reduce the chances geous, horizontal being favoured in practically
to see the same meteor trail from both all cases over vertical. So all in all, amateurs
locations. Precision planning is needed to who dont specialize in satellite work are not
maximize the chances of a successful contact, sacrificing much by using linear polarization.
and this may include elevating the antennas A good tropo combination array for 144MHz
except on extreme horizon paths. The general and 432MHz, interspaced with due care and
system requirements for 432MHz MS are not providing full steerability, has all the perform-
far below those for EME, together with the ance necessary for satellite operation on these
high-speed keying and decoding equipment as two bands.
used on 144MHz, which is essential with the
short bursts typical of 432MHz. MOONBOUNCE
EME (earth-moon-earth) operation is perhaps
AURORA the greatest technical challenge to the VHF/
Radio aurora occurs on all VHF bands, and UHF amateur. Success depends on a number
also on 432MHz. Antenna requirements are of prerequisites, each of which are highly
similar to those for meteor scatter, but they demanding. There are physical limits to
also depend on your geomagnetic latitude and receiver sensitivity, and legal limits to trans-
the typical location of auroras in your part of mitter power. Operating skill can be learned,
the world. A station situated in the north will but the decisive role is played by the antenna
be able to make many contacts using rela- system. It must be large theres no way
tively small antennas, though more gain (and around that and it must be quiet. It must
elevation control) would greatly extend the also be accurately steerable, and should
range. Stations further south can do without preferably be strong enough to survive the
elevation capability, but need as much gain as next winter. Even though many parts today
possible. This is particularly true for 432MHz can be bought off-the-shelf, it still takes plenty
where signal levels are 15 to 20dB lower than of engineering work to turn them into a
on 144MHz. Adjustable polarization could be functioning system. There is a good deal of
useful, but is difficult to arrange, especially in special literature on EME antenna techniques,
a steerable and tiltable Yagi array. and only a general idea can be given here. But
anybody setting out to erect a large VHF/UHF
SATELLITES array for any purpose should also at least
Satellite communication might not be gener- consider the EME requirements before making
ally accepted as a variety of VHF DX opera- the final decisions.
tion, but that is a matter of definition. There is The 50MHz band is not very favourable for
a certain technical challenge in exploring the moonbounce operation. In Europe, power and
extreme conditions which still allow contacts licence limitations preclude operation on this
low power, horizon passes etc. Specialists band or 70MHz in any event; the hurdles are
tend to use antennas with circular polariza- high even with the power levels permitted in
tion, preferably switchable from left-hand to the USA. Theoretically the round-trip path loss
right-hand. Under most operating conditions is around 178dB, so antenna arrays providing
there is a definite advantage over linear about 15dBd should be sufficient. This would
polarization; the reduced spin modulation call for a quad array of four medium-sized Yagis
plus the (theoretical) 3dB gain over an equal- spaced some 10m apart. For example, WA4NJP
length linear antenna can be quite noticeable. has had good success with four eight-element
However, many operators throw away the Yagis of the W1JR design [7,8], and has worked
advantage by improperly mounting or feeding stations using as few as two long-Yagis
their antennas and thereby severely distorting although I really do mean long!
7 21
BEAM ANTENNAS & FEEDLINES
The 144MHz band carries most of the EME handy than on 144MHz. If you have limited
activity, closely followed by 432MHz. New- space or intolerant neighbours, youd do
comers often ask where they should start. If better to try this higher band. Four really long
you have plenty of space and are not afraid of Yagis (eg 10 boom length) at approximately
building a big antenna, 144MHz is easier. Four 3 spacing can produce the necessary gain of
13dBd Yagis spaced about two wavelengths about 23dBd. The antennas must be of
apart will do the job. There are several stations modern low-sidelobe design and have a good
already using arrays of this size for terrestrial front/back ratio to make full use of the low
DX, so if youre contemplating an array of this sky-noise levels at 432MHz. Though contacts
size, consider adding an elevation drive for have been achieved with even smaller arrays,
EME. The main lobe of a four-Yagi array as the recommendation given here should be
described is about 15 wide in both planes, rated as a minimum to be able to hear your
making steering fairly uncritical although own echoes frequently and therefore to work
heavy-duty mechanisms are of course manda- stations with similar or better equipment.
tory. A station equipped along these lines is You cant cut corners with such a small
not top-notch for 144MHz EME but is cer- array for 432MHz EME. You need short, high-
tainly capable of contacting at least 50% of grade cable connections between the
the other active EME stations around the feedpoints and a low-loss power divider
world enough to broaden anyones operat- mounted out in the rear of the array, and a
ing horizons! good preamplifier connected straight on to
Moonbounce operation on 432MHz places the divider. Better still (if you do it correctly)
higher demands on equipment performance. would be open-wire feeders using no power
The path loss is 9dB higher than on 144MHz, divider. The next step up from four Yagis is an
so the antenna system must make up for half array of six, which also allows relatively short
the difference on each end. Still, the size of a cable runs to a single 6-way divider. Compared
sufficiently high-gain array is much more to any 144MHz EME array this is still a rather
7 22
CHOICE OF ANTENNAS
DESIGN GUIDELINES
Before plunging into an antenna construction
project think twice. Make up your mind
which bands and propagation modes are most
important to you. Consider the gain and
pattern requirements and try to combine
them in as few arrays as possible. The chart
opposite may help you make these decisions.
Meanwhile Ill describe a few Yagi designs you
may wish to consider.
7 23
BEAM ANTENNAS & FEEDLINES
7 24
DL6WU YAGI DESIGNS
0.46
0.45
0.44
DIR. DIA.,
0.43
0.42
DIRECTOR LENGTH,
0.41
0.002
0.4
0.003
0.39 0.004
0.38 0.005
0.37 0.007
0.36 0.01
0.35 0.015
0.34
0.02
0.33
1 6 11 16 21 26 31
DIRECTOR NUMBER
slightly improved forward gain. These espe- elements which launch the travelling wave,
cially favourable lengths occur in the Yagis the director lengths reduce in a logarithmic
with ten elements, 14 or 15 elements, 19 or 20 manner, each being a constant fraction of the
elements, and so on at intervals of approxi- length of the one before. Also, thicker ele-
mately five directors or 2 in boom length. ments need to be shorter to give the same
On the other hand, if you are mostly inter- reactance as thinner ones.
ested in gain, just go for the longest boom you To use Fig. 7.7, first select an appropriate
can manage. element diameter. If it doesnt correspond to
As already stated, the performance of a one of those plotted, sketch in your own
parasitic director is closely related to the curve using the plotted curves as guidelines.
reactive component of its impedance. Ele- Then simply read off the length of each
ments with equal reactances are essentially director, in wavelengths, and convert to the
interchangeable. Reactance depends on both frequency of your choice.
electrical length and diameter, so a chart of The driven element and reflector lengths
optimum element lengths and diameters will should both be a little less than 05, though
look very much like a reactance chart. Fig. 7.7 both may need to be adjusted. The driven
shows the optimum length for the first, element length should be adjusted for the best
second, third... director as a function of match to 50, and the reflector length for
element diameter in wavelengths. This chart is best front/back ratio. Neither length is
valid for Yagis with eight or more directors, ie particularly critical as to element diameter
ten or more elements in total. except for very thick elements, and the
As you might expect, the element lengths following table gives suitable starting values
taper along the boom. After the first few (in wavelengths) for experiment:
7 25
BEAM ANTENNAS & FEEDLINES
BOOM EFFECTS
Unless you use only insulating materials to
support the elements, the presence of a nearby
metal boom and other metal fittings will
partially short-circuit the centre of the
element and reduce its electrical length. So
you will always have to lengthen the elements
to compensate. Ignorance of these boom
effects led to Yagis gaining a reputation for
being difficult to reproduce, and later to some
curious attempts to avoid the need for correc-
tions. However, boom-effect corrections are
now quite well understood and you can
approach them with some confidence.
The required correction depends on the
7 26
DL6WU YAGI DESIGNS
element mounting method, and on the corrections are smaller. Fig. 7.8 shows some of
thickness of the boom. It also depends to a the options for element mounting.
smaller extent on the length and diameter of Corrections for elements mounted through
the element itself. But your first option is the centre of a square or round metal boom
whether to use a metal boom at all or some- and securely connected to it (Fig. 7.8A, B) are
thing made of an insulator such as wood or shown in Fig. 7.9 as a function of boom
fibreglass. The attraction of insulating booms diameter in wavelengths. The correction to be
is that they require no element length correc- added to all element lengths is approximately
tions. Wooden booms are often touted as one-quarter of the boom diameter for a boom
being low in cost, but in fact the costs of diameter of 001, and a somewhat larger
buying good-quality, straight-grained, knot- fraction of the boom diameter for thicker
free timber and of weatherproofing it so that booms. A diameter of 001 corresponds to
it remains an insulator are not insignificant. 20mm at 144MHz, so youd need to add about
Fibreglass is a better insulator than wood, but 6mm or 025" to all element lengths if you
is less rigid and much more expensive. chose this mounting method. For elements
If you are using a metal boom, your basic sitting on top of a square metal boom (Fig.
options are to connect the elements securely 7.8C), the correction is about the same as that
to the boom or to insulate them in a fixed for mounting through the middle; but once
mechanical relationship to the boom. Of the again youd need to ensure a solid long-term
two, the all-metal method has been more electrical connection. The correction for
thoroughly explored but is open to the effects elements either touching or almost touching
of corrosion. Insulated mounting schemes to the top of a round boom (Fig. 7.8D) is ap-
side-step corrosion problems are more recent proximately halved. Elements mounted on
and less well characterized, but fortunately the insulators further away from the top of the
boom (Fig. 7.8E) require very little adjust-
ment, the correction dwindling rapidly from
Fig. 7.9. Boom-effect correction to be added 50% of the through-boom value to 5% or less
for elements passing through the centre of a for elements mounted one boom-radius away
metal boom, with a metal-to-metal from the boom surface.
connection at each side (as in Fig. 7.8A and Another mounting method which has
7.8B), or connected across the top of a
recently gained popularity is to mount the
square boom (as in Fig. 7.8C)
element through the centre of the boom on
insulating shoulder washers. The correction
required depends on the type of washer, and
Fraction of
boom diameter also whether metal keepers or pure friction
0.6
are used to hold the element in place (Fig.
0.5 7.8F, G). K1FO has published corrections for
elements of 3/16" diameter at 432MHz, for a
CORRECTION
7 27
BEAM ANTENNAS & FEEDLINES
commercial antennas from which you might The new value for L is 9142mm, which is
salvage the fittings. Fortunately its quite easy 25mm longer than the original 35", as youd
to measure the corrections using a two- or expect for a thinner element. The program
three-element Yagi with an insulating boom: disk mentioned earlier has a simple BASIC
just replace the original director with a test program called ELE to automate this entire
element mounted in your chosen manner and process.
change its length to restore the original VSWR Whatever element diameters you select, the
conditions [13, 14]. best material is generally aluminium alloy
preferably a corrosion-resistant grade, espe-
ELEMENT DIAMETER cially if you live in an industrial area or near
In presenting the design tables for DL6WU the sea. Welding rod of 4mm diameter is an
Yagis, Ive already mentioned the effects of excellent choice for both 432MHz and even
element diameter (thickness). The same effects 144MHz, although it may not be available in
apply if you want to build some other pub- the metre lengths required for the latter band.
lished Yagi design using elements of a differ- Hard-drawn enamelled copper wire of about
ent diameter. This often happens because 3mm diameter can be used for 432MHz
continental Europe works in metric sizes, the elements. Brass is not a good material for
USA works in inches, and the UK is caught antenna elements, because it is prone to
between the two which means that whatever fatigue cracking. Although stainless steel is
size you need, it will always be out of stock! springy and highly resistant to corrosion, it is
The way to tackle this problem is to remem- also lossy; its use may cost up to 05dB of gain
ber that elements having equal reactance are at 144MHz and even more at 432MHz.
interchangeable. So you first calculate the Element diameters for 144MHz and below
reactance of the original element, and then depend critically on whether your winter
calculate the length of the new element climate includes storms which build up thick
having a different diameter but the same and heavy layers of ice. If so, the elements
reactance. must be strong enough to support the ice
A formula for element reactance is: loading, which rules out thin aluminium rods
2l 2L and tubes. Ice storms are fortunately rare in
X = 430 3 log10 - 320 - 1 + 40W most of the UK, so element diameters as thin
D l
as 9mm (3/8") can be used down to 50MHz.
where D is the element diameter and L its Elements for 50 and 70MHz made from thick-
length. This is a bit of a mouthful, but is easily wall 9mm tubing can withstand practically
handled by a programmable calculator or a any gale, if reinforced at the centre with a
home computer. 12mm sleeve (no length correction is neces-
Lets assume that we start with a 144MHz sary if the sleeve is less than about 200mm
director element of 3/8" diameter, length 35", long).
and we want to convert this to use 4mm- By the way, if you are modifying someone
diameter aluminium welding-rod. Inserting elses practical design, youll need to remove
the numbers into the formula (dont forget to his boom-effect correction before applying the
use in inches, the same as the other dimen- above formulae and then apply a new correc-
sions) gives a reactance of 791. Now work tion for your own element mounting method.
the formula backwards to calculate a new The hardest part of this is to estimate some-
value for L, given X and the new D=4mm. one elses boom-effect correction, though
fortunately more and more designers are
mentioning this in their published articles.
L=
( X - 40 ) + 1
l
2l 2 MAKING THE ANTENNA WORK
430 3 log10 - 320
D Adjustment of a DL6WU or any other type of
7 28
DL6WU YAGI DESIGNS
Yagi should be straightforward, especially if The gain reduction is less rapid if the
you know what to expect. elements are all too short, so its better to err
First, check the VSWR. All the DL6WU Yagis on that side than to fall over the steep cliff
give a fairly good match to 50, although this on the HF side of the gain-frequency curve.
can be improved in any particular case by Rain will also move the antenna HF on its
minor adjustments to the lengths of the gain-frequency response, so all the DL6WU
folded dipole and the first director, which Yagis contain a 1-2% element length allow-
functions mainly as a matching element. Note ance to guard against this possibility. K1FO
that I said minor changes in those two has found that stacking his own highly-
element lengths, to improve a match thats optimized Yagis tends to detune them in the
already reasonably good. If the antenna shows same HF direction; he believes that antennas
absolutely no sign of matching properly, that are already close to the gain cliff can be
theres a fault somewhere in its construction pushed over when stacked in arrays, with
and you will need to find it and fix it. Other disappointing results [6, 12]. If part of the
types of antennas which require matching process of optimization has been to move the
adjustments are best dealt with by pointing Yagi to the top of its gain curve, any change
the antenna towards the sky and adjusting the due to interactions or rain will obviously be
driven element while its comfortably within even worse.
reach (Fig. 9.43).
Any major faults in antenna construction
will show up in the radiation pattern. As a
preliminary check, look for a symmetrical E-
plane pattern with deep nulls at 90 and
reasonably well-suppressed minor lobes. You
cant tell too much from the rear lobe,
although it should be well-suppressed, but the
first sidelobes can be very revealing. The
following remarks apply only to long multi-
element Yagis; Yagis with six or fewer ele-
ments tend not to have first sidelobes at all, or
else they are completely hidden by the 90
nulls.
All the DL6WU long Yagis have first E-plane
sidelobes about 17dB down, which should be
separated from the main lobe by a distinct
null. If the nulls are too deep and the first
sidelobes are too well-suppressed (19dB or
more), the antenna is operating on too low a
frequency, ie its elements are too short. This
can easily happen if youve underestimated
the electrical shortening effect of the boom.
On the other hand, if the first sidelobes of a
long Yagi are poorly suppressed (15dB or less)
or merge into the main lobe, the antenna is
operating on too high a frequency, ie its
elements are too long. Either case all ele-
ments too long or all too short will produce
less than the maximum attainable forward
gain.
7 29
BEAM ANTENNAS & FEEDLINES
WHICH COMMERCIAL
YAGI?
7 30
WHICH COMMERCIAL YAGI?
YAGI PERFORMANCE
144MHz
Design/type Boom Ele Gain (dBd) Gain/exp E H 1sl F/B Notes and
() NEC Claim (dB) () () (dB) (dB) references
Design/type Boom Ele Gain (dBd) Gain/exp E H 1sl F/B Notes and
() NEC Claim (dB) () () (dB) (dB) references
Youll also surely have noticed that there is great trouble to make realistic and verifiable
scarcely a single instance of over-modest performance claims for their current produc-
claimed gain! Some claims are grossly inflated, tion antennas. This is a highly creditable
especially for the older designs and for certain policy, especially since new antennas may have
manufacturers. It is ironic that many of the lower claimed gains than the products they
antennas with highly exaggerated claimed replace. A manufacturer who has had the
gains are actually quite good, so the advertising courage to take this step should not be re-
hype was unnecessary. Some manufacturers, warded by clever people asking embarrassing
for instance Antennes Tonna, are now taking questions about their old catalogues!
7 31
BEAM ANTENNAS & FEEDLINES
FEEDLINES,
TRANSFORMERS
AND BALUNS
A chapter on antennas would be
incomplete without consideration of the
feedlines connecting them to your station.
of the conductors themselves as a source of
loss. As the power dissipated is proportional to
the square of the current, higher values of Z0
After all, a 3dB cable loss means converting are more favourable. For example 2mm wires
half your precious power into heat, and to spaced 100mm apart have a characteristic
make up for that loss youd have to double the impedance of 552. Assuming a /D ratio of
size of your antenna at least. Impedance 25 to be acceptable, this line could be used on
transformers are important because they are HF and up to 70MHz with excellent results.
used as power dividers in stacked arrays. Ill For 144MHz, the spacing must be reduced to
also have something to say about the much- avoid radiation losses, and rather than use
misunderstood subject of baluns. thinner wire its generally better to accept the
slight increase in resistive losses due to the
FEEDLINES lower value of Z0. Such an open-wire line
In 99% of all cases, feedline today means would still outperform half-inch coax as far as
coaxial cable. But it wasnt always so; VHF loss is concerned.
amateur radio began with home-made open- The situation on 432MHz is a little worse,
wire feedline, and open wire has recently made though open wire still can have lower losses
a comeback in the highly demanding area of than coaxial cable. A further reduction in
moonbounce antennas. So lets begin by taking spacing is dictated by the wavelength, but it
a critical look at both types of feeder. also becomes reasonable to increase the wire
Open-wire feeders or parallel lines are diameter to make the feeder semi-rigid and
characterized by their spacing and wire almost self-supporting. With a spacing of
diameter. These parameters determine the 25mm and 2mm wire, Z0 is reduced to 386.
characteristic impedance (Z0) by the simple Other problems then emerge, like keeping the
relationship: spacing constant, preventing dirt from
forming current bridges on the spacers, and
D
Z0 = 276 log10 the field distortion by the spacers which can
r
upset your calculations of impedance and
where D is the centre-to-centre spacing and r velocity factor. Even so, there is still a case for
is the wire radius. using open-wire feeders for interconnections
In principle, open-wire feeders can have within an array of several Yagis for 432MHz
very low losses. Apart from the insulating EME, where 01dB may make a noticeable
spacers, the dielectric between the lines is air difference. 432MHz marks the upper-fre-
and only a vacuum has lower dielectric losses. quency limit of parallel-wire feeders. I wont
Radiation losses are surprisingly low provided even discuss TV ribbon, which has no place in
that the wire spacing is a small fraction of a serious VHF/UHF work.
wavelength, leaving only the ohmic resistance Turning to coaxial cable, the advantages are
7 32
FEEDLINES, TRANSFORMERS & BALUNS
7 33
BEAM ANTENNAS & FEEDLINES
the inner conductor flows only on the surface, main cable run, and look after it well, because
so the entire conductor doesnt need to be its one of the most important parts of your
solid copper. A solid copper conductor is station.
preferable in small cables because it has lower
losses than a stranded conductor, but in larger IMPEDANCE TRANSFORMERS
cables the centre conductor is often either a AND POWER DIVIDERS
copper coating on an aluminium core or a Transmission lines find widespread usage as
hollow copper tube. matching devices in VHF/UHF equipment.
The table of cable data is intended mainly to Although in principle they can accomplish
illustrate the differences between various cable any kind of complex impedance transforma-
types rather than as a complete reference. It tion, we amateurs rarely have instruments
also serves to show that RG213 (URM67), the suitable for measuring complex impedances at
old standby, should really be used only where VHF and UHF under outdoor conditions so we
it cannot be avoided. The other 50 cables are try to make our antenna impedances straight-
manufactured and sold under a variety of forward and non-reactive. We can therefore
names (Flexwell, Cellflex, Heliax etc). Except manage very well with simple quarter-wave
for the low-cost H100/W103/9913 the best transformers for matching to the feedline.
cables are of rugged construction with solid
corrugated copper shields. For amateur COAXIAL TRANSFORMERS
installations the polyethylene and PTFE foam Transformation between real (ie non-
types are the easiest to use. reactive) impedances is easily checked by
Special highly flexible versions of these VSWR measurement and substitution of loads
cables (eg Kabelmetall HCF) are ideal where by terminating resistors see Chapter 12 for a
frequent bending is necessary, although the full explanation. To transform between two
connectors are heavy and costly. real impedances Z1 and Z2, all one needs is a
Summing up the practical implications of quarter-wave line with the characteristic
all these various kinds of coaxial cables, it is impedance Z0 = Z1 Z2 . Coaxial transformers
advisable to make the long cable run from for outdoor use are most easily made from
stiff, low-loss cable and use a short piece of square tubing, with cylindrical wire or tubing
flexible line to bridge the rotator. If the bridge for the centre conductor (Fig. 7.10). The
is made using good connectors (a pair of N- characteristic impedance of such a line is:
types has less than 005dB loss at 432MHz)
D
that is a tolerable sacrifice. Z0 = 138 log10 1 08
d
Be prepared to spend good money on the
COAXIAL CABLES
MATCHED ATTENUATION OF 50
Type Atten (dB/100m) OD CC Dielectric Notes
50MHz 70MHz 144MHz 432MHz (mm) (mm)
7 34
FEEDLINES, TRANSFORMERS & BALUNS
7 35
BEAM ANTENNAS & FEEDLINES
you use these tricks, make sure the effects the horizontal cross-piece of feeder depends
apply equally to both sides of the power on the horizontal stacking distance for the
division; you can only do this for two-way or particular Yagis in use. Shorter Yagis which
half-wave four-way dividers. require horizontal spacings in the range 19-
You can leave the ends of the inner tube 22 can use a 20 (electrical) cross-piece as
open. The ends of the square tubes are also shown in Fig. 7.12A. This transforms the
field-free if they extend a few centimetres feedpoint impedance of the array of four Yagis
beyond the connectors, so they only need to to 50, which can be connected to the coaxial
be sealed against water ingress. It would be feeder via a /4 sleeve balun. For longer Yagis
nice to make one end-cap from something requiring horizontal stacking distances of 22-
transparent, so you can see at a glance if any 25, the cross-piece should be 25 (electrical)
water has crept in. which gives a feedpoint impedance of 200,
The following table shows the impedances the same as an individual Yagi.
to be used for quarter-wave and half-wave 50 The velocity factor relating the physical to
combiners, and the corresponding D/d ratios. the electrical length can be very high for
Since we all seem to have access to differing open-wire feeders as high as 0975 if the
sizes of tubing, you probably need to take a dielectric spacers are few in number and
pocket calculator to the metal shop and work located away from voltage maxima. Measure-
out what can be done with the materials ment of the velocity factor is very simple.
available. Connect a length of the open-wire line to a
Z0 D/d low-power RF source (eg by a 4:1 coaxial
Two-way /4 354 167 balun), terminate the far end and monitor the
Three-way /4 289 150 VSWR in the coax line from the RF source. If
Four-way /4 250 141 you lay an insulated screwdriver across the
Four-way /2 500 213 open wire and slide it along, you will see a
Six-way /2 408 183 disturbance in the VSWR which repeats itself
The largest number of Yagis to be joined in every electrical half-wavelength.
a single combiner is six. For an eight-Yagi Doing away with baluns, cables, transform-
system, make two groups of four and then ers and connectors saves a lot of weight, loss
connect the two groups to a two-way and money. But there is a price for everything;
combiner at the central feedpoint. A sixteen- open-wire feeders are quite critical in dimen-
Yagi system will require a total of five four- sions and installation and they can also be
way combiners plus at least 25 N-type plugs weather-sensitive and mechanically unstable
and sockets... quite a strong argument for unless youre very careful. And open-wire lines
open-wire feedline! cannot be lengthened or shortened very
easily, so youd better be sure about your
OPEN-WIRE FEEDLINE velocity factor and stacking distances before
The inherent balance of parallel open-wire you commit yourself.
lines makes them well suited for both inter- When working with open-wire feeders,
connections and impedance transformation. small impedance errors tend to accumulate
Yagi antennas with 200 folded-dipole driven and the bandwidth of satisfactory perform-
elements are ideal for feeding with 200 open ance is usually very narrow; fortunately, this
wire. Fig. 7.12 shows two schemes for combin- doesnt matter for DX antennas. Be prepared
ing four Yagis in this way, with very simple to do a lot of experimenting, but also be
matching to 50 coaxial feeder. Since the prepared for some benefits: you can trace the
Yagis and the open-wire feedline have the voltage maxima and minima by just waving
same characteristic impedance, any vertical your hands around the feedline and watching
stacking distance can be used without affect- the VSWR, and you can also fine-tune the
ing the impedance matching. The length of impedance match by bending the wires in the
7 36
FEEDLINES, TRANSFORMERS & BALUNS
right places! Naturally, these hands-on 9 Avoid high VSWR in any section of feeder,
adjustments should be made using extremely to minimize the effects of rain and frost at
low power (Chapter 12). Most open-wire feed high-impedance points. The two arrange-
systems require an adjustable impedance ments in Fig. 7.12 involve a VSWR nowhere
transformer somewhere in the system, greater than 2.
probably at the central feedpoint. This can be The overall result may be well worth the
combined with the symmetry transformer effort. EME operators have reported up to 3dB
(balun) required for the transition to coax. improvement in noise performance over crude
The rules for installing open-wire feeders coax feed in large 432MHz arrays, although
have been set down by DL9KR and G3SEK in that is extreme. More often you could expect
K2UYHs 432MHz Moonbounce Newsletter. an open-wire feed system for a large 432MHz
1 Use heavy-gauge enamelled copper wire, eg EME array to be at least as good as a coaxial
3mm, to make the feedline largely self- system made from heavy hard-line and
supporting. Avoid clear polyurethane enamel; probably cheaper, lighter and easier to
it doesnt stand up to sunlight. The best grade construct. But dont expect miraculous results
of enamel is dark brown, and very difficult to in horizon modes and on lower frequencies. If
scrape off for soldering. you want an all-purpose array for both ends of
2 Buy soft-drawn wire, and stretch it out the band and in any kind of weather, open-
straight immediately before use. For heavy wire feed is probably not for you.
wire, youll need a stout post or tree and a
towing hitch on the car! BALUNS
3 Use only straight runs of feeder and make Most antennas are symmetrical, ie balanced
all joints at right-angles. This is vital. with respect to ground, so a transition to
4 Pre-stretch the individual wires, and then unsymmetrical coaxial cable must be provided
construct each section of feeder to the correct somewhere in the system. A balance-to-
length before installation. Keep the two sides unbalance transformer or balun is usually
of each section of feeder exactly the same fitted at each individual antenna feedpoint, or
length. at the central feedpoint if open-wire line is
5 Use a minimum of spacers made from solid used within the array. There are many types of
PTFE rod, and avoid placing them at voltage balun; some of them involve impedance
maxima. transformation as well, although that is not
6 To make a straight joint between two wires, essential to the balun function.
use a small sleeve made from brass or copper There are a lot of misconceptions about
tubing. To make a T-joint, drill a hole across baluns and whether you really need them, so
the end of a piece of tubing. Slip one of the
wires through the hole and the other wire I1 = I2
into the open end of the tube. Crimp the
wires in place and then use an enormous
soldering iron. Waterproof all joints with
polyurethane spray; although susceptible to
sunlight, polyurethane varnish has the great
advantage that it doesnt need to be scraped
off before soldering if repairs should be I3
necessary at a later date.
7 Keep the feeder well away from other
conducting objects. Fig. 7.13. Currents flowing on the inside and
outside of a coaxial cable. Currents I1 and I2
8 Avoid using mechanical supports for the
on the inside of the cable are always equal. If
feeder, other than the antennas and the the system is unbalanced, a current I3 will
junction to the main feedline. flow on the outside of the outer sheath
7 37
BEAM ANTENNAS & FEEDLINES
7 38
FEEDLINES, TRANSFORMERS & BALUNS
7 39
BEAM ANTENNAS & FEEDLINES
[7] Ray Rector WA4NJP, Update on 6 meter [21] John Nelson GW4FRX, In Practice: Water-
EME. Proceedings of the 22nd Conference of the proofing. Radio Communication, January 1989.
Central States VHF Society. ARRL (1988). ISBN [22] Mike Gibbings G3FDW, Moxon slopes at
0-87259-209-X. VHF and other thoughts. Radio Communica-
[8] Joe Reisert W1JR, VHF/UHF World: tion, May 1988.
7 40
144MHz
8 1
144MHz
THE SUFFOLK
144MHz TRANSVERTER
by Sam Jewell G4DDK
8 2
SUFFOLK 144MHz TRANSVERTER
144MHz RF 28MHz
Filter Mixer Term.
RF input amp IF output
Crystal. 116MHz
Splitter
osc. osc. output
spent in getting the design right using account. With an IP3 of +20dBm for the HF
TCALC, and in measuring the performance of receiver, 0dBm for the receive converter and a
individual circuits to meet the requirements of gain of 20dB in the receive converter, the
the optimized block diagram. system IP3 will be 3dBm. The spurious-free
The TCALC analysis of the Suffolk receive dynamic range (SFDR see Chapter 5) will
converter is explained at length in Chapter 5. depend on the receive systems noise figure; if
The converter was designed to be used with a this is 2dB, the SFDR will be about 93dB.
modern high-performance HF transceiver. If Obviously a poorer intermodulation intercept
you use it with an older HF rig, the perform- in the HF receiver will lead to a smaller
ance of your receiving system may well fall dynamic range for your whole 144MHz
short of that obtainable with modern trans- receiving system. All that I could do was to
ceivers. With the Suffolk, it will almost always make the Suffolk front-end sufficiently robust
be the HF radio which limits the overall that it wont let your HF receiver down.
receive performance, not the transverter itself. The quest for an even better signal-to-noise
After close examination of many HF trans- ratio a smaller degradation of S/N, that is
ceiver reviews, I decided that it was reasonable will inevitably lead to poorer system dynamic
to expect an input third-order intercept of range. The balance between sensitivity and
+20dBm at 28MHz, together with a noise dynamic range achieved in the Suffolk receive
figure of 10dB. If your HF radio has worse converter is close to optimum with todays
performance than this, you will not achieve band conditions and electronic devices. So
the best possible overall performance using how is this balance achieved?
the Suffolk receive converter. The key lies in the use of a modern dual-gate
Given the above level of HF receiver per- MOSFET for the RF amplifier. These devices
formance, what can we reasonably expect are capable of low noise figures and high
from the receive converter? We learned in dynamic range simultaneously. You may be
Chapter 4 that a receive noise figure of surprised to discover that I did not use
approximately 22dB is adequate for most GaAsFETs. Although they are capable of very
terrestrial operation at 144MHz. With this low noise figures, GaAsFETs require low-loss
noise figure, a degradation of less than 3dB in input circuits which are not very suitable for
signal-to-noise ratio may be expected as printed-circuit construction; equally, the
compared with an idealized noiseless receiver. dynamic range of GaAsFETs is not very good
Dynamic range is important when the unless negative feedback is used. At 144MHz,
144MHz band is crowded with stations, as the best overall performance comes from dual-
during an opening or contest. What dynamic gate MOSFETs such as the BF981 from Philips
range can we expect and more importantly, Components.
achieve? An input third-order intercept (IP3) To assess the repeatability of the MOSFET
of 0dBm is an acceptable target for the receive amplifier design for the Suffolk, I built and
converter, but is itself misleading unless the tested six units. A Hewlett Packard HP8790
IP3 of the following HF radio is also taken into noise-figure meter with HP346A noise head
8 3
Circuit
8 4
receive
Fig. 8.2.
converter
the Suffolk
diagram of
+135 volt
P1 RF stage
C3
P2 Oscillator
C13
144MHz
R4
R3
R8
C4
C2 R2
R1 T1 L7
M1 L6 C15
C25 C26
IC1
C19 R23
C27
R15
C20 R25
C24
T2 L10 L11 R24 R26
P5 LO OUT
R13
L9 C28
C18 R18 R20
C21
R11
C23
TR2 TR4
TR3 C32
R16 C30 C31
R21
R19 C29
R12 C17
R14 X1
R22
Cx C22
R17
Gnd
SUFFOLK 144MHz TRANSVERTER
was used to measure noise figure and insertion ring mixer takes some beating. The biggest
gain, and an HP 8568B spectrum analyser to disadvantage seems to be the high cost of the
check stability. Only one amplifier gave a better devices. For this reason the mixer
noise figure in excess of 1dB, all the others specified has been deliberately limited to one
giving a value between 06 and 09dB. Inser- of the medium dynamic-range types. If you
tion gain measured 26dB 1dB on all units, have worked through the dynamic-range
apart from the one that was giving a high calculations for the Suffolk in Chapter 5, youll
value of NF; when the MOSFET was changed realise that there is little to be gained in this
in the suspect amplifier, its performance fell design by using a higher dynamic-range
into line with the other units. This MOSFET mixer. Having said that, I recommend the full-
was probably a victim of static damage caused specification SRA-1 rather than the budget
by mishandling prior to use. Dynamic-range SBL-1. The reason for this is the assured
measurements on the same amplifiers indi- performance of the SRA-1, the cheaper device
cated an input third-order intercept in the having a slightly wider tolerance. If cost is all-
range 0 to +4dBm. At no time was I able to important, use the SBL-1; you may never
obtain an IP3 as high as this with any of the notice the difference in performance. The
dual-gate GaAsFET RF stages I tried. mixer is slightly overdriven by the local
Optimum dynamic range is achieved in the oscillator to obtain a small but significant
MOSFET amplifier with a drain load imped- increase in dynamic range. The mixer is
ance of about 500. Matching this to the specified for +7dBm LO injection, giving an
filters 50 characteristic impedance is IP3 of approximately +12dBm. With +10dBm
achieved with a trifilar-wound broadband of LO input, IP3 rises to about +15dBm.
transformer. A 3:1 winding ratio achieves a 9:1 If the mixer is to operate correctly, all its
impedance transformation, which is just close ports must be terminated in a non-reactive
enough. With this drain load the insertion 50 load at all frequencies generated in the
gain measured 26dB when the BF981 amplifi- mixing process. This can be extremely difficult
ers were adjusted for lowest noise figure. to arrange, particularly when you consider the
Fig. 8.2 shows the circuit of the Suffolk wide range of frequencies involved. If the
receive converter and local oscillator. As a termination is reactive, mixer conversion loss
144146MHz bandpass filter (image rejection will increase and intermodulation perform-
filter) to follow the MOSFET RF amplifier TR1, ance will deteriorate. The IF port is particularly
I initially intended to use a commercial helical critical in this respect.
filter but was disappointed with its perform- The simplest broadband termination is an
ance, particularly the insertion loss of 8dB for attenuator, but to be effective it should have a
the three-stage filter. I also found that the value of at least 10dB and preferably 20dB. At
return loss (Chapter 12) increased dramatically the LO port I compromised at 6dB to gain the
outside the passband, leading to a poor advantage of the higher LO power level. But
termination for the preceding RF stage, and even 6dB of attenuation at the IF port would
problems in maintaining its dynamic range. have caused major difficulties in achieving the
As a result I designed a three-stage bandpass desired overall noise figure and dynamic
filter with top-capacitor coupling which range. Diplexers are commonly used to
exhibits almost ideal characteristics for the overcome this problem but most types of
intended purpose. Input and output imped- diplexer are difficult to align without access to
ance were designed to be 50 so that the filter a network analyser, so they cannot be recom-
can be easily aligned and tested. The filter is mended for home construction.
isolated from the mixer by a 50 attenuator To overcome these problems, the active IF
which also serves to optimize the overall RF termination has become popular. These
gain (Chapter 5). devices are chosen for their broadband 50
For consistent high performance the diode- input impedance and high dynamic range.
8 5
144MHz
A popular choice is a grounded-gate RF power the power output, which is then divided to
FET, for instance one of the Texas Instruments the required level for each mixer. The feed-
P8000 series. These devices perform well but back amplifier design used here is based on a
can be difficult to obtain. My own choice was design published in the excellent Solid State
a monolithic microwave integrated circuit Design for the Radio Amateur [2]. Both shunt
(MMIC), as used extensively in other parts of and series negative feedback are used to
the transverter. The MMIC chosen for use in achieve the required stable gain with good
the receive converter at IC2 has high dynamic input return loss. A lowpass filter follows, to
range, low noise and a constant 50 input reduce the harmonic level, after which the
impedance from DC to well over 1GHz. Just as output level at 116MHz is +16dBm. The power
important as their electrical characteristics, splitter has a loss of 6dB, giving two outputs of
MMICs are simple to use, low in cost and easy +10dBm. If the transmit converter is not used,
to obtain. it is necessary to terminate the unused output
What is the input impedance of your HF in 50. Failure to do so will lead to excessive
receiver at 28MHz? It is almost certainly not oscillator drive to the receive mixer, which
50, so in order to present a reliable 50 load will increase the conversion loss.
impedance to the MMIC I have included a A full TCALC design analysis of the Suffolk
simple diplexer at the output of the trans- front-end is presented in Chapter 5. The
verter [1]. The demands upon this diplexer are calculated overall gain is 185dB; the noise
far less than those at the mixer IF output, so figure is 225dB and IP3 is 52dBm. The
alignment is no problem. In addition to measured values for the real-life transverter are
providing a proper impedance termination for given later.
the MMIC, the diplexer provides a degree of The system noise figure of 225dB is ob-
bandpass filtering and a well-defined source tained at the receive converter input, and
impedance for the input filters of the HF therefore does not take into account the
receiver. contribution from the feeder between antenna
Frequency-translation stability in the and converter. Cable loss should be kept to
converter is important, especially for moon- the minimum possible, to minimise the
bounce and meteor-scatter where being on the additional noise and to reduce power loss on
right frequency can make the difference transmit. If your HF radio has a lower noise
between success and failure. It is very impor- figure than 10dB, the overall system noise
tant to use a high-stability crystal in the local figure can be significantly lower than 2dB
oscillator. A cheap crystal will almost certainly but remember that this will probably be at the
lead to frequency drift. A fifth-overtone cost of dynamic range.
116MHz crystal is used in a Butler oscillator Because 50 interstage impedance matching
(TR2-TR3). This circuit is better known as a has been used throughout, it is possible to
harmonic generator, but experience has change various stages to optimize the per-
shown it to be the best choice for frequency formance to meet your own particular require-
setting and reliable starting as well as low ments. But please remember that the Suffolk
noise and high stability. A small number of design as published is already optimized for
constructors have reported stability problems terrestrial DXing. Any improvements that
with the oscillator. The addition of an 18pF you attempt will inevitably result in a reduc-
ceramic plate capacitor (Cx) has cured the tion in performance somewhere else.
problem.
Local-oscillator drive for the transmit TRANSMIT CONVERTER
converter is also supplied from the receive The transmit converter consists of two
converter. Because of this, a higher output separate units the frequency converter and
level is required than can be supplied from the the power amplifier. The frequency converter
oscillator alone. An amplifier (TR4) increases provides an output of approximately 100mW
8 6
SUFFOLK 144MHz TRANSVERTER
116MHz
osc. input
ALC
detector
with extremely good intermodulation- output level at which the transverter operates
distortion (IMD) performance. This may be linearly. This is considerably less than the flat-
enough for some purposes such as driving a out maximum power see Chapter 6. Your
second transverter to some other band such as voice could drive the transverter beyond its
1296MHz, and in fact the original design work linear power range, resulting in distortion, and
was inspired by just that need. The transmit ALC prevents this from happening. Once ALC
converter is not designed as a stand-alone is incorporated in the transmitter chain, it
unit; it requires local-oscillator drive from the becomes simple to add some additional
receive converter. circuitry to protect the output transistor in the
For most amateurs, 100mW will not be event of a poor match at the antenna connec-
sufficient output power its hard to make a tion. ALC is achieved by a PIN-diode attenua-
dent in a pile-up with this level of QRP. The tor (D1) at the IF input, which can reduce the
second unit, which I will describe shortly, is IF drive to the transmit mixer by at least 20dB
therefore a linear amplifier designed to take if necessary. A separate variable attenuator,
the low power up to 10W. RV1, is used to set the initial IF drive to a
The block diagram of the transmit converter reasonable level so that the ALC doesnt have
is shown in Fig. 8.3, and the full circuit in Fig. to do all the work.
8.4. Starting at the transmit converter input, Following the PIN attenuator, a broadband
many modern HF radios have extremely low MMIC amplifier (IC1) provides 8dB of gain to
transverter-output drive levels, while others increase the IF level into the transmit mixer. If
have no transverter output but can only your HF rig has an even lower output than the
deliver their full HF power. The Suffolk can transmit converter is designed to accept, the
accept HF drive levels up to a few hundred MMIC can be changed for a higher-gain
milliwatts (+24dBm), or down to about one version. Details of this alternative are given
hundredth of a milliwatt (20dBm). It is later.
relatively easy to cope with the excess drive A diode-ring mixer (M1) converts the 28
all you need is an attenuator. Very low levels 30MHz IF input to 144146MHz, using local
of drive need an amplifier at IF to increase the oscillator drive at a level of +10dBm from the
level to something more usable. Greatest receive converter board. Mixing from 2830 to
versatility is therefore achieved by using a 144146MHz requires care in the choice of
wide-range variable attenuator at the transmit mixer, since an in-band product will be
IF input, followed by an amplifier to establish generated at 5 x 29MHz = 145MHz. This fifth-
the correct drive level for the mixer. order product is only inside the amateur band
It is good practice to incorporate automatic when the mixer is driven between 288 and
level control (ALC) in a transverter. ALC 292MHz, but at other IF drive frequencies it is
allows the transverter to be set up for a specific still present. The 144146MHz bandpass filters
8 7
144MHz
RV3
GAIN
C34 R17
P5
7 ALC
6
2
P2 IC5 R18
P6
ALC Output SWR
RV2 4 3
ALC SET
C33
L14
C32 C31
R14 C5
R4 R5
TR1
C4
R3
L2 L4
L1
P1
C9 C10 C12 C13
Tx IF D1 3/4
1
Input RV1 IC1 M1 IC2
C7
C1 C2 C3 C6 C8 L6
ATT
R1 L5
L3
R2 2/5/6/7 8 C11
L7
can do little to remove this product, so you are orders must be suppressed to avoid splatter
relying on the mixer to avoid generating it at (see Chapter 6). The target for the mixer
excessive levels. Diode-ring mixers provide the output was set at better than 50dB for third-
highest level of fifth-order rejection of any order products, and once again a low IF drive
type of mixer I tested, but in order to achieve level ensures the best results. This creates a
this the IF drive level must be kept low. problem of its own. The mixer has typically
Third-order distortion must also be kept low 6dB conversion loss, so assuming that the IF
to ensure a clean-sounding signal, and higher drive is kept to 13dBm maximum, the
8 8
SUFFOLK 144MHz TRANSVERTER
P7
+135V
Switched
C41 R21 Supply
P10
TR3
PTT
D4 R22
C39 P8
R19 4 8 R24 Tx
7 D2 R23 Supply
R25
IC6 3
R20
2 C38
6 D5 TR4
5 1
P9
C36 C40 Rx
D3 Supply
C37
C29
C30 R13
C28
T1
R6 R7
R10
P4
C25 L13
Low
L8 power
L9 Tx
R9 C23
Output
C17 C18 C20 C21
C14 C15
8 9
144MHz
116MHz. Two further MMIC amplifier stages, towards cutoff, reducing the bias current for
IC3 and IC4, raise the 144MHz output to a the PIN diode D1. This in turn reduces the
higher level, and a second BPF ensures that mixer drive, with a consequent reduction in
any last remnants of spurious outputs are transmit-converter output to the value
effectively removed. determined by the setting of the ALC SET
A BFR96 amplifier stage with negative potentiometer RV2. When the modulation
feedback [2] provides over 100mW PEP output peak is over, C35 slowly discharges through
with third-order IMD products typically 46dB R16 to restore full transmitter gain gradually.
below PEP (40dB below either of two tones). The values chosen for R16, R17, R18 and C35
The output stage is followed by a lowpass filter give what I believe are optimum ALC attack
with a cutoff frequency of 150MHz. and decay times. If you want to change them,
Transmit/receive power switching is pro- the section on alignment tells you how.
vided by the additional circuitry associated Optimum control of modulation peaks is
with TR3 and TR4. On receive, TR4 conducts achieved by feeding ALC voltage all the way
as a result of forward bias through D5, R24 back into the HF transmitter itself. If the
and R23. When a press-to-talk (PTT) ground is power and attenuation levels are set up
applied to P10 to initiate a changeover to correctly, all the ALC is then provided by the
transmit, TR3 is forward-biased, removing the HF transmitter, leaving the transverters own
forward bias from TR4 and hence switching internal ALC loop with very little to do. Most
off the receive-converter RF stage. TR3 also HF transmitters require a negative-going ALC
provides power to the transmit converter and voltage for gain control, and the Suffolk
linear amplifier connected to P8. provides this facility. To generate the required
The diode D4 at the PTT input provides polarity from a positive operating voltage
isolation from positive voltages that might requires the use of a polarity inverter. IC6
appear on the PTT lead, and C39 decouples generates a high-frequency pulse stream
any stray RF. In extreme cases it may prove which is connected to a diode pump consist-
necessary to add more decoupling here, to ing of C36, D2 and D3. This circuit produces a
prevent unwanted switching to transmit in rectified and smoothed voltage of approxi-
the presence of strong RF fields from other mately 10V to the operational amplifier IC5.
transmitters. The output of IC5 is made available at P2 for
Amplification of the low-level ALC and connection to the ALC input of the HF radio.
VSWR detector output voltage is provided by
the FET-input operational amplifier IC5. Gain LINEAR AMPLIFIER
is set by the ratio of RV3 to R15. By making This section describes the linear amplifier
RV3 variable the gain can be adjusted to deal producing 10W PEP output for less than
with detector output from either the 10W 100mW input. A Mitsubishi M57713 amplifier
amplifier or the barefoot transmit converter. module is used for convenience, reliability and
ALC operation calls for controlled attack its small size. These devices are available from
and decay of the ALC attenuation. These time several distributors and the cost is competitive
constants are determined by the values of R16, with a discrete-transistor design.
R17, R18 and C35. Rising power at the ALC For the constructor, the main advantage of
detector, produced by a peak of modulation, the module is the low risk of transmitting a
causes a negative-going detector output signal with severe intermodulation distortion
voltage. Initially C35 is discharged and or parasitic oscillations. Most people who have
therefore appears as a short-circuit to the tackled transistor VHF power amplifiers have
detector voltage. As C35 charges through R17, experienced some problems with stability,
the voltage at the non-inverting input of IC5 linearity and (all too often) the destruction of
goes negative. The amplified negative-going an expensive transistor. I experimented
voltage from the output of IC5 biases TR1 extensively with a 10W PA using discrete
8 10
SUFFOLK 144MHz TRANSVERTER
M57713
IC1
C2 C4 C6
L2 L3 L4
P2
D1 Output
Input D2 C9 C10
C1 C3 C5
C11 R1 C12
P1 +135V Unswitched
P3
L1
C7 C8
Fig. 8.5. Circuit diagram of the M57713 amplifier
transistors, and its performance was truly instability, and the intermodulation perform-
excellent except when it oscillated! After a ance will be adequate, though it will not reach
lot of effort, I reluctantly decided that I could the very high standards we recommended in
not guarantee good performance under all Chapter 6. Having said that, I should also
circumstances and changed over to the hybrid point out that many commercial transceivers
PA module. These devices give little trouble use just this kind of PA, and when modules
provided the correct supply and bias voltages with better intermodulation performance
are used and the RF drive level is not ex- come along in the future as they surely
ceeded; the power amplifier is extremely easy ought to the Suffolk will accept them. And
to build and requires no alignment. The dont forget that the transverters overall
disadvantage lies in the intermodulation intermodulation performance is also influ-
performance. As with any linear amplifier, enced by earlier stages in the transmit con-
modules display high levels of IMD if verter. If the bandpass filters in the transmit
overdriven, and at maximum rated output converter are not aligned properly, intermodu-
power the level of intermodulation products is lation in the poorly terminated MMICs can be
just about tolerable. But unlike most amplifi- a problem. Follow the alignment instructions
ers, you cannot produce a super-clean signal carefully!
from a module merely by under-driving it. The amplifier module is bolted to a heatsink
When the drive is reduced, the levels of the on the transverter rear panel, with its connect-
odd-order intermodulation products with ing leads soldered to the amplifier printed-
respect to the main signal do not decrease circuit board. The circuit (Fig. 8.5) is almost
significantly until the output has fallen below ridiculously simple. An integrated-circuit 2A
about 1W PEP. I suspect that this unusual regulator (IC2) provides a 9V bias supply to
intermodulation performance may be due to the module (IC1). The linear amplifier is
deliberate design compromises in the switched to transmit mode by turning on the
interstage and output matching. bias regulator via TR3, which is located on the
So the hybrid module approach is a compro- transmit-converter board. To avoid having to
mise. There should be no problems caused by switch high currents, collector voltage remains
8 11
144MHz
Fig. 8.6.
General
layout
inside the
case, shown
with old
issue A PCBs
connected to the module during both trans- on the transmit-converter board, and are used
mit and receive. for ALC control as already described. Protec-
The modules RF output is connected via a tion against output mismatches is achieved by
directional coupler to a 150MHz lowpass filter, feeding the VSWR signal from D2 back into
designed to reduce the level of all harmonics the ALC loop to reduce the drive power.
to less than 60dBc. The directional coupler
provides negative voltage outputs from D1 CONSTRUCTION
and D2, corresponding to forward and Almost any modern construction technique
reflected power. These voltages are amplified can be used to assemble the transverter, but
for guaranteed results I recommend the use of
printed-circuit boards made according to the
designs presented in the following section.
In line with the modular approach used for
each of the transverter units, three separate
printed circuit boards are used in the construc-
tion of the transverter. These boards are:
Fig. 8.7. Front panel of the Suffolk transverter 1 Receive converter (issue C)
2 Transmit converter (issue C)
Fig. 8.8. Rear panel of the Suffolk transverter 3 Linear amplifier
8 12
SUFFOLK 144MHz TRANSVERTER
8.11). The top of each board is left fully each hole on the groundplane side to prevent
copper-clad to act as an RF groundplane, and short circuits.
the interconnection tracks are etched on the A second groundplane is used to intercon-
reverse side. Components are generally nect some grounded leads on the track side of
mounted on the groundplane side with their each board. Component leads which need to
leads passing through the board and soldered be grounded are soldered on both sides of the
on the track side. Where component leads board. This form of construction ensures
pass through the groundplane without need stability in VHF/UHF applications without the
for grounding, the copper is removed around expense of a board with plated-through holes.
Fig. 8.9(a).
Full-size PCB
mask for the
receive
converter
(underside)
Fig. 8.9(b).
Full-size PCB
mask for the
receive
converter
(topside)
8 13
144MHz
The linear-amplifier board differs from the Centurion AE3, which is 280mm wide,
converter boards in that all components are 120mm deep and 65mm high. It consists of
mounted on the track side, because the two pieces; a plain aluminium U-shaped base
groundplane side of the board must be close and a durable brown epoxy paint-finished
to the heatsink. Top and bottom groundplane steel top. The two parts of the case are held
bonding is provided by short wire links together by four screws. Some problems were
through the board. encountered in finding a suitable heatsink for
Figs 8.12 shows the general layout inside the the amplifier. Eventually I decided to use a
case. A U-shaped aluminium screen separates readily obtainable heatsink which is larger
the amplifier from the rest of the transverter. than required, and to cut this down to the
Feedthrough capacitors decouple the supply right size for the intended case. I recommend
leads and ALC/VSWR detector outputs where that you get a local engineering workshop to
they enter or leave the screened amplifier do the cutting for you, unless you really enjoy
compartment. hard work with a hacksaw and file! The Siefert
Short M25 bolts and nuts secure the lower KL-139 75SW heatsink (black anodised finish)
lid of each tinplate box to the bottom of the needs to be cut down to 64 x 105mm to fit on
case. Two screws are used for each lid. The the rear panel.
linear amplifier is secured to the rear panel For a professional appearance, you need to
and heatsink by four M3 screws which pass paint the bare aluminium parts of the case. I
through the board and case, and are tapped chose Coral Beige paint, which provides a very
into the heatsink as shown. pleasing effect when lettered with black or
The recommended case for the Suffolk is the white rub-down transfers. Holes need to be
8 14
SUFFOLK 144MHz TRANSVERTER
drilled in the front and rear panels of the case masking tape to mark and drill the holes, and
to mount input and output coaxial connec- paint the panels after drilling.
tors, power switch, power connectors, indica-
tors, mounting pillars etc. To avoid scratches ASSEMBLY
on the front panels, cover the panels with Assembly of the printed-circuit boards should
Fig. 8.11.
Full-size PCB
mask for the
M57713
linear
amplifier
(topside;
underside is
fully copper-
clad)
8 15
144MHz
Linear amplifier
Transmit converter
Receive converter
Receive converter
8 16
SUFFOLK 144MHz TRANSVERTER
Fig. 8.14.
Component
layout for the
receive
converter
depend on the diameter of the capacitors to be that two small areas of the PCB at the corners
fitted. need to be removed, corresponding to the
Before marking the drilling points, bring the overlapping edges of the tinplate box. Care-
two sides of the tinplate box together but do fully file away these small areas of board. The
not solder them together at this stage. Decide PCB should now fit comfortably in the box. If
which side of the temporarily assembled box not, carefully file the edges of the PCB to
is to be the top and mark both L-shaped sides allow a comfortable fit. Now spot-solder the
to remind yourself later. Now measure down PCB into one half of the box, aligning the PCB
9mm from the top in line with P1 and P2, and groundplane to the line you scribed 17mm
mark. This is where the holes for the feed- from the top edge of the box.
through capacitors to P1 and P2 will be Jig the other half of the box into place and
drilled. Similarly, mark down from the top spot-solder the PCB groundplane to the box.
12mm in line with P3, P4 and P5. This is Check that the top and bottom covers fit
where the input and output socket holes will snugly on to the box sides. Carefully bend the
be drilled. lid edges, if necessary, to achieve a good fit.
Measure down 17mm from the top and When you are happy about the fit, carefully
scribe a line all around the inside of the box. seam-solder the PCB into place. This will
This line marks the position of the PCB require a 50W temperature-controlled solder-
groundplane when it is seam-soldered in ing iron. Anything less with inevitably lead to
place. Drill the holes in the box sides. Do not poor results and unsightly appearance. Excess
solder the box together yet. Carefully place flux can be removed with white spirit or
the PCB over the box, with the groundplane acetone. Do be careful when using either of
towards the previously marked box top. Note these chemicals keep them away from flames
8 17
144MHz
Fig. 8.15.
Component
layout for the
transmit
converter
and skin. that the coils seat properly on the board with
Solder (or screw) the sockets and feed- the correct spacing. Take care to mount all the
through capacitors to the box. coils in the bandpass filter the same way
All resistors and RF chokes in the receive round, with the ungrounded ends of the
converter are mounted flat to the board with windings closest to the board. Solder the
10mm lead spacing. Where a lead has to be grounded end of each coil to both top and
soldered to ground, remember to solder it to bottom groundplanes, having first removed a
both top and bottom of the board. Capacitors small amount of plastic from the base of each
are mounted vertically with short leads; once coil to allow access for soldering to the top
again, solder grounded leads on both sides of groundplane.
the board, taking care not to overheat or The input coil L1 should be wound carefully
fracture the body of the capacitor. The top- as shown in the component list. Do not use
coupling capacitors in both the receive and tinned copper wire; only silver-plated or
transmit converter bandpass filters are each enamelled copper wire of the correct size will
made from two 18pF ceramic capacitors give the lowest noise performance.
connected in series. These may be replaced by Care is needed when winding the toroidal
a single tubular capacitor of 09pF if you can transformers T1 and T2. For practical hints on
find one of suitable size. winding toroidal transformers, see the panel
The coils in the bandpass filters are on page 9-10. Be sure to spread the windings
mounted side by side as shown. Cut off the around the whole core, keep the interconnec-
small lugs on the sides of the coil formers, so tions short and double-check that they are
8 18
SUFFOLK 144MHz TRANSVERTER
M57713
Output
L4 P2
IC1 C10
C9
L3
C9
C1 C3 C5 L2
Input
C4 C6
P1 C2
C11
D1
L1
D2
R1
IC2 C12
C7
C8
78S09
P5 P4 P3 P6
ALC +135V +135V SWR
switched unswitched
correct. The start of transformer T1 (wire 1) wire between P1 and P2 to their respective
goes to the drain of TR1, and the finish goes feedthrough capacitors.
to the junction of R4 and C4. The junction of This completes assembly of the receive
wire 2 and wire 3 is the tap, and connects to converter. Crop all component leads short on
C5 and C6. The junction of wires 1 and 2 is the etched side of the board, and carefully re-
not used; it should be carefully positioned check all connections and component
away from the PCB groundplane. Transformer placings before applying power. Setting-up the
T2 is wound in the same way as T1, except receive converter is covered later in the
that only two (bifilar) windings are required. section on alignment. I recommend that you
Solder the mixer, voltage regulator, crystal, do this before assembling the transmit
transistors and MMIC into place, taking care converter, since the receive converter can be
to get the leads the right way round. This used to align the bandpass filters on the
applies particularly to the mixer, which can be partially-constructed transmit converter.
difficult to remove once soldered into place.
The correct orientation of the mixer is shown TRANSMIT CONVERTER
in the component layout. The name MCL Assembly of the transmit converter is similar
should appear in the corner as shown. The to the receive converter. Fig. 8.15 shows the
metal can of the crystal is grounded by two component layout. The transmit converter fits
short wires, as shown in Fig. 8.14. Solder into a similar size box to the receive converter.
lightly and very quickly to the rim of the Mark the tinplate box sides to accommodate
crystal can. the IF input, local oscillator input and trans-
Solder short lengths of tinned copper wire mit output sockets. Also mark the box for
between P3, P4 and P5 to their respective feedthrough capacitors to connect P2 and P5
socket spills. Solder short lengths of insulated to P10. All holes should be in line with their
8 19
144MHz
Heatsink
Fig. 8.17. Mounting the M57713 amplifier to the rear panel of the transverter case. Note the
U shaped metal screen, and that the PA module is mounted directly on the heatsink through
a cut-out in the rear panel
respective P pins or holes and either 9mm or place around them as shown. These screens
12mm from the box top as described for the will need to be made from off-cuts of sheet
receive converter assembly. Assemble the box since they are not available commercially.
and PCB as described for the receive converter. Each screened enclosure is 22mm x 38mm x
Next assemble the two bandpass filters, 18mm high, and two are required. Align the
exactly as in the receive converter; once again, filters as described later, before assembling the
make sure that the ungrounded ends of all the rest of the converter.
coils are closest to the board. When the filter It was found necessary to mount a few
coils and capacitors are mounted on the resistors vertically on this board as a result of
board, tinplate screens can be soldered into limited space in some areas of the PCB. IC5
28MHz in
+135 volt
supply
LED or meter
Local oscillator Display
Input
144MHz in
P3 P2 P5 P7 P3 P1 Negative
P2 ALC
output
Receive converter P9 Transmit converter P4
P1
P4 Switched
28MHz out Press to talk
receive converter P10 input
supply
P5 P6 P8
Switched
bias Low power
transmit
P5 P6 P4
P3
P2 Linear amplifier P1
144MHz out
Denotes DC connection
Fig. 8.18. Interconnection diagram for the
boards comprising the Suffolk transverter
8 20
SUFFOLK 144MHz TRANSVERTER
and IC6 can be mounted in sockets if required. have most of them in your shack already, not
only to comply with the licence requirements
POWER AMPLIFIER but also because they are invaluable if you
Fig. 8.16 shows the simple component layout. want to build your own equipment! Chapter
Start construction of the amplifier by solder- 12 contains plenty of good ideas for special-
ing short lengths of tinned wire through the ised VHF/UHF test equipment.
holes marked in the PCB. These through- You will need:
board links are marked by black dots in Fig. Multimeter
8.16, and they bond the top and bottom Sensitive absorption wavemeter covering
groundplanes together and ensure amplifier 100 to 150MHz
stability. File the blobs of solder almost flat on 50 RF power meter(s), 10mW to 20W
the reverse side of the board. Solder all 05W 50 attenuators of 6dB and 3dB
remaining components into place as shown, Two low-power 50 terminations
except for the amplifier module itself. Three 100-200mm thin coaxial leads with
When all the other components are in suitable sockets
place, connect +135V to the switched (P4) pin The receive and transmit converters are best
and check that +9V appears at the regulator aligned before they are finally fitted into the
output. Any more than 9V could damage the transverter case. If a fault is found, its much
amplifier module, so rectify any problems easier to sort it out while you have easy access
before soldering it into place. to both sides of the board.
Carefully fit the amplifier board to the
heatsink, and position the module into the RECEIVE CONVERTER
cut-out in the board. Check that it fits prop- Begin alignment with the receive converter.
erly. Now remove the module, smear a thin, Connect the oscillator and RF stage supply
uniform layer of heatsink compound on to pins together (P1 and P2). Connect the short
the mating face of its metal flange, and screw coaxial leads to the RF input, IF output and
it back to the heatsink. The amplifier module local-oscillator output connection points on
can now be soldered into place. Fig. 8.17 the PCB. Connect 50 terminations to the
shows how the amplifier is wired up behind its oscillator output and antenna sockets.
screen, and mounted through a cut-out in the Connect +135V to the converter and check
rear panel of the transverter case. that the current drawn is no more than about
130mA. If it is significantly higher, check for
WIRING UP short circuits, wrongly placed components,
Each board has its input and output connec- faulty components etc. When you are satisfied
tion points indicated by a P on both the that all is well, proceed to the alignment of
circuit and layout diagrams. Fig. 8.18 shows the converter.
how these points should be connected Tune your wavemeter to 116MHz and place
together to produce a complete transverter its pick-up coil close to L9. Rotate the core of
but please read the instructions on alignment L9 until the meter indicates oscillation at
before you wire the units together and tie 116MHz. Switch the converter off and on, and
everything down into the case! All DC check the oscillator restarts reliably. If it fails
connections should be made with plastic- to restart, turn the core slightly and recheck.
covered stranded wire. All RF interconnections Final setting of the core to set the oscillator
should be made with thin 50 coaxial cable exactly on 1160000MHz can be done later.
such as RG174. Now place the wavemeter coil close to L10
or L11 and check that the meter indicates the
ALIGNMENT
presence of a strong 116MHz signal. If you
Some basic items of test equipment are needed have a power meter capable of reading 10 to
to align the transverter properly. You should 20mW full-scale, connect this in place of the
8 21
144MHz
50 local oscillator termination and confirm converter to cover the segment of the band of
that the output is approximately 10mW; then interest. In any case, the noise figure should
replace the termination. finally be optimized at the DX end of the
Connect a receiver to the converter IF band. No preamplifier should be required for
output. Tune the receiver to 29MHz, and you terrestrial DXing unless the feeder loss is very
should detect at least a small increase in noise high in which case, change the feeder!
output when the converter is switched on. Although optimization at the LF end of the
Set the cores of L2, L3 and L4 level with the band may leave the receiver slightly deaf for
tops of their respective formers. A further satellite working between 1458 and
small increase in noise output will be noticed 1460MHz, you will probably be using a
if the filter and the RF stage are both working. preamp for this application anyway.
Slightly screw in the core of L3 by up to two
turns. The noise should show a further peak. TRANSMIT CONVERTER
Slightly retune the cores of L2 and L4 for Alignment of the transmit converter is easy,
maximum noise. especially if you align the two bandpass filters
Adjust the core of L8 to increase the noise before completing the rest of the construction.
output. This peak is very broad, and may be When the two filters have been built, connect
hard to identify for certain. short coaxial leads to the input and output of
Connect a 144MHz antenna to the con- one filter, and connect an attenuator to each
verter RF input. Tune CT1 to increase the lead. If you have access to the necessary test
noise output from the receiver. The proper equipment, this can be used to align the filters
setting for CT1 is the most difficult adjust- for a nice symmetrical response right across
ment in the whole transverter, since it needs the range 144 to 146MHz. It should be
to be set for optimum signal-to-noise ratio. possible to obtain a response flat within 1dB.
This setting does not coincide with maximum Attenuation at 116MHz should be about 50dB,
noise. CT1 is best adjusted with the aid of an and greater than 70dB at the image frequency
automatic noise-figure meter. However, not of 88MHz.
everyone has access to one, so some alterna- If you dont have access to sophisticated test
tive methods are needed. The best amateur equipment, you could align the filters using a
method is to use an alignment aid such as the 1000Hz tone-modulated AM signal generator
G4COM design mentioned on Chapter 12. If and the detector and selective AF level meter
you have one, youve probably already read described in Chapter 12. Otherwise you will
the instructions in the article, so I wont waste have to align the transmit-converter filters by
space by repeating them here. Failing that, listening to off-air signals using the receive
youll have to use a weak off-air signal. converter. Whichever way you do it, use the
Judging small differences in signal-to-noise highest possible values of 50 attenuators on
ratio by ear is very difficult, so I recommend either side of the filter. Insufficient attenua-
that you use a weak FM signal as described in tion will lead to gross errors in filter align-
Chapter 5 for setting up the gain of preampli- ment, and intermodulation in the transmit
fiers. converter.
The adjustment procedure described above Once the two filters are aligned, finish
is likely to result in the converter being peaked building the transmit converter while trying
in a relatively narrow frequency range, not to disturb the filters youve just so care-
somewhere between 144 and 146MHz. If you fully adjusted!
have access to a sweep generator, a network Connect the short coaxial test leads to the
analyser or a spectrum analyser with tracking IF input, RF output and local oscillator inputs.
generator, the bandpass filter can be adjusted Adjust RV2 so that its rotor is at the ground
to give a flat response within 1dB across the end of its travel. The setting of RV3 is not too
144-146MHz band. Otherwise, just re-peak the important at this stage.
8 22
SUFFOLK 144MHz TRANSVERTER
8 23
144MHz
for the required RF output, which should be use the external ALC control to reduce the HF
no more than 10W. drive to more acceptable levels. You can
If the ALC attack and decay times do not generally afford to do this, because the Suffolk
suit your voice characteristics you may alter is designed to accept very low levels of
them by changing the value of R16 and R17. transverter drive. If you cannot check the 28
R16 sets the decay time D according to: 30MHz drive signal with a spectrum analyser,
an output level of 10 to 15dB below the
D = C35 x R16 seconds
maximum available will normally ensure a
while R17 sets the attack time A according to: clean source signal.
If low-level spurious signals are noticeable at
A = C35 x R17 seconds
57kHz either side of the transmitted carrier,
With the values given, the attack time is these are probably due to the negative-voltage
220ms and the decay time 330ms. These are generator on the transmit-converter board.
fairly gentle times, because I have tended to Extra decoupling capacitors are probably
err on the safe side so as not to cause rough- required in addition to C33 and C34. Try a
sounding audio. However, if you do modify 100F electrolytic between pin 4 or 8 of IC6
the time constants, dont be tempted to use and adjacent ground. This should be a 16V
the ALC of the Suffolk to increase your talk working capacitor with the positive connected
power because thats not what it is intended to the IC pin.
for.
R18 sets the VSWR protection reaction time MODIFICATIONS
P as: Several modifications are possible to the basic
Suffolk design. These are easy to do, thanks to
P = C35 x R18 seconds
the modular nature of the transverter.
This time constant is best kept short, for The transmit converter was designed for use
obvious reasons, though if the value of R18 is with modern transceivers which often feature
too small it can affect ALC operation. very low-level transverter drive outputs. The
External ALC to the HF transceiver can be Suffolk will accept 2830MHz input levels
taken from P2 on the transmit converter. between 20 and +24dBm (10W to 250mW)
Almost all HF radios require a negative-going for full power output at 144MHz. Lower levels
voltage for ALC control, and the voltage can be catered for by changing IC1 (the
provided by the Suffolk should be adequate. A transmit IF amplifier) from the specified MSA
20k potentiometer may be connected 0404 to an MSA 0304, at the same time
between P2 and ground to provide an adjust- changing the bias resistor R4 to 270. This
able external ALC voltage. To set the external will extend the range of HF drive levels down
ALC, begin by adjusting the transverter as to 24dBm (4W).
previously described, with the external ALC During the time the receive converter was
disabled. Next connect the external ALC to being developed, a new MMIC was an-
the HF radio and adjust the 20k nounced. The MSA1120 (soon to be followed
potentiometer so that the 144MHz output just by the lower-cost MSA1104) appears to have
starts to fall from the previously set output been aimed specifically at mixer terminations.
power. This ensures that the transverter is not It features a gain of 125dB, a noise figure of
being overdriven, and that the ALC circuit of 35dB and an IP3 of +18dB at 28MHz, making
the HF radio is in overall charge. it a worthwhile replacement for the MSA0404
Experience has shown that many HF (IC2) in the receive converter. However, the
transmitters are capable of much more increased gain will reduce the system dynamic
transverter output drive than specified by the range unless it is compensated by increasing
manufacturer. However, this is often at the the attenuation between the BPF and mixer.
expense of spectrum cleanliness, so you can TCALC shows that an increase to 9dB of
8 24
SUFFOLK 144MHz TRANSVERTER
Fig. 8.19.
Spectrum of
the Suffolk
transverter
transmitter
output,
showing as
far as the
third
harmonic. All
harmonics
are more
than 60dB
down on 8W
output, and
all higher
harmonics
are below the
analyser
noise floor
Fig. 8.20.
Output
spectrum of
the Suffolk
transverter
showing two-
tone
intermodulation
performance
at 8W PEP
output.
Third-order
products are
more than
30dB down
(24dB down
from either
tone). Fifth-
order
products are
more than
40dB below
PEP
8 25
144MHz
8 26
SUFFOLK 144MHz TRANSVERTER
TRANSVERTER
Control 144MHz 28MHz Antenna
PTT
IF Relay
Tx Rx Rx Tx
Optional
connection
to external
linear
amplifier
careful to check at every stage of the interfac- simply need a 20dB attenuator (10W in,
ing. If your questions about interfacing are of 100mW out; see Chapter 12). If the HF
the most basic kind, you should ask yourself transmitter is capable of high power but also
whether youre really ready to build your own has an RF power-output control, it can be
transverter. turned down to 10W or less in order to drive
Perhaps the most common interfacing the transverter although you also need to
problem is how to separate the HF transmit think what would happen if you forget to turn
and receive paths. If you are fortunate, or the power down from 100W! The resistors in
chose your radio well, the manufacturer will the attenuator need to be capable of dissipat-
have done this for you, and you can follow ing almost all the available HF power, and care
Fig. 8.21. Otherwise you will need to make must be taken to ensure low leakage around
your own provisions. the attenuator and its relay switching.
Transceivers with a single HF antenna HF transceivers with independent trans-
socket are generally the most difficult to deal verter sockets are much easier to interface.
with, since the transmit and receive paths Usually a transmitter IF output of between
need to be separated, and also a large amount 10W (20dBm) and 250mW (+24dBm) is
of unwanted transmitter power needs to be provided, with the lower levels being typical
lost in an attenuator before connection to the of many modern designs, and the Suffolk is
transverter. Separating the receive and trans- designed to cope with this entire range of
mit paths requires a relay switching box with levels. Simply connect the HF radio transverter
appropriate TX/RX changeover sequencing. output to the Suffolk 2830MHz TX IF input
Even a short spike of HF power into the (P1), using a suitable length of miniature
transverter input will do it no good at all! A coaxial cable. Try not to make this lead too
changeover relay will be required to bypass long, or you may get breakthrough of signals
the attenuator on receive, and for fail-safe on the 28MHz band as a result of cable
operation that relay must be energised only leakage. A European 144MHz DX record
on receive. many years ago was later identified as a direct
The Suffolk transverter can accept a maxi- 28MHz contact between two leaky
mum of 250mW input at 2830MHz. Low- transverters!
power HF transceivers such as the Kenwood HF transceivers with transverter outputs
TS120V are fairly straightforward, and you generally require some form of switching to
8 27
144MHz
disable the main HF PA. This can be done and this description have both been thor-
through the Suffolks accessory socket; consult oughly tested by a number of volunteer
the transceiver handbook for details. constructors, to whom we should all be very
The receiver connection should preferably grateful. As a result, I hope that the Suffolk
be via a separate transverter receiver socket. If design will be a success with home construc-
the receive transverter input is via the same tors.
transverter socket as on transmit, you will As a result of comments from a number of
need to provide independent TX/RX switch- constructors, the two converter boards have
ing, which is fairly straightforward. But if the been redesigned to allow them to be mounted
receiver input is via the main HF antenna in screened tinplate boxes. The opportunity
socket, you really ought to fit a relay box or has also been taken to clean up the layout and
modify the HF transceiver. Otherwise, sooner correct a few minor errors. At the time of
or later, you are going to stuff 100W of HF writing the receive and transmit converter
power into the receiver output of the Suffolk. boards are issue C. The linear amplifier board
It is often quite easy to identify a redundant is still issue A.
socket on the back of the HF transceiver (the For anyone who has never used a well-
phone-patch output, for example) which can designed transverter or transceiver at VHF
be connected by coaxial cable directly to the before, the Suffolk can take a little getting used
receiver input, bypassing the antenna to. There is no excess gain, so dont expect
changeover relay. your HF transceivers S-meter to hover at S9
A more sophisticated arrangement catering on noise. If you are easily impressed by lots of
for several transverters connected to the same noise, youre probably not getting the best out
HF radio would use a switching box to of your DX system. Youd do well to re-read
connect the TX and RX IF to the wanted Chapters 4 and 5, and get used to the fact that
transverter via miniature screened relays or a weak signals can be perfectly copiable without
rotary selector switch. The switching box also moving the S-meter.
needs to disable the HF PA, and to route the
PTT and ALC lines to the appropriate trans- REFERENCES
verter. [1] The diplexer circuit was devised by H. Paul
If you can, arrange the switching so that all Shuch, N6TX. It first appeared in Ham Radio,
the transverters can remain powered while on February 1977 and is used by kind permission
standby, so that their local oscillators are of the author and magazine editor.
already stabilized on frequency. A problem [2] The broadband Class-A negative-feedback
can arise here, due to strong-signal pickup in amplifier used in both the receive converter
one of the unused transverters feeding and transmit converter appeared in Chapter 8
through the limited isolation of the IF switch of Solid State Design for the Radio Amateur by
and being heard on the HF receiver. The Doug DeMaw W1FB and Wes Hayward
solution in the case of the Suffolk is to switch W7ZOI, and is used by kind permission of the
off the receive converter during standby while publishers, the American Radio Relay League.
retaining power to the local oscillator. Sepa-
rate supply connections and unused pins on
the accessory socket allow for this option,
which I have implemented in my own station.
SUMMARY
I have presented a detailed description of a 28
to 144MHz transverter which is easy to build,
but has the high performance necessary for
todays crowded band conditions. The design
8 28
SUFFOLK 144MHz TRANSVERTER
8 29
144MHz
8 30
THE W1SL 144MHz POWER
AMPLIFIER
by John Nelson GW4FRX
8 31
144MHz
VG1 VG2
FT FT
VDR
RF C
OUTPUT
L2a
INPUT
Cn L3
C4 TUNE
L1
C2 C3
L4
Cn
C1 L2b
LOAD
C5 RFC1
VDR
RFC
C7
FT FT FT FT
C6
Fig. 8.22. Circuit of the W1SL power amplifier. See Fig. 11.5 for additional screen-grid
protection components
you try to wind the power up. People have resistors and inductors as shown in Fig. 11.5.
tried all sorts of clever ways to get round this The W1SL circuit shown in Fig. 8.22 includes
loading problem, with Pawsey stubs and all only the VDRs for clarity, but no doubt you
sorts of fearsome devices being pressed into can reason the rest out for yourself.
service. In fact, the original design seems to
work perfectly well provided that you enlarge CONSTRUCTION
the coupling loop considerably, as shown in The amplifier is built in two 17" x 8" enclo-
the diagrams and construction notes which sures, following the general layout of Figs 8.23
follow. and 8.24. The anode box is 3" deep; although
As with almost every 4CX amplifier design the original grid box was only 3" deep, you
in the history of the world, the original W1SL need closer to 4" depth to accept the mouth of
included a low-value series resistor in each a decent-size blower as recommended in
screen feed in this case 150. For Class-C Chapter 6. The assembly is completed by top
work, screen regulation doesnt matter so and bottom covers and a 19" rack panel,
much and the resistors would help to isolate ideally 5U or 6U tall. All this is standard
each screen from the common power supply. metal-bashing and you should be able to
But if you include these resistors in a modern make that, or get it made locally, without too
W1SL intended for Class-AB1 use, you in- much bother. If youre a dab hand in the
stantly negate the effort which you put into workshop you could make a W1SL chassis
building a screen supply which could source entirely from sheet, bar and angle stock.
and sink vast amounts of current with practi- For the non-specialist metal-basher, the
cally zero variation in output voltage. Such anode line is probably going to be the most
resistors are also bad news if you have a difficult part of the whole amplifier to make.
flashover. If 150 is the first thing the current Dont let that stop you, though my first two
surge sees on its way out, the voltage on the 144MHz amplifiers were W1SLs, and in both
screen grid will streak up and the screen cases the nearest I came to workshop facili-
decoupling capacitor will blow faster than you ties was possession of one junior hacksaw and
can say tetrode. Id recommend that you some extremely old and blunt files. The
protect each screen using VDRs, parallel workshop was the floor! All very character-
8 32
W1SL 144MHz POWER AMPLIFIER
Fig. 8.23. Top view of a partly-constructed W1SL amplifier. Still to be added are the front
panel, anode-tuning capacitor, output coupling and top cover
Fig. 8.24. Underside view of a partly-constructed W1SL amplifier, minus its DC wiring
8 33
144MHz
Fig. 8.25.
Leading
mechanical
dimensions
of the anode
line and
tuning
arrangement
(dimensions
in inches)
building, of course, and another argument to the anode line, the grid lines, the coupling
defeat those who moan about the impossibil- loops and the tuning and loading capacitors
ity of home-brewing anything, let alone big follow in sequence towards the left-hand end
amplifiers. If you or yours draw the line at of the chassis. Fig. 8.25 shows the anode line
metal filings in the carpet, you should still be and the vanes which form the stators of the
able to find someone who can run you up a anode tuning capacitor. The line is supported
set of W1SL metalwork in the lunch hour. on stand-off insulators as shown, and is
Think of it as an initiative test in the true connected to the anodes of the valves by tabs
amateur spirit. and ring clamps. Elongate the mounting holes
The layout is quite uncritical. Having drawn in the anode line to allow the valve clamps to
a centre-line along the top of the lower be drawn up firmly before tightening the line
chassis, mark the centres for the two valve down on to the stand-offs.
bases 2" from the right-hand edge, and 2" The flapper which forms the rotor of the
apart. Drill pilot holes and use these to mark tuning capacitor is a box-shaped arrangement
the top cover. All the other major components mounted on a quarter-inch fibreglass rod (Fig.
8.25). Vertical fins on the anode lines provide
Fig. 8.26. extra capacitance to the ends of the flapper.
Output The rod is supported by a slow-motion drive
coupling loop
on the front panel and a bearing at the rear of
for the W1SL
amplifier the anode enclosure. Its best to mark the
(dimensions positions of the two holes for this rod after
in inches) mounting the anode line.
The output coupling loop is shown sepa-
rately in Fig. 8.26. It starts half-way between
the U-shaped closed end of the anode line
and the tuning flapper, follows the centre-line
of the anode line around the U (dodging past
the stand-off insulator) and exits at the
opposite half-way point. In other words, the
8 34
W1SL 144MHz POWER AMPLIFIER
8 35
144MHz
which connect to the valves are bent apart, The DC connections to the grid lines are
and each is flattened and drilled to accept a made at the points of minimum RF voltage,
screw (Fig. 8.24). The opposite ends are about half-way along. If you want to make a
soldered to C2 and C3. The original W1SL was proper job of it, temporarily attach the grid
presumably designed with the American 144 feeds at roughly the right place, tune the lines
148MHz band in mind, whereas the UK and to resonance, and slide an insulated screw-
Europe have only 144146MHz. Because of driver along one line to find the point of
this, and also because of the fact that youll minimum disturbance. Then move the DC
probably be using the amplifier right at the grid connections permanently to that location
bottom end of the band for most of the time, on both lines. This is not at all critical, so long
Id recommend that you lengthen the grid as both sides are the same. Dont spend too
lines from the original W1SL dimensions by at long on the job, though, because not much
least ". If you use a Jackson U102 25+25pF for air is going to go through the valves while
the grid tuning capacitor and make the grid youve got the bottom off the grid compart-
lines 11" long before bending, youll certainly ment try and jury-rig something to ensure
be OK. 144MHz should tune at slightly more that they get at least a whiff of air over the
than half mesh, and 146MHz at slightly less seals and grid spigots whilst youre twiddling.
just what the doctor ordered. Because modern SK610 and SK620 bases
From this latter point you will gather that have better isolation than those which were
the grid tuning of a W1SL is exceedingly around in 1971, you will probably need a
sharp. The original write-up didnt mention a much smaller neutralizing capacitance than
slow-motion drive and there wasnt one in the was shown in the original article. Try just " of
pictures, so one perforce has to assume that the braid from some UR43 coax, soldered to
Messrs McMullen and Tilton both had the tops of the feed-through insulators which
muscular control and reflexes about a million come out underneath the anode line (Figs 6.8
times better than mine. Unless youre either a and 6.9). Chapter 6 explains how to adjust
Tornado pilot or a concert harpsichordist, fit a their position to neutralize the amplifier.
10:1 drive to the grid tuning capacitor; you Incidentally, the wiring from the various
should then find the grid tuning just about electrodes of the valves to the multiway
manageable. The anode tuning is quite sharp connector was done in UR43-type coax in the
as well, and a slow-motion drive also helps to original. If you enjoy a challenge and dont
lock the mechanically unbalanced flapper in mind spending hours and hours getting the
place. Dont forget that the bearing for the bending radii right, you can do a supremely
grid tuning shaft must be airtight since the beautiful job by using semi-rigid coax
grid compartment is pressurized, so choose a especially if you polish the outer screen with
drive such as the fully enclosed Jackson 10:1 Dura-Glit or Brasso.
job which doesnt leak. With a few demon tweaks along these lines,
You could use a similar component for the you should find that the W1SL is supremely
input loading control as well, for the same easy to get going and a joy to use. If your
reason. blower is big enough, you shouldnt have to
The input coupling loop isnt at all critical touch the anode or grid tuning for months on
in the W1SL. You should easily achieve a 1:1 end and once youve set the input VSWR
VSWR with the arrangement shown in Figs and balance you can forget about them until
8.23 and 8.27. The coupling loop is about 2" you have to change the valves. Incidentally,
long and " wide, and sits about " below the Ive also used the W1SL design as a basis for
grid lines, half-way along the straight part. several amplifiers using 4CX600JAs and
The original design used copper strip, but here 4CX1500As, and they all worked a treat.
again 14SWG enamelled copper wire or
thereabouts works perfectly well.
8 36
ANTENNAS FOR
144MHz DX
Fig. 8.28. The 10-element DL6WU Yagi for 144MHz. Above it are two other DL6WU Yagis, a
23-element for 432MHz and a 49-element for 13GHz. All three antennas are fed with folded
dipoles and half-wave coaxial baluns
8 37
144MHz
8 38
AN T E N N A S F O R 1 4 4 M H z D X
Fig. 8.29. E-plane pattern of the 10-element DL6WU/PA0MS of around 6m, and is intended for really
Yagi, calculated by the NEC program. The front/back ratio serious 144MHz DXers and contesters who
should be well over 20dB, and the first sidelobes at about have the masts, towers and rotators to handle
18dB relative to the main lobe such a monster. With a gain of about 136dB,
even one of these Yagis makes a highly
Fig. 8.30. Two different mounting methods for elements on a effective tropo antenna, and four will put you
25mm (or 1") square boom
in the moonbounce league. Although it is
actually quite an old design, the table on page
7-31 shows that this Yagi still out-performs
many others of a similar boom length.
This Yagi departs from DL6WUs general
design philosophy that one reflector should
be sufficient, by having dual reflectors above
and below the plane of the rest of the ele-
ments. The overall boom length is chosen to
coincide with a natural high front/back ratio,
and at the optimum frequency this can be in
excess of 30dB. Fig. 8.31 shows the approxi-
mate E-plane pattern. According to the
guidelines in Chapter 7, suitable stacking
distances for horizontal polarization would be
37m vertically and 395m horizontally. The
8 39
144MHz
Fig. 8.31. E-plane pattern of the 15-element DL6WU Yagi, calculated by the NEC program. The front/back
ratio should be about 30dB, and the first sidelobes should be about 18dB relative to the main lobe
8 40
CHAPTER 9
50 & 70MHz
9 1
50 & 70MHz
A HIGH-PERFORMANCE
50MHz TRANSVERTER
by Dave Powis G4HUP
9 2
50MHz TRANSVERTER
Fig. 9.1. Block diagram of the G4HUP 50MHz transverter, showing division into PC boards
noise from a resistive load, with due attention strong-signal handling. A noise temperature of
to accurate impedance matching. The lowest 600K equates to a noise figure of about 5dB at
antenna noise temperature measured was the input of the receive converter (the loss
650K, and 1000K was more typical. This is and noise contribution of the feeder can be
lower than would be expected from Chapter ignored in this context).
4, but I would not expect to find many quieter Turning to the HF receiver, current rigs have
locations than mine at 50MHz; suburban or typical noise figures of about 10dB at 28MHz,
more exposed rural environments would be with an input IP3 of about +20dBm. Working
much noisier. Although I have used 600K as into such a receiver, the 5028MHz receive
the target for the system noise temperature, converter therefore has to achieve an overall
you are free to optimize the receiver sensitiv- NF of 5dB and a system IP3 of 0dB as already
ity for noisier environments, and this will specified. Thus the receiver block diagram
bring corresponding improvements in the begins to take shape.
Fig. 9.2. TCALC analysis of the 50MHz receive converter and HF receiver
9 3
50 & 70MHz
An overall block diagram of the 50MHz therefore to be avoided wherever a low noise
transverter system is given in Fig. 9.1. As well figure is important, this was something of a
as showing the major blocks, it also indicates shock. However, thorough investigation failed
the way I have partitioned the design into a to turn up any problems with the measure-
number of modules, each on its own PC ment technique, and certainly the perform-
board. You can build whichever units you ance of the receiver seems to agree with the
need: a receive converter, a complete low-level calculations using this value of noise figure.
transverter, or even a stand-alone 10W PA. The output load of the BF981 is another
toroidal transformer L3 (by now I was thor-
RECEIVE CONVERTER oughly converted) which is connected to the
AND LOCAL OSCILLATOR capacitive input matching of the filter. Two
The design of the receive section of the identical two-pole sections are used. Within
transverter was carried out using the concepts each section, a common inductance forms the
explained in Chapters 4 and 5, and the coupling element; this configuration is not
TCALC analysis is shown in Fig. 9.2. The often seen in amateur filter designs, but at
paper design was backed up by many these frequencies it is easier to control than
measurements on individual stages, to make the more usual top-coupling capacitance. It is
sure that the expected performance can be sometimes possible to print the bottom-
achieved in practice. coupling inductors on the PCB, but the
Working through the signal path of the necessary inductances would have taken too
receive converter (Fig. 9.3), the input circuit much space so I settled for single-turn loops
for the BF981 RF stage (TR1) uses a toroidal for L6 and L9. The insertion loss of the filter
transformer. A lot of time was spent investi- varies between 5 and 6dB over the frequency
gating low-noise matching into the BF981, range 495 to 525MHz.
using different arrangements of tuned circuit A 4dB resistive attenuator is used to get rid
aimed at minimising losses, but the noise of excess gain before the mixer. It also serves
figures measured were far from satisfactory. In to define the impedance into the mixer
desperation, and against my intuition, I tried a correctly, since the return loss of the filter is
toroidal transformer and the first sample only around 15dB over the bandwidth quoted
gave a noise figure better than 04dB at above. If you wish to adjust the overall gain of
50MHz! Having been educated to regard the receive converter, to optimize the gain to
magnetic cores as inherently lossy, and your particular RF noise environment, do it by
Fig. 9.3.
50MHz
receive
converter
circuit
9 4
50MHz TRANSVERTER
Fig. 9.4.
22MHz local
oscillator for
the 50MHz
transverter
(on same
board as
receive
converter)
altering the value of this attenuator. This will against supply-voltage variations. The ampli-
ensure that the strong-signal handling fier following the oscillator is a broadband
attributes of the receive converter are not stage with 16dB of gain. A simple two-pole
compromised. The mixer is an SRA-1 diode filter reduces all unwanted signals to better
ring (or the less expensive SBL-1), driven at an than 50dBc, and a resistive power divider is
LO level of +10dBm for greater dynamic used to provide a separate +10dBm output for
range. This is followed by an MSA0404 MMIC the transmit converter.
as an active terminator, and then by a diplexer
to provide a broadband load to the MMIC TRANSMIT CONVERTER
see the Suffolk in Chapter 8 for further details. AND CONTROL SWITCHING
The local oscillator is on the same board as The circuit of the transmit converter is shown
the receive converter, and its circuit diagram is in Fig. 9.5. RV1 gives some preset control over
shown in Fig. 9.4. The oscillator is a Colpitts the input signal level, and the PIN diode
type using a bipolar transistor, and a voltage attenuator (D1) forms the control element of
regulator (IC2) is used to maintain immunity the ALC loop. The ALC range is in excess of
Fig. 9.5. 28 to
50MHz
transmit
mixer circuit
9 5
50 & 70MHz
Fig. 9.6. 50MHz driver amplifier and TX/RX switching (on same board as transmit converter)
30dB, and is more than enough to protect the tion products better than 40dBc. The 44MHz
power stages from overdrive or damage. An second harmonic of the LO is better than
MMIC (IC1) is then used to bring the signal 55dBc, and the 100MHz second harmonic of
level up to around 15dBm to drive the mixer, the wanted signal is better than 65dBc.
which is another SRA-1 or SBL-1. Although it Indeed, the transmit converter on its own
is possible to drive the mixer harder, the IMD would provide an excellent drive source for a
performance deteriorates. The following second transverter to 13GHz or above.
terminator, diplexer and filter are similar to Transmit-receive switching is mainly
those in the receive converter, though the performed electronically, using PNP power
diplexer has been simplified because the filter transistors (TR4 and TR5) as the switching
itself provides a reasonable termination over elements. These are driven by two NPN
the wanted band of frequencies. devices (TR6 and TR7) arranged as a long-
Two broadband Class-A transistor stages are tailed pair, to ensure that transmitter and the
used next (Fig. 9.6). The first has a BFR96 receiver cannot both be on at the same time.
(TR2), running at a collector current of 40mA,
and the second a BFG34 (TR3) at an IC of PA MODULE
100mA. These relatively high currents are The availability of the Mitsubishi M57735
necessary to maintain the IMD performance linear RF power module in the UK makes
to specification. The use of feedback in these possible the simple construction of a reliable
stages enables the gain to be controlled, and and robust PA. I have inadvertently operated
in fact would allow the design to function at mine into an open circuit with no damage
almost any frequency from around 10MHz up although I dont recommend this as a regular
to about 180MHz, simply by changing the operating practice! The major drawback is the
driving frequencies and the filters. The high price in the UK, but a few mistakes with
addition of the trapped low-pass filter ensures power devices in a discrete-transistor PA
a very clean output, with no unwanted signals would soon add up to the same overall cost.
greater than 55dBc for a 15dBm input signal The other drawback of these modules is that
at 28MHz. Under these conditions the output their IMD performance is not outstandingly
is 200mW PEP with third-order intermodula- good, as explained in Chapter 8.
9 6
50MHz TRANSVERTER
Fig. 9.7.
Vcc
50MHz
modular IN OUT
L1
power
TxB COM
C4
amplifier C3 C5 C6
IC1
circuit
C1
C2
3 2 4
C7
5
IC2
1
RF
RF R2
o/p
i/p
R1 R3
Fig. 9.7 shows the circuit diagram for the PA, choke is used to make the DC connection
including the bias control circuit. Switching between the two. The DC bias input is
between transmit and receive is by applying similarly decoupled.
the bias; the supply to the collectors of the
transistors is permanently connected. The TXB VSWR AND LOWPASS
(switched 12135V transmit) line used to FILTER BOARD
power the transmit converter also switches the This unit (Fig. 9.8) contains the VSWR detector,
supply to the 78S09 bias regulator IC1. the main antenna changeover relay and the
Looking at the RF path, provision is made lowpass filter. The VSWR bridge provides
for a resistive attenuator at the input to the forward and reflected DC outputs simultane-
module, giving a further adjustment point to ously, and has been tested up to 100W of RF in
ensure optimum performance. The output this configuration. The values of resistors R3
signal has a DC blocking capacitor, and then and R4 are specified to give a suitable signal to
will be cabled off to the VSWR and lowpass the ALC amplifier at the 10W output of this
filter board. The DC supplies to the two transverter. A CX-120P PCB-mounted coaxial
transistors inside the PA module are brought relay is used as for antenna changeover; it is
out to separate pins (2 and 4) and each must driven by the PTT line.
be adequately decoupled at audio and RF. A The lowpass filter is an implementation of a
Fig. 9.8.
VSWR and
lowpass filter
circuit for the
50MHz
transverter
9 7
50 & 70MHz
design by G4CXT. Like the VSWR bridge, the bridge, the ALC input on the transmit con-
LPF has been tested up to 100W RF and can be verter board, and the HF driver. Fig. 9.9 shows
used in other designs. The specified ceramic- the circuit diagram. The forward and reverse
plate capacitors are satisfactory for use with the VSWR detector outputs are the inputs to the
10W PA module, but silver-mica capacitors op-amp IC1, and its feedback resistor is
must be used at the 100W level. The insertion adjustable from the front panel to provide a
loss of the LPF on receive is about 02dB. power-output control. The output signal from
The entire unit is mounted inside a tin-plate IC1 goes to the PIN diode attenuator in the
box to maintain the integrity of the filtering. transmit converter, and also to a rear-panel
connector to provide ALC control to the HF
ALC AMPLIFIER BOARD rig. IC2 and its associated components form a
The function of this board is to provide the standard circuit to generate a negative supply
interfacing between the output of the VSWR voltage for IC1.
Fig. 9.10.
Front view of
the G4HUP
50MHz
transverter
9 8
50MHz TRANSVERTER
Fig. 9.11.
Inside view of
a prototype
G4HUP
50MHz
transverter.
The modular
PA is slightly
different
from that
described in
the text.
Fig. 9.12.
Rear view of
a prototype
G4HUP
50MHz
transverter.
The
recommended
heatsink is
larger than
the one
shown here.
9 9
50 & 70MHz
board in its tin-plate screening box on one choice is yours. As shown in the rear view
side, and the ALC amplifier board on the (Fig. 9.12), I used an N-type for the antenna
other. (Note that the transverter in the connector, BNCs for the IF connections, a
photographs was a prototype with slightly multi-way D-type for the PTT and ALC, and
different circuits and board layouts from those 4mm sockets for the power. Within the
given here.) transverter I used miniature SMB connectors
Connectors are another area where the for coax. All of these fit in with my own shack
9 10
50MHz TRANSVERTER
standards, and I expect youll want to do the Alignment is a fairly painless process,
same for yourself. thanks to the use of standard coils wherever
There are no real problems in construction possible. It is obviously easier if you have access
of the boards. The only point of difficulty may to some test equipment, though you can get by
be in winding the toroidal transformers the with just a multimeter and some means of
panel opposite gives a few hints. measuring RF power at levels down to about
Fig. 9.13
Interconnection
diagram for
the G4HUP
50MHz
transverter
9 11
50 & 70MHz
10mW, such as the simple milliwattmeter 9.3 and 9.4. Fig. 9.14 shows the PCB mask for
described in Chapter 12. the underside of the board. The top of the
Basic DC checks must be carried out first. board is an un-etched copper groundplane,
Connect each board in turn to a 125-135V and no top-side mask is required; simply
regulated supply (following the connection remove the copper from around the non-
information given in Fig. 9.13) and check for grounded holes using a 3-4mm drill held in
correct DC conditions before proceeding on to your fingers. There is also a partial ground-
the RF alignment. plane on the underside of this board, so some
ground connections need to be soldered on
RECEIVE CONVERTER AND LO both sides. Fig. 9.15 shows the component
The circuits for this board were shown in Figs layout. All components are generally available,
Fig. 9.14. Full-size PCB mask for the 50MHz receive converter and LO board (underside view;
top side is un-etched copper groundplane)
Fig. 9.15. Component layout for the 50MHz receive converter and LO board
9 12
50MHz TRANSVERTER
and suggested sources are given in the compo- coils (L4, L8) level with the tops of the cans,
nents list. and those of the inner pair (L5, L7) a half-turn
Build the LO section first, including the further out. Just make sure that the compo-
amplifier and filter. You will have to remove nents you have used are exactly as specified.
the side lugs from the two S18 coils for the In practice, the benefits obtained by more
filter, so that they can be mounted close accurate adjustment of the filter are minimal.
together on the PCB. The side of the coil with If you have access to the necessary test
the lower shoulder is the ungrounded end. If equipment and cant resist the urge to tweak,
you have a receiver which covers 22MHz you go ahead and enjoy optimizing all four
can check that the oscillator is functioning. interacting adjustments.
Use a milliwattmeter to check the local
oscillator output level and peak the LO filter TRANSMIT CONVERTER
(start off with the two cores just proud of the AND CONTROL
top of the plastic coil body). You should find Figs 9.5 and 9.6 showed the circuit for this
that the LO output to the transmit converter board. The style of construction is very similar
is +10dBm, and if this is correct ( about 1dB to that of the receive converter; Fig. 9.16
max) it is reasonable to assume that the RX shows the PCB mask and Fig. 9.17 the compo-
LO drive is also correct. If you have a fre- nent layout. Begin construction by mounting
quency counter, adjust L13 to give exactly the components associated with the TX/RX
22000MHz; if not, youll have to wait until switching and the ALC circuit, as these are the
you can receive a beacon transmission. most densely packed. Then work logically
Now build the rest of the receive converter, through the mixer, filter and amplifiers to the
taking care to insert TR1, the mixer and IC1 output, taking care to insert the mixer, ICs
the correct way round. The two coupling coils, and transistors the correct way round. The
L6 and L9, are each 20mm of wire in a U- two loop coupling coils in the filter (L6 and
shaped loop. L9) are identical to those in the receive filter.
Connect your HF receiver to the IF output As in all construction projects, ensure there
and a 50MHz signal generator to the input of are no short circuits, particularly around the
the board, with the TX LO output terminated TX/RX switch since some of the components
in 50. Set all the cores of the filter to be just there are very close to one another. Connect
proud of the can tops, and the cores of the +12513V and ground and check the DC
mixer termination coils also just proud of the voltages. Check the action of the TX/RX
coil formers. Set the signal generator to switch in response to a DC ground applied to
501MHz, CW, and tune the receiver until you the PTT pin of the board.
find a signal at about 281MHz. Having heard The RF alignment is quite simple. Set the
your first signal, peak it using C1. If no signal cores of the diplexer just proud of the top of
appears, even at the highest setting of the the coil body, and set the filter cores as
signal generator, do not adjust the filter cores detailed for the receiver. Set the core of the
but look instead for a fault. Once you have output trap filter just below the top of the coil
located the signal and peaked C1, you can body. To confirm the operation of the trans-
then attempt to set up the filter. mit converter, connect a source of 28MHz
Since the transmit and receive filters are drive (up to about 200mW) to the input with
identical, the same procedure can be used to RV1 set to maximum resistance. Connect the
align them both, although in practice they LO input from the receive converter, and a
hardly need alignment at all. A number of suitable load to the output.
filters have been built to this design; in all Adjust RV1 to obtain a suitable output
cases an entirely acceptable performance indication at 50MHz. Using a diode probe or
between 50MHz and 52MHz has been ob- milliwattmeter, you should be able to register
tained by setting the cores of the outer pair of at least 100mW at the output of the transmit
9 13
50 & 70MHz
converter board. Prototypes have typically burrs round any of the holes, and do not
given 200 to 250mW output at 50MHz. overtighten the bolts or tighten them un-
evenly. Try out the assembly before you solder
POWER AMPLIFIER any components to the board, to make sure
The power-amplifier board is designed so that that everything fits all right without stressing
the entire assembly of components, rear panel the leads of the module. Two ways to destroy
and heatsink is held together by the screws these modules are poor assembly practices and
which hold the board in place, along with inadequate heatsinking, so be careful.
those securing IC1 and IC2. In the prototype, Fig. 9.7 showed the very simple circuit
the heatsink was drilled and tapped for M3 diagram all the complication is inside the
bolts using the PC board as a template. The module. Figure 9.18 shows the PCB mask for
same heatsink can be used as in the other the power amplifier board, and Fig. 9.19 the
transverters in Chapters 8 and 10, but I chose component layout. Note that all components
not to cut down the heatsink to fit behind a are mounted on the top side of the board,
low-profile case. Make sure that there are no which is single-sided and effectively uses the
Fig. 9.16. PCB mask for the 50MHz transmit converter, driver and switching board (underside
view; top side is un-etched copper groundplane)
Fig. 9.17. Component layout for the 50MHz transmit converter, driver and switching board
9 14
50MHz TRANSVERTER
9 15
50 & 70MHz
The diameters of the holes depend on the testing to be done in this module, except to
components available, eg the group of four confirm the operation of the relay.
holes on the side of the box are for the
feedthrough capacitors. ALC AMPLIFIER
Provided that the coil dimensions have Fig. 9.9 showed the circuit diagram for the
been followed closely there is no alignment or ALC amplifier board, and the PCB mask and
component placement are shown in Figs 9.23
and 9.24. There is little that needs comment
Fig. 9.22. Drilling diagram for the case
here, except to note the polarities of the
housing the 50MHz VSWR-LPF board
electrolytic capacitors and the diodes of the
negative-rail generator.
The panel-mounted OUTPUT LEVEL control
shown on the diagrams and drawings is in fact
two potentiometers RV1 mounted on the
panel and the preset RV2 mounted piggy-
back fashion on to the tags of RV1 (Fig. 9.13).
This allows the transverter to be set up so that
maximum on RV1 gives the full output from
the unit without running into saturation.
These two components should be wired so
that minimum resistance occurs at the
clockwise position, corresponding to maxi-
mum output.
To test the ALC board, simply confirm that
the negative voltage generator IC2 is giving
around 115V, measured at pin 4 of IC1.
OVERALL TRANSMITTER
ALIGNMENT
Assuming that the PA is working correctly, the
only aspect of the transmitter requiring
further alignment is the setting up of the ALC
levels. To do this, it is necessary to disable the
ALC loop by setting RV2 of the transmit
9 16
50MHz TRANSVERTER
9 17
50 & 70MHz
POWER AMPLIFIER
RESISTORS
All 025W 2%
18R 1 R2
300R 2 R1, 3
ALC AMPLIFIER
CAPACITORS
RESISTORS Sub-miniature ceramic plate, Philips 682 and 629 series
All 025W 2% or 5% 10n 4 C2, 4, 6, 7
10K 3 R4, 5, 6 Solid aluminium electrolytic
22K 1 R2 33/16V 3 C1, 3, 5
220K 1 R1
330K 1 R3 SEMICONDUCTORS
78S09 1 IC1
CAPACITORS M57735 1 IC2
Sub-miniature ceramic plate, Philips 682 and 629 series
10n 2 C5, 6 INDUCTOR
L1 10 turns 4mm diameter, 04mm
Tantalum enamelled wire, close wound
22/16V 1 C1
10/16V 3 C2, 3, 4 MISCELLANEOUS
Heatsink Siefert KL-139 75SW, black anodized
SEMICONDUCTORS finish (Schaffner EMC, Micromark or
CA3140 1 IC1 GCA Electronics)
555 1 IC2 Solder tags and braid
1N4148 2 D1, 2 Bolts and washers
9 18
50MHz TRANSVERTER
9 19
50 & 70MHz
THE CRAY
HIGH-PERFORMANCE
70MHz TRANSVERTER
by Dave Robinson G4FRE/WG3I
AMPLIFIER PCB IF
RF BANDPASS RF AMP &
STAGE FILTER MIXER DIPLEXER
OUT
TRANSVERTER PCB
XTAL
ANTENNA ANTENNA AMPS SPLITTER
OSC
RELAY
IF
LOW-PASS ALC POWER AMP & MODAMPS DIODE
TX MIXER
FILTER DETECTOR AMP FILTER & FILTERS ATTENUATOR
IN
9 20
CRAY 70MHz TRANSVERTER
+12V RX
C3 R4
C 32
C13
T1
R3
R8
C2 R2
TR1 L5
RF C2 C15
C22 C26
IC1 +12V
R16 C29
C23 C25 C27
R12
R22
R13 L8
RFC4 R21
C20 R23
R19 LO to TX
L7
C28 CONVERTER
TR2 C24
R20
C19
TR3
TR4
C18
C21
C30 C31
RFC3 X1
R18
C17 R14
R15 R17
R11
Gnd
Fig. 9.26. Circuit of the 70MHz receive converter and local oscillator
9 21
50 & 70MHz
RV3
GAIN
C34
R26
7 FWD
6
2
IC5 R27
ALC Output REV
RV2 4 3
ALC SET
C33
C35
R24 R25
C45
R30 C43
R33
TR5
C42
R31
RFC7
RFC5
28 IN
FROM C40 D4 3/4 C44 C47 C48 C50 C51
C41 1
SK3
RV1 IC3
M2
ATT L11
R32 RFC6 L10
2/5/6/7 8 C46 L9
C49
Local Oscillator
the mixers to improve the intercept point. load is a toroidal transformer, whose con-
A dual-gate MOSFET is used in the RF tained magnetic field isolates the input and
amplifier stage to obtain high dynamic range output of the FET and helps avoid instability.
and low noise figure. Although it is capable of The output from the RF stage passes through a
sub-1dB noise figures on 70MHz, in this 3-pole bandpass filter and a 13dB attenuator,
design a compromise is made between noise to the mixer. The resistor values used in the
figure and gain to obtain the best strong- attenuator may be adjusted to provide opti-
signal handling performance. The input mum performance in your particular situa-
circuit of the RF stage is an air-spaced coil in tion. For further details see Chapters 4 and 5.
order to obtain a reasonable Q. The output The mixer M1 is an SBL-1, which was chosen
9 22
CRAY 70MHz TRANSVERTER
+12V
D1 RLA
A1
R28 4 8
D2 ETT
7 C72
IC6 3
R29
2 C38 +12V
6 RX
5 1
C36
D3
C37
Gnd
T2 RFC9
R34 R37
R42
L15 TX
C62 C67
C61
OUT
RFC8 R36 R43 C66
Fig. 9.27. Circuit of the 70MHz transmit converter and ALC generator
for its high performance-to-cost ratio. If the MMIC over a wide frequency range. This
more expensive SRA-1 is available, this may be arrangement ensures that the receive convert-
used as a direct replacement. An MSA0404 ers overall performance is not affected if the
MMIC is used to ensure a good wideband 50 input impedance of the HF receiver is not 50.
termination for the mixer, and is provided with Fig. 9.27 shows the circuit diagram of the
a bias current of 40mA. The output of this transmit converter, switching and ALC
MMIC feeds the receiver output socket (SK2) generator. The transmit converter requires a
via a diplexer. The diplexer has a broad 28MHz drive signal at around +7dBm, which
frequency response, which gives little selectiv- is fed in through the PIN diode D4. The
ity but ensures a good termination to the function of this component is to act as a
9 23
50 & 70MHz
+12V
RFC100
TX
IC100 C106
C120 C111 C110 +12V
C107
C108 C121 R107
C112
R106
C104 RFC103
+12V
C109 TX
R102 C105
C119
C114
D103 RLB
RF in L100
L101
CT102 B2
TR100
CT101
C100 CT103 C117 C118
CT100 C102 R104 R105
C103
GND
C122 C123
VF VR
Fig. 9.28. Circuit of the 10W PEP 70MHz power amplifier. All component numbers start at 100
voltage-controlled attenuator. This is followed diode D5 in the 28MHz feed to the transmit
by M2, another SBL-1 (or SRA-1) mixer, which mixer. This ALC provision reduces the risk of
is also provided with +10dBm of local oscilla- overdriving the transverter. IC6 (an NE555) is
tor drive at 42MHz. This mixer too is termi- used to generate a negative voltage rail for the
nated in a wideband manner by an MMIC, an operational amplifier.
MSA0404 biased at 40mA. The output is The circuit diagram of the power amplifier,
filtered by a three-pole bandpass filter and is filter and ALC generator are shown in Fig.
then amplified by a further MMIC, an 9.28. To improve its linearity and reliability
MSA0304 biased to 20mA. Following a further the power amplifier uses an under-run TP2320
three-pole filter is a two-stage wideband or 2N6083, either of which is rated at well
feedback amplifier using a BFR91A and a over 10W output power. IC100, D100 and
BFR96S. Discrete amplifiers are used from this associated components provide a regulated
point since MMICs do not offer the necessary low-impedance bias supply to maintain the
linearity and output capability. The BFR96S PA transistor in the linear portion of its
was chosen for its high collector-current characteristic. Connected to the output of the
rating and hence higher linear power output PA stage is a directional coupler providing
and is biased to a quiescent current of 70mA. rectified samples of the forward and reverse
It should not be replaced with a BFR96, which powers. These are fed back to the transverter
has inferior characteristics. The output stage is board to generate the ALC voltage. A further
followed by a low-pass filter with a cut-off lowpass filter with a cut-off frequency just
frequency of 80MHz to clean up the output above the 70MHz band is included before the
spectrum. antenna changeover relay, to reduce any
A CA3140 operational amplifier (IC5) unwanted harmonics produced in the PA or
generates an ALC voltage from the forward directional coupler.
and reverse voltages provided from the PA The complete components list for the
board. This ALC voltage is used to control the transverter board is given on page 932,
amount of attenuation provided by the PIN followed by the components list for the PA
9 24
CRAY 70MHz TRANSVERTER
board. To ease identification, all PA compo- the PA board is shown in Fig. 9.32; the
nents are numbered from 100. underside is left completely copper-clad. All
the PA components are mounted on the track
CONSTRUCTION side of the board as shown in Fig. 9.33,
The transverter consists of two printed-circuit allowing it to be attached directly to a
boards. The first contains the local oscillator, heatsink and case as shown in Fig. 9.34.
receive mixer, transmit mixer, ALC amplifier I chose to mount the whole transverter in a
and negative-supply generator; this is desig- large die-cast box (Fig. 9.31) although a more
nated the transverter board. The second board conventional type of case would be equally
contains the transmit amplifier, ALC detector, suitable. Before inserting any components on
low-pass filter and antenna changeover relay. the transverter and PA boards, their mounting
This is designated the PA board. Both boards holes (indicated by the letters H in Figs 9.29
are etched on double-sided PC board. and 9.33) should be drilled in the case using
The track layout of the transverter board is the bare boards as templates. Further details of
shown in Fig. 9.30a, and the component mounting the PA board and components are
Fig. 9.29. overlay in Fig. 9.29. The top side of the given later.
Component transverter PCB is left copper-clad (Fig. 9.30b), Begin assembly of the transverter board by
layout of the the copper being counterbored from around fitting R23. Then cut the four tin-plate screens
Cray 70MHz the holes where non-grounded leads pass and solder them to the board, leaving a notch
transverter through. The track layout of the top side of for R23. Copper-clad PCB does not provide
board
9 25
50 & 70MHz
Fig. 9.30a. Full-size PCB track layout of the Cray 70MHz transverter board, underside view
adequate screening in this application and 9.30. The MMICs have their ground leads bent
should not be used. Solder in all the resistors, upwards and soldered to the top surface of the
followed by all the capacitors. Mount all the board. Both of the SBL-1 mixers are then
moulded coils next, oriented so that the long soldered in place, the M in the MCL label
connection leg is grounded. The transformers on the can being over pin 2. The cans of the
are then wound as described in the compo- mixers should also be soldered to the top of
nents list (see page 910 for advice on wind- the board; do this quickly using a large, hot
ing toroids), and are secured to the board with iron. Wire links are required to join the top
a small amount of adhesive. The semiconduc- and bottom surface on the transverter board
tors are then soldered in place; the pinouts are as indicated by the letters X in Fig. 9.31. To
the same as shown in Fig. 8.13 for the Suffolk improve stability, RFC8 is soldered between
144MHz transverter. Great care should be the two relevant copper pads on the track side
taken not to damage the fragile leads of the of the board. The board is mounted in the box
MOSFET and MMICs, most of which are on four 8mm M3 metal spacers.
soldered on the track side of the board. A The PA board is mounted on the heatsink
clearance hole is drilled to locate the body of using M3 x 6mm cheese-head screws, the
each device, using the centres marked on Fig. heatsink having been tapped M3 in the
9 26
CRAY 70MHz TRANSVERTER
9 27
50 & 70MHz
Fig. 9.32. Full-size PCB track layout of the Cray 70MHz power amplifier board (top
view; underside is un-etched copper groundplane)
Fig. 9.33. Component layout of the Cray 70MHz power amplifier board
RFC100
+12 +12 + C110 +12 TX RX ANT
TX
IC100 +
C108
R107
C106
C107
+
C112
C120 RLB
+ C111
RFC103
C121
R106
RV100
H
R101
R102
H C119
C113
RFC101 C114 D103
L104
R100 + C104
+ C109
C118
C105
RFC104
RFC102
R103
H L103
H
CT100 E
C102
C117
C123
E
C115
CT102
F R105
H
C122
C116
D101
R104
CT103
H H
9 28
CRAY 70MHz TRANSVERTER
Fig. 9.35.
Interconnection
diagram for
the Cray
70MHz
transverter
RX IN TX OUT PA IN
+12V TX
+12V
PTT 28 IN
+12V +12V RELAY
28 OUT 12V TX
F
RX OUT
R
ANTENNA
R F +12V
0V
9 29
50 & 70MHz
9 30
CRAY 70MHz TRANSVERTER
indicates 30mA. The current should increase Cray transverter, a few configuration changes
gradually; any sudden jumps are a sign of will be needed. The 70MHz drive to the
instability, which should be remedied before external PAs input may be obtained from the
proceeding. Apply 28MHz drive to the antenna socket as normal. The coax lead
transverter, which should result in a slight connected to the RX OUT terminal of the PA
increase in collector current and possibly board should be transferred to a coaxial socket
indications of output power. Adjust CT100 mounted on the rear of the transverter box,
and CT101 for maximum collector current, and the receive signal from the PAs own
and then CT102 and CT103 for maximum RF external antenna changeover relay is con-
output power. None of the capacitors should nected to this socket. If the external
tune sharply, but all should show a peak. The changeover relay has a 12V coil, a suitable
available output should be something over switched voltage may be obtained from the
10W; however, to minimize the risk that the 12V TX terminal on the transverter board via
transverter is being overdriven, adjust RV1 a spare pin on SK4.
until the output power shown on the power
meter has just started to drop, and leave it at PERFORMANCE
that setting (see panel on page 6-7). The measured noise figure of a series of receive
The last task is to set up the ALC circuit. converters has been found to be around
Firstly ensure that the detector circuits are 22dB, suggesting that R5, R6 and R7 could be
working by measuring VF and VR, the for- changed to provide even more attenuation in
ward and reverse voltages from the VSWR the receive path. Rejection of a 14MHz image
detector, when the transverter is producing signal (42MHz minus 28MHz) is around 80dB.
full power output into 50. VF should be over The receiver has been found to be stable under
05V and VR should be much less than VF. a wide variety of conditions including an
Now connect VF and VR to the appropriate open-circuit input. All harmonics and spuri-
points on the transverter board. Adjust RV3 ous transmitter products are greater than 60dB
(GAIN) for maximum voltage of 8V at the down on the wanted signal, which is suffi-
output of IC5, and then back it off slightly to cient when the transverter is driving an
a reading of say 78V. Now adjust RV2 (ALC antenna directly. It is advisable to use an
SET) and note that the output power starts to additional lowpass filter if a further stage of
fall as TR5 reduces the bias current through amplification is used, or if the transverter
D4. Adjust RV2 for the desired RF output, board is used without the PA board. A copy of
which should not be more than 10W. the lowpass filter on the PA board would be
suitable for 100W output or more, if UNELCO
INTERFACING or SEMCO cased-mica capacitors are substi-
Interfacing of the Cray transverter to HF tuted for the ceramic-plate types.
transceivers is exactly the same as for the Two-tone measurements of the linearity of
other transverters in this book. The descrip- the transverter demonstrate that, with care in
tion of the Suffolk 144MHz transverter on page alignment, the third-order intermodulation
8-26 gives full details. If your HF transceiver products are 38dB down on 10W PEP. This
can only provide about 10dBm (100W) value is quite fair, and is better than that of
from its transverter socket, additional ampli- modular PAs used on the other bands. To
fier stages will be required in the TX IN feed. improve on this figure the transistor would
This may take the form of further MMICs have to be biased in class A, or else a bipolar
employing the same circuit as IC3. As the power transistor or a VHF power FET would
circuit is operating at low signal level, it may have to be used with a higher-voltage supply.
be built on a small piece of PCB and mounted
in the transverter box.
To interface an external amplifier to the
9 31
50 & 70MHz
9 32
CRAY 70MHz TRANSVERTER
9 33
50 & 70MHz
SOLID-STATE
POWER AMPLIFIERS
Chapter 6 is pretty scathing about the vast surplus sources are manufacturers drop-outs
majority of solid-state power amplifiers, and or early versions. Select a device which is
quite rightly so. And yet it is possible to build designed for your intended purpose. For
transistor PAs of quite acceptable linearity, example, transistors for Class-C (FM) opera-
with a little care and attention. The Big tion may produce lots of power in Class AB for
Mistake is to try and do it on the cheap. SSB, but they seldom produce more than
Manufacturers for the amateur market say that around a quarter of their rated Class-C power
we, their customers, expect solid-state PAs to with acceptable linearity. For the lower VHF
be cheap, so we only get what were prepared bands at least, you can use transistors specifi-
to pay for and theres some justice in that cally designed for SSB operation. If you really
remark. As a breed, we often seem to expect must use surplus devices, at least make sure
professional products at amateur prices. that theyre the right ones for the job!
Happily, that vicious circle is now begin- Try to get away from the amateurs insist-
ning to break. There is a concerted drive to ence on 138V supplies. Devices intended for
raise the standards expected from power 28V or even 50V supplies are more efficient,
amplifiers of all kinds, and information on have more gain, are more linear and are easier
how to build good solid-state PAs is beginning to match. Also, the lower-current power
to appear. John Matthews G3WZT has written supplies are easier to make. How much high-
two excellent articles on how to design and power SSB mobile operation do you intend to
build solid-state power amplifiers for 50MHz do anyway?
and 144MHz [1,2]. Even a cursory glance will Produce a gain budget before you start and
reveal the vast difference between his designs always try to have a few dB in hand. Its good
and the average black brick. The devices are to be able to put a 13dB attenuator pad at the
run well within their linear range, using 28V input of the amplifier, since this will reduce
or 50V supplies, and higher powers are the effects of varying input impedance upon
achieved by combining identical amplifiers. the previous stage.
Although transistors still cannot reach the Use passive components which have
ultra-smooth intermodulation standards of a adequate Q at the frequency of interest. For
really good pair of tetrodes (Chapter 6), a well- example, Philips film-dielectric trimmers are
designed solid-state power amplifier can be as often satisfactory in 200W amplifiers! Mica
clean as the normal run of valve PAs. But we compression trimmers are lossy; although
repeat: dont expect it to be cheap. they survive because they can also dissipate
Another exponent of high-quality solid- more heat, they are not a good choice unless
state PAs is Dale Harvey G3XBY, who pro- you really need a big capacitance variation.
vided the following design information. Padding capacitors should have low loss and
low inductance. ATC transmitting chip
COMPONENTS capacitors are certainly good up to 300W.
Use only new transistors they are worth the Try to keep coils in the clear. The field
money. Some transistors obtained from from the collector series coil in a 70MHz
9 34
SOLID-STATE POWER AMPLIFIERS
BLW96 PA was enough to melt a film trimmer voltage divider, well over an amp of standing
placed near the open end of the coil. Use current has to flow down the bias chain in
large-diameter wire for the collector RF coil order to provide anything near adequate
even 0141" semi-rigid coax gets hot owing to regulation. Most amplifier circuits just dont
the large circulating currents and the direct do that, so the bias voltage starts to fall on
thermal connection to the transistor. On the peak current demand. The operating point of
other hand, the collector DC feed coil carries the transistor then moves towards cut off and
much lower circulating currents and is much the collector current no longer rises in
less critical. response to the RF drive. The result is gain
compression and splatter on every speech
RF BYPASSING peak.
RF power transistors require thorough bypass- As the power transistor heats up, the voltage
ing of their collector and base bias supplies, at drop across its base-emitter diode decreases
all frequencies from AF to UHF regardless of and the collector current will start to rise. This
the actual frequency of operation. Otherwise in turn heats up the transistor, the collector
they will oscillate and thats a cast-iron current rises still more, and before you know it
guaranteed promise. To achieve a low imped- the transistor has gone into thermal runaway
ance over such a wide frequency range, several and fried itself. So part of the function of the
different types of capacitors need to be base-bias circuit is to prevent any tendency to
connected in parallel. Starting at the VHF/ thermal runaway by reducing its output
UHF end of the spectrum, use one or more voltage as the power-transistor junction
physically large ceramic capacitors of 1nF or temperature increases. This so-called thermal
less, preferably chip types but certainly with tracking is usually achieved by using a
the shortest possible leads. Take care of the semiconductor junction (often simply a diode)
middle frequencies by connecting a 100nF in the bias circuit as a temperature sensor.
polycarbonate capacitor in parallel, plus A lot of diodes have been trailed over the
possibly a 10nF ceramic for good luck. tops of power transistors in efforts to improve
Finally, add a 100F tantalum or solid- thermal tracking. But its the temperature of
aluminium capacitor for LF and AF not an the RF power-transistor junction that really
ordinary aluminium-foil electrolytic because matters. Outside the semiconductor wafer
its inductance is too large for effective opera- itself, the best thermal contact with the
tion beyond AF. However, since even tanta- transistor junction is at the mounting flange
lum electrolytics have appreciable inductance, or on the heatsink. The temperatures in those
parallel resonances with the smaller capacitors locations can be sensed by screwing a tab-
could produce holes in the broadband mounting power transistor directly to the
bypassing capability youre trying to achieve. metalwork. If the bias circuit calls for a diode,
So the tantalum capacitor should always be use a PNP transistor with its base and collector
separated from the rest of the RF components tab grounded. A bias supply with good
by a ferrite bead to kill any HF/VHF reso- thermal tracking can help prevent the slow
nances. onset of thermal runaway caused by long
periods of full-power operation. However, if
BIAS CIRCUITS the power transistor decides to run away
Many transistor PAs are let down by their suddenly for whatever reason the bias
base-bias circuits, which fail to provide supply cant do a thing to stop it. In such a
enough current on peaks of drive. For exam- case, the underlying problem is thermal, not
ple, if the peak collector current is 10A, the electronic. Either the heatsink isnt big
bias circuit must provide its stabilized voltage enough, or the late lamented RF power
at something in the region of 0205A. In the transistor wasnt in proper thermal contact
traditional circuit, a simple resistor-and-diode with it.
9 35
50 & 70MHz
Fig. 9.36. Circuit for 50MHz and 70MHz power amplifiers by G3XBY
9 36
SOLID-STATE POWER AMPLIFIERS
the cutter. You can of course etch the copper film of heatsink compound, just enough to fill
instead. Leave plenty of clear insulation any irregularities and use the proper stuff,
around the output end of the collector coil, not ordinary grease! Finally, apply the correct
where high RF voltages are developed. torque to the mounting screws. If you dont
Mount the board flat on to the bottom of a have a torque screwdriver, tighten the screws
die-cast box, with the lid acting as a screening moderately, but not too hard. The screws
cover. Dont even think of using the box itself should be merely holding the transistor in
as a heatsink; cut away both board and box so place; if the mating surfaces are properly flat,
that the flange of the transistor bolts directly it shouldnt be necessary to pull the transistor
to the heatsink. If the heatsink has one flat down on to the heatsink.
side which is bolted to the base of the box, the Make sure that the depth between the PC
thickness of the box plus that of the board is board and the heatsink is not too great (see
just right for a standard flange-mounting RF above). Otherwise youll split the top off the
power transistor (see below). ceramic package as you tighten down the
Pay particular attention to the grounding device. This could also happen if you solder
arrangements close to the power transistor. To the device into place before tightening the
get the expected power out of the device, mounting screws. A broken power transistor is
good emitter grounding is crucial. The scheme not only expensive but also dangerous all RF
in Fig. 9.37 works well; G3XBY has used it in power transistors contain beryllium oxide,
seven amplifiers in the 50150W range, and which is toxic if ingested, inhaled or rubbed
every one has produced the expected gain into cuts. In case of problems, DO NOT blow
within 1dB. out the dust! Mop it up with a damp cloth,
and dispose of the cloth and the device in an
MOUNTING POWER industrial-waste skip. A useful tip to aid the
TRANSISTORS removal of transistors without breaking them
Mount the power transistor(s) directly to a open is to bend up one corner of each tab
heatsink of the correct rating [3], with the before soldering, so that you can use pliers to
recommended mounting hardware. For the peel it back if necessary.
most efficient heat transfer, make sure that
both mating surfaces are flat. Remove any POWER SUPPLIES AND
high spots from the heatsink and the power CONTROL CIRCUITS
transistor flange using wet-and-dry paper on a Always use regulated power supplies. If the
flat supporting block. Apply only a very thin voltage sags on peak current demand, you will
9 37
50 & 70MHz
lose power and the linearity of the amplifier COMPONENTS FOR 50MHz
will be limited. An overvoltage crowbar AND 70MHz
circuit is a very worthwhile feature in the POWER AMPLIFIERS
power supply. The BLW96, for example, has a
RESISTORS
VCE rating of 110V (DC+RF), which means that 15R 1 RV1: Multi-turn trimpot
it will run quite safely from a 50V supply, but 68R 1 R3: 05W
100R 1 R2: 20W
anything above 55V is likely to blow the 5k6 1 R1: 1W
device when RF drive is applied. CAPACITORS
Control circuits should take care of the 47p 4 C3, 12, 14, 16: 300V transmitting
chip capacitors, ATC or Waycom
transmit/receive sequencing and also protect 57p 4 C1, 2, 13, 15: 300V foil trimmers,
Philips type 200-8090-8003
the transistor against excessive VSWR. The 1n 3 C4, 7, 8: ceramic chip
articles by G3WZT [1,2] contain some very 1n 1 C11: ceramic feedthrough
100n 5 C5, 9, 17, 18, 19: polycarbonate
good ideas about power supplies and control 100 1 C10: 64V tantalum electrolytic
220 1 C6: 63V tantalum electrolytic
circuits for solid-state power amplifiers.
INDUCTORS
L1 50MHz 2 turns
IMPEDANCE MATCHING 70MHz 15 turns
6mm inside diameter
Tune everything up in a 50 environment. 18SWG tinned copper wire
For intermediate amplifiers, pay careful L2 6 turns through Philips FX1898 six-
hole ferrite bead
attention to the input matching of the L3 50MHz 8 turns, 12mm inside diameter,
18SWG enamelled
succeeding stage; and for the power amplifier, 70MHz 5 turns, copper wire, close wound
make sure that the antenna is well-matched. L4 18SWG wire through 2 ferrite beads
L5 50MHz 4 turns
Otherwise, all the tuning and matching 70MHz 3 turns
12mm inside diameter, 12SWG
adjustments will interact, your power meters copper or 0141" solid coax,
will not read correctly and you can never preferably silver-plated, turns spaced
1 wire diameter
understand what is going on. If you are using
SEMICONDUCTORS
combined amplifiers, matching errors will BLW96 1 TR1: Philips (Gothic Crellon Ltd)
cause unequal power sharing between the two BD135 1 TR2
BD228 1 TR3
amplifiers. So check the VSWR into and out of 1N4007 1 D1
every stage; if its worse than about 14, do
something about it.
impedance transformations. In other words,
POWER AMPLIFIERS FOR high-power solid-state amplifiers at these
50MHz AND 70MHz frequencies are still an area for the more
These two amplifiers are each capable of over advanced experimenter.
100W PEP output. They use a 50V transistor
specifically designed for linear operation, the
Philips BLW96. Fig. 9.36 showed the circuit,
which is essentially the same for both bands.
Note that this is only a bare circuit, presented
to give you a few ideas. Having invested a
large sum in the RF power transistor, you are
expected to look after it by providing switch-
ing and VSWR protection as described above.
The coil dimensions may require a little
experimentation, because the values of these
small inductances depend very much on the
arrangement and lengths of their leads.
Similarly it may be necessary to add or remove
47pF chip capacitors to achieve the correct
9 38
ANTENNAS FOR 50MHz
AND 70MHz
Since even a minimum-sized DL6WU long optimum. Electrically short Yagis can be very
Yagi is very long indeed at a wavelength of frequency-sensitive, unlike their longer
6m or 4m, this chapter gives some designs for counterparts used on the higher bands. Since
shorter Yagis. 50MHz and 70MHz DX operation takes place
The mechanical construction adopted for in narrow frequency segments, an error of less
50MHz and 70MHz Yagis depends largely on than 1% in element lengths can shift the
the weather conditions they are likely to have optimum performance right outside of the DX
to experience. For portable use, mostly in sub-band. And a 1% error in the length of a
summer, a very lightweight form of construc- Yagi element represents only 30mm at
tion can be used, but building a Yagi to 50MHz, or 20mm at 70MHz.
survive several years of Scottish winters is It is a common misconception that an
quite another matter! The Yagis described here antenna must be resonant or on frequency
use 127mm (05") diameter elements and if the VSWR is good. This is untrue. Even if
25mm (1") square booms. They are light the antenna has been adjusted to a perfect
enough for portable use, but should also stand 50 match it can still be hopelessly mistuned
up to the rigours of winter in most of the UK. for pattern and gain, as is regrettably demon-
In parts of the world which are subject to ice strated by many commercial Yagis. The only
storms the construction might need to be way to get the antenna working properly on
strengthened to withstand thick layers of the frequency you want is to alter all the
heavy ice. element lengths. When you do this, the
VSWR is bound to become worse until you
TUNING THE ANTENNA have also re-adjusted the matching, but you
For the best DX performance, you will always shouldnt regard that as a problem.
need to check the performance of your The easiest way to check the performance is
antenna in your particular installation. to compare the rearward polar pattern with its
Especially at these longer wavelengths, the predicted shape. The rearward lobes of all
antenna is likely to be quite close to ground. It kinds of electrically short Yagis change with
may also have antennas for higher frequencies frequency in a very similar way, as shown in
inside its capture area (Chapter 7), and these the seven polar plots comprising Fig. 9.38a-g.
too can cause detuning. The first thing to note is that although the
For any antenna, the effective electrical plots cover a span of only 15% in frequency,
length of the elements depends on the the patterns change dramatically within that
element diameter and the method of mount- range, so they give a very good indication of
ing on the boom. As explained in Chapter 7, how the antenna is tuned.
these effects can cause the antenna to be When the antenna is operated far below its
completely off-frequency. This leads to optimum frequency, the front/back ratio is
incorrect polar patterns, poor sidelobe per- poor (Fig. 9.38a). As the frequency increases,
formance and probably reduced gain, the gain the backlobe begins to split (Fig. 9.38b-c) and
falling off more rapidly on the HF side of there is generally one frequency at which the
9 39
50 & 70MHz
Fig. 9.38a. Fig. 9.38e.
9 40
50MHz & 70MHz ANTENNAS
Fig. 9.39.
Computed
pattern of
the 8-
element
70MHz Yagi
split becomes deep and the front/back ratio is To tune a Yagi, begin by inspecting the
very high. A small further increase in frequency rearward pattern using test signals from
produces a new HF rear lobe (Fig. 9.38d) another amateur at various points across the
which rapidly grows outwards at 180 (Figs band. Having located the optimum frequency
9.38e-f-g), while the remains of the old LF range, adjust all element lengths in the same
backlobe wither away. Meanwhile the forward proportion to bring the optimum pattern into
gain has been gradually increasing with your chosen part of the band. Ignore the
frequency, although it will eventually begin to VSWR until youve got the pattern right, and
drop very sharply. If you go even further HF, then finally adjust the matching if necessary.
the first director will begin to act as a reflector,
and the antenna beams backwards! FOUR- AND EIGHT-ELEMENT
The true optimum frequency is debatable. YAGIS
Optimizing for the best pattern will generally The table below shows the dimensions for a 4-
involve the sacrifice of a few tenths of a dB of element Yagi for 50MHz and 4- and 8-element
forward gain, but will greatly improve the designs for 70MHz. The 4-element design (Fig.
signal-to-interference ratio. The pattern of Fig. 9.40) has about the highest gain obtainable
9.38d gives quite good suppression of un- from that number of elements, and is quite
wanted signals from the rearward direction in unusually long. It is taken from the excellent
general, and also provides a number of deep book on Yagis by W2PV [4], and the patterns
nulls which can be very useful in eliminating in Fig. 9.38 were computed for this particular
QRM. In fact, any of the three patterns in Fig. design. At the frequency of optimum pattern
9.38c-d-e would be pretty good in practice, (501MHz, Fig. 9.38d), the predicted forward
and between them they will nicely cover the gain is a very respectable 86dBd, which is
DX portion of any VHF amateur band. significantly better than for shorter 4- or 5-
9 41
50 & 70MHz
Fig. 9.41.
Element
construction,
mounting
method and
gamma
match for the
50MHz and
70MHz Yagis
9 42
50MHz & 70MHz ANTENNAS
35
Weatherproof
junction box
for coax
175
50MHz 350
50MHz 100 70MHz 250
70MHz 75
PTFE sleeve
Heat-shrink
Clip
35
All dimensions are given in mm, and the element lengths apply only to the method of construction
described.
9 43
50 & 70MHz
9 44
432MHz
10 1
432MHz
A DXERS TRANSVERTER
FOR 432MHz
by John Wilkinson G4HGT
432MHz RF RF RF RF Mixer IF IF
amplifier filter amplifier filter amplifier filter 28MHz
Amplifier
101.0MHz
Xtal x4 RF Power
oscillators filter divider 404/406MHz
101.5MHz
Amplifier
RF RF RF RF IF
filter amplifier filter amplifier Mixer amplifier Atten 28MHz
10 2
DXERS 432MHz TRANSVERTER
that you yourself need to optimize or adjust. ring mixers are the simplest and best way of
Even so, a high-performance 28432MHz achieving the target of 90dB for the spurious-
transverter is no project for a novice. In spite free dynamic range (Chapter 5). Finally,
of all my attempts to make the design tolerant excellent filtering of all out-of-band receiver
and reliable, there are still a few tricky areas, responses is needed to keep out the image
so my description concentrates on these. I response at 376MHz, direct breakthrough of
have included only the basic outlines of 28MHz, and every other frequency except for
interfacing, control circuitry and mechanical the desired 432436MHz. Commercial helical
construction. If you feel confident enough to filters have been used for both the signal
build this transverter from my description, frequency and the 404/406MHz local oscilla-
perhaps taking a few hints from G4DDKs tor. The problems in developing home-made
Suffolk 144MHz transverter design in Chapter UHF bandpass filters, not to mention describ-
8, wiring up the LEDs and drilling a few holes ing how you could duplicate them, were just
should be no problem to you. too numerous to contemplate.
For alignment, a multimeter and On the transmit side, output levels of 3W,
milliwattmeter (Chapter 12) are required, 10W and 25W PEP are the common standards
together with a wavemeter or frequency for commercial gear. The problems of the
counter. A stepped attenuator is also useful, power-amplifier stage have largely been
though not essential. In case everything does overcome by the availability of hybrid mod-
not go smoothly, you also need to be reason- ules by Mitsubishi. Two variants are available,
ably confident about fault-finding using the with linear output powers of 10W and 25W
aforesaid test equipment. The modular design PEP, and either could be used here. Their only
approach will help in determining any limitation is their intermodulation distortion,
problems quickly. which is adequate but not outstanding
(Chapter 8). Most serious DXers will actually
DESIGN CONSIDERATIONS choose the lower-power 10W module because
I began the design exercise with a survey of it is more than adequate to drive a K2RIW
existing transverter and preamplifier circuits. amplifier in Class AB1.
Surprisingly, there arent that many [1-7] and This transverter could form part of at least
only a couple of them are actually worth four different types of 432MHz system,
serious consideration. Also, it is very suspi- intended for:
cious to see claims of system noise figures 1 Terrestrial DX from home station (good all-
comparable with the NF of the front-end round performance).
device itself, or of power-amplifier gains 2 Contests and DXpeditions (high quality
which are higher than those in the device throughout).
manufacturers published data! I have there- 3 Satellite working (435436MHz coverage
fore taken a pessimistic approach to noise required, and also good receiver sensitivity).
figures and stage gains, in the hope that 4 EME (ultimate high performance, especially
anyone else building these circuits will obtain in receiver sensitivity).
results that will definitely meet my design Receiver sensitivity is important in all four
specification and hopefully exceed it. cases, and the best way to achieve it is to use a
What do you want from this design? For preamplifier mounted close to the antenna;
DX-chasing, low noise figure is important, this will also relax the sensitivity requirements
although with modern devices this is not for the transverter. But there are many
difficult to achieve. At 432MHz a system noise amateurs who either do not have easy access
figure of 18dB is a reasonable design target to the antenna or prefer to have all the
(see Chapter 4). For contest operation a high electronics indoors. For them, the transverter
dynamic range is essential, even at the itself must have a low noise figure.
expense of some sensitivity. At UHF, diode- If you subsequently add a masthead preamp
10 3
432MHz
to a sensitive transverter which already has same lines, I have kept coaxial relays out of
quite a lot of front-end gain, you are likely to the transverter design and simply provided
find problems with strong signals. The logical the necessary switched DC supplies. Some
solution is to have the ability to bypass the operators will want the coax relays at the
first RF stage in the transverter if desired, so masthead, while others prefer to mount them
that if the preamplifier fails you can get back on the power amplifier or even on the shack
on the air with a sensitive system before the wall its your choice.
end of the opening or the contest. Along the The overall block diagram is shown in Fig.
10 4
DXERS 432MHz TRANSVERTER
10.1, and the modular approach is taken one they arise. Its also very encouraging to see
step further by dividing the entire transverter your project showing signs of life within the
into functional blocks with 50 interfaces. A first few hours after starting construction.
lot of constructional articles recommend The transverter is constructed on four
building the whole thing and then smoke double-sided fibreglass printed-circuit boards
testing, but I prefer to build and test each 1 Receive converter
stage before proceeding to the next. This 2 Local oscillator
makes it easier to deal with any problems as 3 Transmit converter, with TX/RX switching
10 5
432MHz
10 6
DXERS 432MHz TRANSVERTER
n = m2 m 1
10 7
432MHz
mixer and feedback amplifiers. The target Fig. 10.3. Full-size PCB layout on the
noise temperature of 150K is achieved almost underside of the 432MHz receive-converter
exactly, and with a good HF receiver the board. The top of the board is a continuous
copper groundplane, etched or counterbored
dynamic range is very respectable.
as shown
The first RF stage (Fig. 10.2) is a dual-gate
GaAsFET, providing a good balance between
performance and cost. Whilst microwave
single-gate GaAsFETs have excellent noise
performance at UHF, the improved sensitivity
is not really necessary unless EME is contem-
plated. The high associated gains of micro-
wave GaAsFETs at lower frequencies also
increase the likelihood of instability possibly
in the gigahertz region and the high input
impedance can also cause problems in
achieving a low-loss match [8, 9]. Therefore an
entirely sufficient solution for 432MHz
terrestrial DXing is the modern dual-gate
GaAsFET. As more of these robust devices are
designed into consumer products, so the price
will fall, making them an even more sensible
choice. Virtually all of the common devices
available (CF300, MGF1200, S3030, 3SK97 etc)
should work well in the design adopted, all
giving gains around 1620dB and device noise
figures below 1dB. This low device noise figure
can then be traded-off in the interests of
improved dynamic range for the system as a
whole, as explained in Chapters 4 and 5.
For both input and output circuits, a length
of silver-plated wire is used as the inductor.
Both ends are bent down to fit through holes
in the PC board and the inductor is spaced
Fig. 10.4. Component layout of the 432MHz
5mm above the board. Taps on the inductors receive-converter board. Note that several
are used for matching, allowing optimization components are mounted beneath the board
of the gain and noise figure. As an alternative, (no component outlines shown)
provision has also been made on the PCB for
capacitive coupling to the top of the input
tuned circuit. A simple tin-plate screen is
needed to isolate the two tuned circuits and
prevent oscillation. The output of the first RF
stage is fed via a resistive pad to a Toko pre-
tuned helical filter which provides significant
selectivity. In the TCALC runs, the loss value
used for these filters has been taken as 3dB
the worst case according to the manufacturer.
I have measured the loss as typically 17dB,
with a 376MHz image rejection of 41dB for a
single two-cavity filter.
10 8
DXERS 432MHz TRANSVERTER
10 9
432MHz
10 10
DXERS 432MHz TRANSVERTER
voltage regulator is working. Then install TR3, stalled on metal pillars in its box. The lead to
reconnect the supply and check the DC the input BNC connector should be as short
conditions. This stage can be checked at RF by as possible, and coaxs braid well soldered to
using it as a simple preamp into the 28MHz the PCB groundplane without any tail.
receiver, the antenna replacing the mixer. If
TR3 oscillates, install the chip capacitor C23 LOCAL OSCILLATOR
beneath the board, directly beneath C24. If While much time is spent discussing the
oscillation persists, double-check the windings merits of receiver front-ends and transmitter
on the transformer L6 with a powerful linearity, the local oscillator is frequently
magnifying glass, and make sure you have overlooked in transverter designs. In fact its
used the correct bypass capacitors with zero specification needs to be determined quite
lead lengths. The output filter L7-L8 is reason- carefully since the LO is responsible for
ably broadband and should need little adjust- frequency accuracy and stability. A local-
ment. oscillator signal for 432MHz will normally be
Before you can go any further with aligning derived from an oscillator running at a lower
the receive converter, you need to have built frequency which is then multiplied to the
and aligned the local-oscillator board as desired harmonic, so any frequency errors
described in the next section. Having done and instability will be multiplied too.
that, solder in the receive mixer. As well as The right design of crystal oscillator will
soldering the pins beneath the board, carefully ensure excellent stability with temperature
solder the corners of the mixer case to the PCB and time, as well as the lowest phase-noise
groundplane, working quickly with a large, characteristics. One or two designs I tried
hot iron. Connect up the local oscillator, and have only worked with the output frequency
with just a simple antenna connected to the slightly offset from the value marked on the
input you may be able to hear local 432MHz crystal, and this undesirable offset is multi-
signals or the third harmonic of your 144MHz plied four-fold at 432MHz. G4DDK recom-
transmitter. mended the Butler oscillator configuration in
Install TR2 and set potentiometer RV1 to Fig. 10.5, and I have been very satisfied with
minimum resistance. Connect power to TP7 the results. At 404MHz the frequency can be
and set up the correct bias point (IC = 20mA, shifted 10kHz without loss of oscillation,
095V across R14) using RV1. Signals should and with only a minimal change in output
be audible with an antenna attached to TP6. level.
Finally install TR1. Modern GaAs devices are Stability of frequency with supply voltage
quite rugged, and fanatical care to protect and load variations is also important. An
against static damage is unnecessary. Power up integrated-circuit voltage regulator provides
and check the drain current (about 25V across better voltage stability than the commonly
R4). The positions of the taps on the inductors used Zener-diode arrangement, and each
should be acceptably close to the optimum for oscillator has its own regulator. By using
gain and noise figure. If the tap is used on the separate buffer amplifiers to feed the transmit
input, C3 must be omitted; alternatively, use and receive mixers, the load presented to the
C3 for capacitive input matching and omit the oscillator output is more constant, reducing
tap. Final adjustment for optimum noise the occurrence of frequency pulling due to
figure is best carried out using a proper noise load variations.
generator and NF indicator (Chapter 12), but Two entirely separate crystal oscillators in
if these are not available you can optimize the the 101MHz region are provided for terres-
signal/noise ratio of a weak but consistent off- trial and satellite operation. Selection is made
air signal (see Chapter 5 for a method using an using DC control to permit split-frequency
FM signal). operation, and circuits can easily be devised
Do a final tweak when the board is in- to connect the two oscillators to the TX/RX
10 11
432MHz
10 12
DXERS 432MHZ TRANSVERTER
10 13
432MHz
10 14
DXERS 432MHz TRANSVERTER
possible, note which is the higher and use it of ZIN and ZOUT to 50 were then extracted
for the receiver local oscillator. from the tables in [11], although some
experimentation was still needed to obtain the
TRANSMIT CONVERTER maximum output from the stage. DC is fed to
This section of the design takes an input signal the collector via an small RF choke L7; it may
in the range 2830MHz and gives an output at be possible to replace this component by a few
432434MHz, or 434436MHz with the turns of enamelled copper wire wound around
alternative LO frequency. There is no reason a low-value resistor. The base bias voltage is
why almost any IF in the range 2070MHz derived using a circuit used many times
could not be used, although consideration of elsewhere, and temperature compensation can
harmonics and other intermodulation prod- be arranged (or at least attempted) by ther-
ucts should exclude certain frequency plans. mally coupling the diodes to the transistor.
The transmit strip is essentially broadband in The voltage is fed to the base via RF choke L6
both its LO and IF sections, while the 432MHz and resistor R21, both of which are needed to
amplifier chain is very narrowband with a ensure stability.
great deal of filtering (Fig. 10.8). So long as the The BFG34 output stage needs to be driven
frequencies and levels at the LO and IF inputs at around +3dBm. It is preceded by a helical
are correct, the output will be on 432436MHz filter (assume 3dB loss) so the driver needs to
and nowhere else. be capable of +6dBm output. Once again
The design of the transmit strip can be adding a 3dB margin for linearity leads to a
approached from either the input or the desired 1dB compression point of +9dBm. The
output, but if linearity is of importance you Avantek MSA0304 MMIC fits the bill nicely
should start by thinking about the output with a 1dB compression point of +9dBm and
level. In this way you will avoid having to 12dB of gain.
squeeze every last dB out of the final stage. For Continuing backwards from IC4 (Fig. 10.9)
compatibility with the Mitsubishi power- another helical filter is preceded by another
amplifier modules, an output power of MSA0304 (IC3) which is again operated well
+17dBm (50mW) is required. It is a reasonable inside its linear region, and this is fed by the
assumption that third-order intermodulation mixer. Using an MMIC directly after the mixer
products will be about 30dBc when an is a good way of providing an accurate
amplifier is operating at its 1dB compression wideband termination, and its noise figure is
point. Following the well-known rule of low enough to ensure a very low level of
thumb (think of a number and double it, ie wideband noise in the transverter. In the final
add 3dB), I therefore aimed for an output design I included a small resistive attenuator
compression point of +20dBm, so that all between the mixer and IC3 to adjust the
intermod products from this stage would be overall gain to the desired value.
well down. With so much spare gain in the amplifier
The BFG34 is a good device for the output stages, it is possible to run the mixer at a very
stage. It has a maximum collector current of low signal level to reduce its contribution to
150mA, so it can easily handle the RF power intermodulation. With the sequence of stages
level required, and being a modern device it described, the RF drive to the mixer input can
has a useful power gain of 17dB at 432MHz. be kept below 8dBm. On the input side of the
Any RF transistor operated at the +20dBm mixer is a simple DC-controlled RF attenuator
power level will have input and output network. Input drive at 28MHz is first attenu-
impedances which are lower than 50. In the ated by the preset potentiometer RV1 and
case of the BFG34, the s-parameters given in then passes through the PIN diode D1. The
the data sheet were converted into the forward DC current through this diode
equivalent real and imaginary components. controls its RF attenuation. This current is
Networks designed for matching these values provided by TR1 which is wired as an emitter
10 15
432MHz
follower to provide a relatively high imped- the Suffolk transverter described in Chapter 8.
ance for the control-voltage line. After the Also on the transmit-converter board is the
attenuator, some amplification at 28MHz is TX/RX switching (Fig. 10.8). I have deliberately
required to restore the signal level; IC2 kept this as simple as possible, and you may
(MSA0404) gives around 8dB of gain. RV1 wish to expand this part of the circuit to
should adjusted to set the transmit strip interface with your existing equipment.
output level to just over +17dBm. If you Chapters 8 and 11 between them give compre-
require an ALC loop to control the output hensive advice. For both transmit and receive,
from the transmit strip, refer to the circuit of pairs of switching transistors give either low
10 16
DXERS 432MHz TRANSVERTER
(50mA) or high (200mA) current capability. I following Fig. 10.10. Use surplus wire to
have used the former for control of the output provide through connections from the
relays and indicator LEDs, with the high- groundplane to the terminals of IC24, TR2
current devices powering the transmit and etc, and remember to solder all ground
receive boards. connections to the groundplane.
Fig. 10.9 shows the etching masks for the Install the regulator IC1 and check that its
underside and the top of the transmit con- output voltage is satisfactory. Next install IC2
verter board. Assemble all the resistors, and ensure it draws the correct current by
capacitors and inductors on the PC board, checking the voltage across R8; for 50mA
10 17
432MHz
standing current this should be 60V. Solder in 0dBm, simply set the drive to the maximum
TR1 and the PIN diode D1. Set RV1 to mid- available. Connect the milliwattmeter to TP3
way. Apply DC power and measure the (Fig. 10.10) using a coaxial lead. Apply 28MHz
voltage across R5 to check that RV2 controls drive and some output at TP3 should be
the current through D1. Leave RV2 at its observable. Adjust RV1 until the level is around
maximum setting, corresponding to mini- 5dBm. Now slowly decrease RV2 and the
mum attenuation through the diode. signal will start to reduce; stop when the
It is important to set the HF drive level before signal level has fallen by around 1dB. From this
installing the mixer. Begin by using a point onwards, any negative-going ALC voltage
milliwattmeter to set the drive from the HF rig applied to TP2 will reduce the drive to the
to 0dBm (yes, only 1mW) maximum, and mixer.
install a permanent attenuator if the level is too Finally, adjust RV1 to limit the mixer drive
high. If the available HF drive level is less than level at TP3 to around 8dBm. The input
Fig. 10.9. Full-size PCB layout on the underside of the 432MHz transmit-converter board. The
top of the board is a continuous copper groundplane, etched or counterbored as shown
10 18
DXERS 432MHZ TRANSVERTER
circuitry is now set up and no further adjust- Apply 28MHz drive cautiously and check for
ments to RV2 should be needed. If the 28MHz output on the power meter. The four trimmer
drive level ever needs to be changed, readjust capacitors are simply peaked for maximum
only RV1 to control the drive to the mixer. signal output. If the above procedure has been
Remove the milliwattmeter from TP3 and followed, it should be possible to adjust RV1
install the mixer, soldering its case carefully to to give an output level of +20dBm before there
the PCB groundplane. With the local oscilla- are any signs of limiting. Finally adjust RV1 to
tor running and connected to the mixer port, reset the output to +17dBm (50mW), the
the rest of the transmit strip can then be built correct level for the modular power amplifier
up stage-by-stage. Having first checked the DC which follows.
conditions on the MMIC IC3 (40mA, ie 60V
across R12), connect the power meter to its MODULAR POWER AMPLIFIER
output point and observe the presence of a One of the more time-saving new products to
signal when drive is applied. Move on stage by emerge recently has been the power-amplifier
stage, until a signal is obtained from the module, with performance aimed specifically
output of F2. at the mobile-radio and amateur markets. At
Next solder in D2, D3 and TR2 and set the the time of writing, Mitsubishi Semiconduc-
wiper of RV3 to ground. Connect the power tors supply a range of modules covering 50
meter to TP8 and the DC supply voltage to 1300MHz, with output powers as high as
TP7. With no RF drive, increase RV3 to give 30W. Versions are usually available for either
15V across R18, corresponding to a standing FM or SSB, the latter simply containing extra
current of 80mA. Any fluctuation in the circuitry to bias all the transistors into Class
voltage across R18 as the potentiometer is AB1.
varied could be due to instability and must be As with a lot of convenience products, one
investigated. Leave the circuit powered-up for or two parameters have to be compromised. In
a few minutes and then reset the standing the case of these PA modules, the compromise
current if it has changed significantly. is that their linearity is not as good as one
10 19
432MHz
might like. At their rated PEP outputs, these amplifier to full output power on SSB. When
modules have third-order intermod products you purchase the module, also obtain the full
of typically 32dB, with equally reasonable five-page data sheet. The PA module will be
levels of higher-order products. Although this the most expensive single item in your
is quite good going for a transistor PA, at transverter, so take your time with this section
lower drive and output levels these intermod of construction to avoid introducing costly
products do not fall away as rapidly as you problems.
would expect. It follows that you cannot The modules are hybrid circuits containing
improve the linearity of a modular PA merely discrete components for matching and
by under-driving it. In contrast, a discrete- biasing, and, thanks to some clever produc-
component PA with lots of variable capacitors tion tolerancing, no adjustments are necessary
can generally be optimized at any chosen when the module is installed. The compo-
power level, producing a much cleaner signal nents are encased by a black plastic housing
but it is equally possible to get the adjust- attached to an aluminium heat-transfer plate.
ments totally wrong! At least with the module In the 430MHz devices there are three transis-
you cant go too far astray. tors which have to be cooled via this plate, so
The manufacturers data show that these the heatsink needs to be somewhat larger
modules can also be driven much harder, to than those for the two-transistor modules
give output powers up to 50% higher than used for lower frequencies. A thermal resist-
their specified maximum. But the intermod ance less than 1C/W is necessary for a 100%
products will then be rapidly degrading and duty cycle and for high reliability under
the extra output power will simply contribute contest, expedition and EME conditions. It is
to the width and splatter of your signal. Dont always worthwhile to use a large heatsink
be greedy settle for the manufacturers rated select the transverter case to suit the heatsink,
linear power. not the other way round!
This design uses the Mitsubishi M57716 DC supplies could hardly be simpler, as Fig.
430MHz amplifier which is rated at 10W. This 10.11 demonstrates. Supply voltage is applied
is more than enough to drive a K2RIW directly to the final transistor stage, pin 4
10 20
DXERS 432MHz TRANSVERTER
Fig. 10.12. Full-size PCB layout on the top-side of the 432MHz PA/VSWR board. The underside
of the board is a continuous copper groundplane
10 21
432MHz
provided to allow DC switching of the ampli- ing the value of R3 if necessary. Also check the
fier. operation of the DC switching by connecting
The supply current can be monitored using TP1 to +125V; the bias voltage should fall to
a voltmeter connected across the low-value less than 2V and the LED should go out.
resistor (R5) in the supply lead. The PA module can then be installed.
Remember to place a thin smear of heatsink
VSWR/POWER METER compound on the flange, and tighten the
Connected directly to the output of the PA screws before soldering the leads. Connect a
module is a printed-circuit VSWR/power dummy load and power meter to the output
meter. Diodes D2 and D3 (Fig. 10.11) rectify connector, and monitor the supply current as
the sampled signal, and must be Schottky- you apply DC power. Little or no current
barrier devices for correct operation. The should be drawn in this state. Connect TP2 to
forward and reflected DC voltages can be a +125V DC supply; the module should draw
fed either directly to a meter, or to an ALC approximately 300mA. Now connect the
circuit as described in Chapter 8 if preferred. output of the transmit converter to the PA
The power-amplifier PC board is single- input using thin 50 coaxial cable (RG174). If
sided (Fig. 10.12), the reverse being un-etched you have set the transmit converter output to
copper. Multiple ground wires connect the +17dBm maximum, it will not be possible to
grounded tracks on the component side to the overdrive the PA. Apply enough 28MHz drive
groundplane. The board is attached to the rear to obtain about 10W RF output; the supply
panel by four M3 screws and held clear by current should be continuously monitored
plastic washers. and should not go significantly above 25A.
Before you solder anything to the PC board, The current should fall back to the 300mA
try assembling the board and the PA module standing-bias level as soon as the drive is
to the heatsink, using two M3 screws and removed. If this does not happen, or if there
tapped holes in the heatsink. Avoid stressing are any sudden jumps in RF output or DC
the package of the module since external current, it is likely that some instability is
forces or mechanical shock can cause internal occurring; this must be investigated and cured
damage. A torque of 5kg/cm on the screws is before putting the transverter on the air.
all that is required to ensure sufficient thermal
contact to the heatsink. Since this may not be
enough to retain the screws against vibration,
use lock washers or Loctite on each
screwhead. A couple of short ground-straps
made from wide copper strip (or solder tags
and the braid from UR43 coaxial cable) are
used to provide low-impedance connections
between the PCB groundplane and the
aluminium cooling plate, which also forms
the modules RF ground.
Once you are happy that the mechanics of
the module and PC board are correct, remove
the module and assemble the rest of the
components on the board (Fig. 10.13). One of
the mounting screws to the rear panel is used
to mount the regulator IC, and insulating
washers are necessary to keep the heatsink tab
isolated. Connect +125V DC to TP9 and
check that the bias supply gives 90V, adjust-
10 22
DXERS 432MHz TRANSVERTER
COMPONENTS LIST
LOCAL OSCILLATOR FOR 432MHz TRANSVERTER
RESISTORS INDUCTORS, RF CHOKES AND FILTERS
All 025W 2% or 5% L1, 3 Toko S18 301AS0500, green (55
10R 4 R9, 11, 24, 26 turn) coil with aluminium core
47R 2 R4, 19 L2, 4 15 turn 4mm diameter, 22SWG
100R 1 R30 enamelled copper
150R 2 R10, 25 L5, 6 1H moulded chokes with axial leads
180R 2 R31, 32 L7, 8 05 turn, as L2, 4
220R 2 R12, 27 L9, 10 10 turns, 3mm diameter, 22SWG
330R 4 R2, 8, 17, 23 enamelled copper
560R 2 R3, 18 F1 405MHz, Toko 252MX-1547A
820R 11 R1, 5, 6, 7, 14, 15, 16, 20, 21, 22, 29
1k5 2 R13, 28 SEMICONDUCTORS
BFY90 1 TR1-4
CAPACITORS LM317LZ 2 IC1, 2
Disc ceramic, Philips 632 series MSA0304 2 IC3, 4 Avantek
1p8 2 C7, 24 BA244 2 D1, 2
5p6 2 C2, 18 1N4148 2 D3, 4
27p 4 C6, 11, 22, 27
100p 11 C9, 14, 25, 30, 33, 35-40 MISCELLANEOUS
1n 14 C1, 3, 4, 5, 10, 15, 16, 17, 19, 20, Crystals 101MHz, 1015MHz (HC33-U)
21, 26, 31, 32 Box Teko TEK S393 (West Hyde 064236)
Tantalum
10 16V 4 C12, 13, 28, 29
Trimmers
2-6p 4 C8, 23, 34, 41 (Farnell 148-157)
10 23
432MHz
10 24
A LOW-NOISE 432MHz
GaAsFET PREAMPLIFIER
by Ian White G3SEK
10 25
432MHz
GaAsFETs also have an undeserved bad loaded Q by using poor-quality trimmers for
reputation for vulnerability to RF damage and C1 and C2, or a lossy and unreliable input
voltage transients. In fact weve been care- connector. The circuit has been designed to
lessly subjecting our old valve and bipolar RF require only small tuning and coupling
stages to such insults for many years, and the capacitances, again in the interests of low
problem is that GaAsFETs wont tolerate that losses, so C1 and C2 must each have a low
kind of bad engineering. If any of the applied capacitance and a high Q. C1 can
DC or RF voltages exceed the breakdown limit typically be a miniature 0525pF PTFE
even for an instant its goodbye GaAsFET. trimmer, and C2 a 055pF glass or quartz
Adequate coaxial relays and proper trimmer although you can really go to town
sequencing of transmit-receive change-over with microwave sapphire trimmers if you
can totally prevent problems of RF burnout, want. To avoid degrading the noise figure,
and the problems of supply-voltage transients both of these capacitors must have solid low-
can equally be avoided by proper design. loss connections.
Chapter 11 gives full details of the power The input connector should be a type N,
supply and station-control features required to SMA or TNC; a BNC connector would soon
protect GaAsFET preamps so this really start to become unreliable in this critical
shouldnt be a problem. application. Another alternative is to elimi-
nate unnecessary coaxial connectors by
CONSTRUCTION building the protection relay (Chapter 11)
The low-loss input inductor L1 consists of an directly into the preamplifier.
86mm length of 025" diameter brass rod, The source resistor R1 will need to be
inside a cavity made from standard 22mm adjusted to set the manufacturers recom-
outside-diameter copper heating pipe. The mended drain current, in the region 10
cavity connects with a small screening box for 30mA. 47 is a good starting value, or you
the rest of the RF components, via a 2mm could use a good-quality 100 trimpot
hole through which the gate lead of the connected to the source bypass capacitor via a
GaAsFET protrudes. The leadless disk capaci- ferrite bead. R2 is a 1k safety resistor
tors C3 and C4 are soldered as close as attached to the opposite source lead to R1, to
possible to the hole, leaving just enough space prevent the source from floating while you
for the body of the GaAsFET (Fig. 10.14). change R1.
Particular care is required to minimize Electrically and mechanically, this preampli-
contact losses in the input cavity, especially at fier design can do justice to the best available
the joint between L1 and the disc which low-noise GaAsFETs. It should be adjusted for
forms the base. Ordinary solder is less effective lowest noise temperature using a G4COM-
than a knife-edge joint formed by type noise figure indicator with a well-
counterboring the end of L1 and drawing the matched 50 noise source (Chapter 12). For
joint up really tight with a steel screw. It is the best possible results, the preamplifier
more difficult to avoid solder in the joint should be given its final adjustments when
between the base disc and the copper tube, installed at the antenna with all its normal
though this is a somewhat less critical point. relays and cables; this can be done by inject-
For long-term performance, the whole ing the noise through another antenna close
assembly should be silver-plated [13]. For less by.
demanding tropo use, a quite satisfactory
input circuit can be made using a 22mm
capillary end-cap instead of the turned base
disc, with soldered joints throughout.
Having taken all that trouble with the cavity
itself, it would be a shame to spoil the un-
10 26
THE K2RIW POWER
AMPLIFIER
by John Nelson GW4FRX and Ian White G3SEK
Fig. 10.15.
Beneath the
lid of an
ARCOS
K2RIW
10 27
432MHz
has marked advantages over parallel on contacts to cause losses and erratic operation.
frequencies near the maximum at which these The grid compartment is almost equally bare,
techniques can be used. But with the strip-line, although the tuning and loading capacitors
the inductance can be made as low as desired, are more conventional.
simply by making the line wider. Any number Knadles original write-up gives the impres-
of tubes can be put in parallel and resonated, so sion that the K2RIW is very easy to build and
long as the intertube resonances (push-pull use, and also that it was a fully-developed
modes) can be controlled... design. However and putting this in the
In this amplifier the parallel grid and plate nicest possible way it isnt and wasnt, and
networks force the RF voltages on the two the K2 is seldom ...routinely duplicated in
tubes to be identical. If one tube has higher its original form despite Knadles use of
emission than the other, it may draw slightly precisely those words in the article. Although
higher current, but this is of little conse- the original was basically sound, there was
quence because it happens to be the condition room for improvement. But the K2 is a subtle
under which the amplifier will produce the and individual device, and some of the
greatest output with this pair of tubes. attempted variations on the original have
The first time you look inside a K2RIW produced mixed results. The basic problems
amplifier (Fig. 10.15), you might suspect that are in the tuned circuits and the cooling,
some portions of it are missing. From above, although by 1992 Varian-Eimac had decided
theres no sign of variable capacitors for that the latter wasnt such a problem after all.
tuning or loading; you only see a pair of valve The K2RIW variant described here was
anodes (with home-made chimneys, even) developed by the late Fred Merry,W2GN, and
and a flat sheet which you presume is the large numbers were sold in kit form by his
anode inductor. The capacitors for anode company (ARCOS) until Fred retired for the
tuning and loading are in there, however; second time at the age of 75! The ARCOS
they are flappers made from flexible phos- design benefited from considerable user
phor-bronze and operated by plastic fishing feedback, two extremely satisfied users being
line, and they live underneath the plate line W1SL and the editor of this book. Another
where you cant see them. There are no sliding variant which works well is produced by
Fig. 10.16. Circuit of the K2RIW 432MHz power amplifier. See Fig. 11.5 for additional screen-
grid protection components
VG2
INPUT FT
VDR OUTPUT
(C3) LOAD
C4
C2
RFC2
L2 L1
C5
C1
TUNE
RFC3 RFC1
FT
VDR
C7
FT FT FT FT
C6
10 28
K2RIW POWER AMPLIFIER
ANODE COMPARTMENT
8
ANODE LINE
5
3
3 1/2
4
AIR
2 5/8
2 1/4 x 9/16
12
1/2
all round 8 3/4
24 fixing HV
holes
3 1/4
2 5/8
1 1/8
2 1/8
3/4
2 9/16
2 9/16
Fig. 10.17. Dimensions of ARCOS K2RIW anode line and anode compartment (dimensions in
inches). The size of the large anode-line holes depends on the type of finger stock used
Gordon High Precision in the UK. sions originally given by K2RIW will not
The original circuit is shown in Fig. 10.16, resonate. Part of the problem was that the
and is about as simple as you can get. Note, original article did not specify minor dimen-
however, that extra screen-grid protection sions that are actually quite important. Consid-
components are required, as recommended in erable difficulties have also been experienced
Fig. 11.15. with anode lines made from double-sided PCB
Many builders have found that anode lines (as recommended by K2RIW) to the extent that
built as accurately as possible to the dimen- two apparently identical lines made from the
same material behaved quite differently one
Fig. 10.18. K2RIW anode compartment, with worked and the other wouldnt. However,
anode line removed to show the tuning and anode lines made from silver-plated solid brass
loading flappers [13] to the dimensions of Fig. 10.17 will just
tune up and go.
With the correct anode-line dimensions, the
dimensions of the tuning and loading flapper
capacitors shown in Fig. 10.17 should also be
correct. Fig. 10.17 also shows where these two
flappers are located, and Fig. 10.19 shows that
end of the anode compartment in close-up.
The flappers are made of springy beryllium
copper, and their natural rest positions are
parallel with the anode line; this represents
maximum capacitance. To tune and load the
amplifier, each flapper is pulled away by a
piece of plastic fishing-line wound around the
shafts of the tuning and loading controls.
There have been some problems in this area,
mainly with melting of the nylon line in the
intense RF field. One constructors solution
10 29
432MHz
Fig. 10.20.
Dimensions
of the K2RIW
grid line and
grid
compartment
(dimensions
in inches).
10 30
K2RIW POWER AMPLIFIER
10 31
432MHz
screen current can run away very quickly and K2RIW, and the anode line is shorter.
the result will be another flashover. RFC1 mounts vertically beside the anode
Many K2RIWs seem to require excessive line and connects to the HV feedthrough C7.
drive. At the supply voltages recommended in Flapper capacitors are made from thin,
Chapter 6 for 4CX250s, an output power of springy phosphor-bronze. C5 (anode tuning)
400W should require just 7W of drive. If yours is secured to the end wall by clamps or blocks,
requires more, its got a problem somewhere. with several screws. C4 (loading) may require
All in all, Richard T. Knadle Jrs brainchild support besides the soldered joint to the
isnt perfect, but if you want to use a pair of output socket (type N). The dimensions of the
4CX250 tetrodes at 432MHz its the only game tuning and loading flappers should be as
in town. If you make unwise substitutions for shown, though some experimentation with
critical dimensions and components, be the width of the loading flapper may be
prepared for a mighty struggle to get it to necessary. The rest positions of the flappers
behave. Follow the ARCOS dimensions given should be " below the anode line and parallel
here and itll run like a charm. 4CX250Rs, BMs to it.
and Bs all work well in the K2, as does the later The input socket (type N or BNC) is
8930 (electrically identical to the 4CX250R but mounted on the grid box, with a hole through
with a larger anode-radiator assembly, so the the rear drop of the base. For C1 and C2, use
anode dissipation goes up to 350W). However, miniature 05pF components with a low
forget the 4CX350 family they really dont minimum capacitance. C1 is mounted on an
want to know at this frequency. L-shaped metal bracket, and C2 on a piece of
insulating fibreglass board (Fig. 10.22). To
CONSTRUCTION NOTES achieve the correct reactance for C2, you may
The anode compartment is an inverted need to use a series capacitor (C3) and/or thin
chassis, 12" x 8" x 3", made of 16SWG wire leads connecting C2 to the input socket
(14AWG) aluminium. Corners should be and the grid line.
soldered, welded or fully overlapped with Heater, screen and control-grid supplies are
aluminium angle inside. The top plate is brought into the grid compartment via
secured to the flanged edge by at least 24 feedthrough capacitors. By rotating the valve
screws; use bushes and machine screws in bases and positioning the feedthroughs
preference to self-tappers if possible. For the appropriately, all wiring in the grid compart-
K2RIW cooling arrangement, the anode box ment can be eliminated (Fig. 10.22). A
must be airtight. V275LA40B VDR is soldered directly to the
The base for the amplifier is a similar chassis screen-grid pin of each valve base. RFC3 is
the right way up, with a base plate, and the mounted vertically from the edge of the grid
top cut out to leave a " flange all round. The line and connects to a feedthrough capacitor
grid compartment is a 5" x 7" x 2" box with soldered to the flange of one valve base; this
its own base-plate. in turn connects directly to the feedthrough
The anode line is made from solid 16SWG capacitor in the box wall.
brass, preferably silver-plated, and supported When using the K2RIW cooling arrange-
on 1" ceramic or PTFE standoffs in the ment, the size of the exit vents for the grid
positions shown. Drill pilot holes for the airflow is critical. One vent should be in the
valves in the anode line and use it as a end of the grid box furthest from the valve
template to drill accurately aligned holes bases, with another identical opening in the
through the top cover, the base of the anode end wall of the base directly opposite. Fig.
compartment and the grid box, having first 10.22 (extreme left) shows a pattern of 90 holes
bolted the two boxes together. Note that the of 1/8" diameter a tedious job but the result
spacing between the valves shown in Figs will be well-cooled valves. The blower inlet
10.17 and 10.20 is smaller than in the original should be the largest possible screened hole in
10 32
K2RIW POWER AMPLIFIER
10 33
432MHz
ANTENNAS FOR
432MHz DX
DL6WU long Yagis are hard to beat for noted above, the particularly favourable
432MHz DX. There are a number of compu- lengths may produce a little more gain.
ter-optimized designs which you could Element spacings are quoted to the nearest
consider [14-17], but most are only optimized 01mm, but are nowhere near as critical in
for specific boom lengths and numbers of practice. By making all measurements from
elements. The DL6WU family of long Yagis, one datum point the position of the reflec-
on the other hand, can be made for 432MHz tor you should easily be able to achieve an
in boom lengths starting at 28m and increas- accuracy of a millimetre or two along the
ing in steps of about 278mm up to any length entire length of the boom. Square-section
you like, and their performance will be very booms are recommended for mechanical
close to the optimum achievable for your convenience, although round booms have a
chosen boom length. An example is shown in better ratio of strength to windload. The
Fig. 8.28, which also demonstrates that the longer booms will require support and
same basic designs, construction and method strengthening against sag and side-winds, but
of feeding can be used from 144MHz up to at this must not affect the electrical behaviour of
least 13GHz. the elements.
The table on page 7-24 shows the possible Element lengths are given for a number of
boom lengths in wavelengths, and the designs element diameters, boom thicknesses and
with 14/15, 19/20, 24/25, 30/31 elements and mounting methods. If you need a different
so on are particularly favourable. These Yagis combination, you can easily run the computer
have naturally good front/back ratios and program for yourself, and this facility is often
potentially higher gain in the forward lobe. provided on the RSGB VHF Committee stand
Such generalizations apply to DL6WU-type at VHF conventions. Element mounting
Yagis only; other types of long Yagis will show methods can be the same as those shown in
similar behaviour, but possibly at different Fig. 8.30, and the appropriate length correc-
boom lengths and numbers of elements. If tions must be applied as shown in the table.
one of these designs suits your requirements Having made the corrections, it is quite safe to
for gain and overall boom length, use it; round the element lengths to the nearest
otherwise, go for the longest Yagi you can whole millimetre. K1FO has reported notice-
possibly manage. able effects on the resonant frequencies of
The table opposite gives more detailed elements due to chamfering the ends [15], and
dimensions for 432MHz DL6WU Yagis up to a 051mm 45 chamfer might be beneficial in
30 elements. The data were generated using a reducing the LF detuning which affects all
computer program [18] written by KY4Z/6, Yagis in wet weather. The fingers of the
W6NBI and G3SEK, which has been validated recommended Heyco nylon bushes are too
by DL6WU. The table also shows the esti- long to fit inside a 15mm square boom, so it is
mated gains and recommended stacking necessary to cut two opposite fingers off each
distances at intervals of five elements; you can bush.
judge the intermediate values for yourself. As The driven element should be a folded
10 34
ANTENNAS FOR 432MHZ DX
dipole about 335mm long overall; within half-wave 4:1 balun (Fig. 7-17), which will
reason, conductor diameter and separation are generally give a good match to 50. As
not critical. The dipole should be fed with a explained in Chapter 7, the match can be
improved if necessary by making small
alterations to the length of the driven element
DIMENSIONS FOR 432MHz and/or the first director. The boom passes
DL6WU YAGIS between the limbs of the folded dipole but is
not connected to it since this may cause
Element Spacing Length Estimated Stacking* asymmetric excitation; hence the size of the
(mm) (mm) gain (dBd) (mm)
boom has no effect on the dimensions of the
Element diameter Vert Horiz folded dipole.
3 1
/16" 40mm /8" The losses in a half-wave balun can be quite
Refl Zero 3404 3406 3409 appreciable at 432MHz unless the balun loop
Driven 1388 see text is made from low-loss coax. Since URM67 and
D1 1908 3008 3027 3053 RG213 are difficult to form into the necessary
D2 3158 2967 2991 3023
small loop, a thinner PTFE hard-line such as
D3 4650 2929 2956 2989
D4 6384 2894 2922 2956 UT141 (RG402) is recommended. The correct
D5 8328 2863 2891 2927 length for UT141 is 233mm, measured over
D6 10409 2836 2865 2902 the shielding, plus a 5mm length of dielectric
D7 12595 2813 2842 2880
D8 14886 2792 2822 2861 exposed at each end [16]. The shield of the
D9 17280 2774 2805 2844 main feedline can be grounded symmetrically
D10 19778 2757 2789 2829 to the boom at the balun if desired.
D11 22380 2742 2774 2815 Weatherproofing and mechanical stability
D12 25087 2729 2761 2803
D13 27863 2716 2749 2791 139 1250 1300 can be problems with the construction of Yagi
D14 30639 2704 2738 2780 feedpoints. Although some constructors seem
D15 33414 2694 2728 2770 to manage with a balun and coaxial connector
D16 36190 2683 2718 2761
D17 38966 2674 2709 2752
out in the open [1416] it is better to use some
D18 41742 2665 2700 2744 153 1450 1500 form of plastic enclosure for the connections
D19 44518 2656 2692 2736 between the driven element, balun and
D20 47294 2648 2684 2729 feedline. These are often available from the
D21 50070 2641 2677 2722
D22 52845 2633 2670 2715 smaller antenna manufacturers who sell spare
D23 55621 2626 2663 2709 163 1650 1700 parts, and for waterproofing they can be filled
D24 58397 2620 2657 2703 with closed-cell aerosol foam which is
D25 61173 2613 2650 2697
available from boating stores. If the antenna is
D26 63949 2607 2645 2691
D27 66725 2601 2639 2686 fitted with a coaxial connector, this may be
D28 69501 2595 2633 2680 170 1850 1900 another point of moisture ingress; in many
D29 72276 2590 2628 2675 installations it is better to dispense with the
D30 75052 2585 2623 2670
* for maximum gain connector and bring the cable straight into
Element mounting methods the waterproofed junction box.
(1) Insulating boom, or elements mounted on insulators at least 1 boom When mounting a Yagi for 432MHz, try to
radius clear of boom use the above element lengths keep the mast and any supporting struts as
(2) Elements mounted in full electrical contact through a square boom,
through the middle of a round boom, or across the top of a square close as possible to the centre-line of the
boom boom to avoid distorting the pattern. Do not
for a 254mm boom, add 156mm to the above element lengths use a mast diameter greater than 40mm (15")
for a 15mm boom, add 66mm to the above element lengths
(3) Elements mounted on insulating bushes through the middle of a round
and do not use a clamp which stands the
or square boom boom a long way off the side of the pole. And
for a 254mm boom, add 78mm to the above element lengths having taken so much care with the metal-
for a 15mm boom, add 33mm to the above element lengths work, dont ruin it all by asymmetrical
positioning of the coaxial cable.
10 35
432MHz
10 36
C H A P T E R 11
POWER
SUPPLIES
&
CONTROL
UNITS
11 1
POWER SUPPLIES & CONTROL UNITS
POWER SUPPLIES
& CONTROL UNITS
by John Nelson GW4FRX
11 2
ANODE SUPPLIES
11 3
POWER SUPPLIES & CONTROL UNITS
resulting composite HV capacitor isnt more electrolytics in the stack, with the smelly
particularly critical and anything above about and tiresome consequences already discussed.
30F will do. What working voltage do we New good-quality wirewound or 2W carbon-
need? Remember that in a capacitor-input film or metal-film resistors for this job dont
filter the peak voltage on the reservoir capaci- cost much why tempt Fate?
tor is 1414 times the transformer secondary Even though the equalizing resistor chain
voltage so a 01800V AC secondary, for will act as a bleeder resistor for the HV reservoir
example, will apply no less than 2545 volts to capacitors, and so will the resistors across the
the capacitor. A margin of at least 25% is rectifier diodes, its still best to play safe and fit
required for safety, so you need the working a back-up of about 1M across the entire
voltages of the capacitors to add up to at least capacitor stack. Since a 1M resistor will
3200V. This means that youre looking for dissipate about 5W at this high voltage, and
something like eight identical capacitors of normal resistors will flash over at more than a
more than 240F and at least 400V working. few hundred volts, you need to use a chain of
Anything else we need to specify about the resistors in fact a near-duplicate of the chain
reservoir capacitor? Yes, ripple current an of equalizing resistors. If youre using a single
important point, and one neglected by several large HV capacitor, pay even more attention to
commercial valve-amplifier manufacturers. the bleeder chain and connect it directly across
The ripple-current rating of the capacitor the capacitor, independently from any other
should be about 13 times the DC current to wiring. An isolated, fully-charged paper
be drawn from the supply, ie at least 650mA capacitor can remain lethal for weeks...
for a pair of 4CXs. Inadequate ripple-current
rating is a major reason why amateur HV CIRCUITRY
capacitor stacks blow up theres a lot of Fig. 11.1 is a typical circuit for a HV power
internal heating which culminates in a loud supply. The precise details will depend on
explosion, fountains of white stuff everywhere what kind of transformer you can find, but all
and one of the most revolting smells on earth. the rest of the circuit is pretty standard. Ill
So a safety factor of at least 100% in ripple- concentrate on features that help ensure a
current rating is no bad idea. long and happy life for the two most impor-
Bearing in mind that youre after reliability tant items the HV transformer and yourself.
and safety with no drama, the HV reservoir- In HV supplies, fuses cannot give perfect
capacitor stack is another area where Id protection all they can do is to confer some
strongly recommend you to grit your teeth, degree of safety. At the end of the day the
smash the piggy-bank and buy new compo- rectifiers, reservoir capacitors and all the other
nents. Try the STC Electronic Services catalogue bits and pieces can be replaced if the HV
for some excellent modern specimens; eight power supply throws a major wobbler, but if
330F 400V capacitors cost about 60 in 1992, you lose the HV transformer youre in for a
and ought to last for the next 50,000 hours or prolonged period of depression and mourn-
so. Their ripple current rating is 17A at 85C, ing. No-one will rewind it and you dont have
which is ample. To equalize the voltage which a spare; to make matters worse, the new one
appears across each electrolytic, so that each you get after several months of hard searching
one bears its proper burden and no more, each wont fit in the existing box, so the entire
capacitor requires a resistor connected in supply will have to be rebuilt. Meanwhile,
parallel with it. The value should be about youve missed the Perseids, the autumn tropo
100k (not critical, so long as theyre all the and the only aurora this year. So do your best
same) and the wattage rating should be as high to protect everything else in the HV box but
as you can fit in. The reason for the conserva- be totally neurotic about protecting the HV
tive rating is that failure of an equalizing transformer.
resistor will probably involve failure of one or Protect the primary against overvoltages
11 4
ANODE SUPPLIES
COMPONENTS
B1 Bridge rectifier, consisting of 20
1N5408 diodes (1000V PIV, 3A), each
wired in parallel with a 680k 2W
resistor and a 10nF 500V ceramic
capacitor.
C1 10,000F 25V electrolytic
C2-C9 330F 400V electrolytic capacitors,
ripple-current rating 1A or more
D1, D4 1N4007, 1N5408
D2, D3 BYX97-1600 or similar diodes with
40-50A current rating (or a cross-
connected 20-25A packaged bridge
rectifier)
FL1 Mains filter (optional), 15A rating
FS1 5A mains fuse
FS2, FS3 1A High Rupture-Capacity fuses
NE1 240V AC neon indicator
R1 2k2 25W
R2 500R 50W
R3 R10 100k 2W, wired in parallel with C2
C9
R11 68R 50100W
R12 R20 22k 12W
R21 1k0 2W (2000V reads 10V on the HV
meter)
R22 1R 250W, eg five 4R7 50W in parallel
(see text)
R23 Multiplying resistor for a 05V meter,
to read 05A anode current
RLA 24V DC, 25A switching
S1 DPDT switch (or 15A contact breaker)
SK1 PET100 or similar coaxial HV socket
(not an RF type!)
T1 HV transformer (see text)
VDR1, VDR2 V250LA40B or similar VDR (250V RMS
wkg, 140J capacity)
ZD1 15V 10W
Fig. 11.1. Circuit of a 2kV HV supply with soft-start and bridge rectifier. Current is measured
using the 1 resistor R22 which connects the HV-minus rail to ground. The bleeder resistor
chain R12R21 acts as a 1/200 voltage divider for the HV meter, and can also be used to
provide an HV present logic signal
and nasty spikes by mains filtering and a primary of the mains transformer with
suitably-rated VDR, which will keep the something suitable. A soft-start circuit will
rectifiers happy as well. Incidentally, does the avoid the heavy initial current surge, and will
shack light go dim and the transformer go allow you to fit a standard fuse (not anti-
thunnnnnng when you hit the switch? That surge) of around 5A rating. That will give
awful thump is associated with a fair amount some protection to the components upstream
of mechanical and electrical stress on the of the transformer if, say, the primary goes
secondarys insulation. Its much better to use short-circuit to the core. Regrettably, this
a soft-start system like the one in Fig. 11.1, so failure mode isnt uncommon in elderly HV
you just hear a purposeful hum which transformers which have had a hard life in the
increases in amplitude as the high voltage hands of their previous owners. Next, fit a fuse
gently comes up. in each leg of the rectifier stack (Fig. 11.1).
Fuses, as I was saying... First of all, fuse the They will protect the transformer secondary if
11 5
POWER SUPPLIES & CONTROL UNITS
the rectifier stack decides its tired of life, or if circuits and all manner of assorted abuse and
the reservoir capacitor takes it into its head to all thats ever needed replacement have been
go dead short-circuit which block paper the two fuses.
capacitors sometimes do, unfortunately. These While were pondering the HV supply, how
fuses must be of the high rupture-capacity are you going to measure the anode current of
(HRC) type, with a ceramic body which is the valves? Bit tricky, this. For a start, youll
filled with sand or something similar. The have to forget about trying to measure the
older 125" fuse is best for this particular separate anode currents of a two-valve ampli-
application, though a modern 20mm UK fier directly because both anodes are hooked
mains fuse will just about manage. DO NOT straight to the tank circuit. Nor can you get at
try to use a glass-bodied fuse in this position. the cathodes, because theyre both solidly
If you do, and theres a short-circuit fault, two grounded at the valve base and they have to
things will probably happen. The first will be be. Dont even think about trying to use
a beautiful whitish-blue arc inside the fuse; bypass capacitors here. I suppose you could
this arc may drop a few tens of volts, but it hang a 0500mA meter in series with the
certainly wont be the open circuit that all the output of the HV supply, if you didnt mind
hard-pressed components are begging for. The its internal metalwork floating about at a
arc will last for a second or so before the fuse couple of kV . On second thoughts well
explodes, spraying fine glass particles every- forget that bright idea too...
where. By that time, all the rectifiers in the The right way to meter the anode current is
stack will probably have resigned in protest, to catch it where its close to ground potential,
possibly taking with them a few equalizing ie in the negative return of the HV supply.
resistors and even some of the reservoir This is very easy. Instead of grounding the
capacitors. At this point the phone will ring negative side of the HV rectifier and the
and someone will tell you that it looks as reservoir capacitor, you return them to an HV-
though tropo is opening up to HB9 and Italy. minus rail (Fig. 11.1). Then connect a low-
Such a pity that its just turned 6pm on value resistor between the HV-minus rail and
Saturday evening and youve got no spares... the common chassis ground. 10 is a con-
Theres not a lot of advantage in fitting a venient value for a legal-limit 4CX pair,
fuse in the DC output of the supply itself because when you connect a 005V meter
well, thats not strictly true but, amongst across this resistor, youre measuring anode
other things, youd need a rather specialized current on a scale of 0500mA.
high-voltage HRC item which is difficult to By keeping the metering circuitry down at
get. Standard 250V AC cartridge-type HRC ground potential, youve made everything
fuses in this position may work or they may much safer. You can now move the meter
just arc and explode under short-circuit itself into the control unit on the operating
conditions; they work well enough in the HV desk, leaving the HV PSU on the floor, and
rectifier stack because of the quasi-AC wave- you wont need special HV cable trailing
form, but straight DC is usually too much for around the place. Very convenient, very safe
them. The best practical solution is to fit a arent we clever? Yes, fine if everything works
large wirewound resistor in the DC output to properly but there could be a problem.
limit the worst-case short-circuit current until Specifically, suppose theres a dead short from
the other fuses blow. For a number of years the HV line to chassis suppose, for instance,
my standby 25kV supply has used a 68 that an HV decoupling capacitor fails, or
100W resistor in the output, with a 1A anti- somebody drops a screwdriver into an open
surge ceramic HRC fuse at each end of the anode compartment (yes, it has happened). In
transformer secondary as I mentioned earlier. the basic supply weve been describing, a
This much-loved box has survived countless phenomenally high current will promptly
deliberate and accidental flashovers, short- flow through the 1 resistor which will
11 6
ANODE SUPPLIES
probably not take kindly to this onslaught mention generating the HV for working a fair
and may well opt out of existence in a more amount of assorted DX and the occasional
or less spectacular fashion. If that resistor contest. The supply is still there, and so am I.
blows before the HV fuses, or if there are no By the way, the dangers youre trying to
HV fuses to blow, the HV-minus rail will fly avoid could also occur if the HV-minus lead
off to 2kV. Somehow I dont think either the drops off the resistor and diodes, while still
meter or a large portion of the associated remaining connected to the meter. To prevent
wiring will survive that, and you might not this possibility in the FRX supplies, the meter
live to fight another day either, if you touch and the HV-minus rail are connected to
the wrong things in the wrong order. opposite ends of the 1 busbar. And just in
Come hell or high voltage, the HV-minus case it really isnt your day, locate the series
rail must stay close to ground potential. You resistor for the anode-current voltmeter
can ensure this by several means. One is to inside the HV supply. If all else fails, this will
connect a couple of big silicon diodes in make sure that high currents at HV-minus
inverse parallel across the 1 metering cant get out into the meter extension wiring.
resistor, as shown in Fig. 11.1. By big in this Bearing in mind Nelsons Rule If you can
context, I mean capable of carrying the full lift the HV transformer, it isnt big enough
prospective surge-current of a dead short- its as well to make the HV PSU and the
circuit. This is why I use a 68 surge-limiting amplifier as separate entities. So the next
resistor in the HV side; it limits the worst-case problem is: how are you going to get 2kV
prospective current to somewhere around safely to the amplifier? The best high-voltage
40A, which meaty diodes can easily handle. connectors easily available in the UK are the
With these two components across the 1 PET 100 series, which are of screw-on coaxial
resistor, the voltage across it can never exceed construction and can often be found in scrap
1V or so. The meter wont worry about this nucleonic gear as well as in the component
mild overload, since a 05V FSD meter will catalogues. For high-voltage cable the easiest
cope with 1V for ever. solution is ordinary UR43 coax, which has an
However beefy your diodes, the 1 meter- adequate DC rating for the job and goes
ing resistor must still be incredibly reliable straight into a PET 100 plug. This way every-
and extremely conservatively rated. If you saw thing is nice and safe, and any cable or
the 1 resistor in the main GW4FRX HV connector fault is most likely to lead to the
supply youd probably laugh your head off. HV going short-circuit to ground and blowing
Its made up of five separate chassis-mounting fuses rather than doing any personal damage.
5 100W Mil-spec Welwyn components in After all, your HV supply can cope with dead
parallel, with one end of each taken individu- shorts to ground now, cant it?
ally to chassis ground with a large bolt and Lets now suppose that youve solved all
solder-tag, extra-nasty shakeproof washers, these problems and built the anode HV
and locknuts to finish the job. Theres a supply. You finally managed to obtain a really
common 14SWG ground busbar with its own superb transformer, which is fused in all the
separate ground connection as well, just to right places and which has nice soft-starting,
make sure. One BYX97-1600 and one BYX97- mains filtering, VDRs and a duvet to keep it
1600R (DO-5 stud-anode and stud-cathode warm at night. Youve put together a tasty
diodes, rated 47A average, 800A surge 50p rectifier stack and spent some folding stuff on
each from a rally) are paralleled across the a decent reservoir capacitor and bleeder
resistor. This probably sounds amazingly Over resistors. Youve built it well, it seems to work
The Top. But everything still works, despite a treat and youre now basking in the rosy
the supply having been used for several years glow that comes from knowing youve done
of tests involving all sorts of abuse and some the job properly. Well done Im proud of
extravagantly violent flashovers not to you. Now answer me this
11 7
POWER SUPPLIES & CONTROL UNITS
11 8
SCREEN SUPPLIES
diodes; or an active shunt-regulator with regulator chain. The best way to prevent this
valves or transistors. from taking out all the devices at once is to
Unfortunately, almost every published use a VDR across the output of the supply, in
screen supply for 4CX tetrodes harks straight addition to the VDR between the screen pin of
back to the Days of Old when all amplifiers the valve-base and ground.
ran in Class C for AM, FM or CW. In those Another highly desirable feature for any VR-
days, of course, VHF/UHF SSB was a gleam in tube or Zener screen supply is a large electro-
the eyes of a few pioneering stalwarts. The lytic across the output. As well as helping to
thing about SSB, of course, is that it requires take the sting out of a flashover, a capacitor of
the power amplifier to run in Class AB and about 50F will dramatically improve the
in the face of the much more demanding dynamic screen regulation at higher audio
requirements of Class-AB service, those old- frequencies, and hence the intermodulation
fashioned screen supplies just dont cut it. performance. Even if youre not prepared to
Lets start at the beginning and examine change your old-fashioned screen supply for
why this is so. Voltage-regulator tubes or something better, the rest of us on the band
Zeners look simple and attractive, but most will be grateful if you add this capacitor as a
published circuits are thoroughly inadequate. brute-force method of improving its dynamic
Neon stabilizer tubes such as the OA2 and regulation.
OB2 seem to perform quite well, but you need The next step towards state-of-the-art
four in series-parallel for each valve in order to intermod performance from your 4CXs is
sink the necessary 50mA of standing current active screen regulation. One very well-known
through them. Youd therefore need eight circuit by Andy Wade G4AJW [2] featured two
tubes for Class AB1 and some fancy switching separate supplies, each consisting of an op-
for Class C and how are you going to vary amp driving a valve shunt-stabilizer. It was
one screen voltage slightly to balance the two widely duplicated, and although purists would
valves? argue that the EL84 used was somewhat
Zener diodes look more promising, since overrun, the G4AJW design will outperform
you can adjust the number of components in anything using VR tubes or Zeners. Another
each separate Zener chain to set up the right active supply, this time using transistors
screen voltage and standing anode current for instead of valves, was published by myself in
each valve. But you still need to send about Short Wave Magazine; however, a later devel-
50mA of standing current down each chain, opment of this appeared in Radio Communica-
so you need to use 10W Zeners and a big heat- tion for December 1987 and January 1988 as
sink. Suppose you decide to use four 82V part of a comprehensive screen and control-
devices per chain together with a few smaller grid supply complete with control logic. Well
ones to balance the standing currents messy, examine this power supply and control unit
but possible. The slope resistance of these (PSCU) in detail at various points in this
diodes will be a few ohms apiece, and with a chapter.
current variation of about 50mA the dynamic Meanwhile heres a simpler screen-regulator
regulation wont be brilliant. With care in circuit designed specifically for upgrading
selecting your Zeners and a lot of trial and existing power supplies.
error you can just about manage 05V regula-
tion. ACTIVE SCREEN REGULATOR
Remember that VR tubes and Zener diodes Figure 11.2 shows an active screen-regulator
are both extremely noisy and will require for a single 4CX. It was developed by Melvyn
copious bypassing (but be careful not to turn a Noakes G4JZQ, initially as a retrofit item for
chain of VR tubes into a relaxation oscillator). early Heatherlite amplifiers which had no
Remember also that if you have a flashover, screen-supply regulation at all! This little gem
an awful lot of kV will appear at the top of the uses power MOSFETs for both the current
11 9
POWER SUPPLIES & CONTROL UNITS
Fig. 11.2. Screen voltage regulator for a source and the shunt stabilizer, and its
single 4CX250/350, by G4JZQ. Relay performance for Class AB1 is as follows:
switching is provided between output
Output voltage range: 350V 10V
voltages suitable for Class C and Class AB1,
and each voltage is adjustable. Large dots Max source current: 30mA
show interfaces to a PC board Max sink current: 30mA
Output short-circuit current: 52mA approx.
COMPONENTS (with load resistor, 66mA approx)
D1 1N4003 etc
Ripple rejection: >80dB
D2, D3 1N914, 1N4148 etc Output noise: <05mV pk-pk
DB1 2KBB100R (1000V PIV 2A bridge)
Regulation at 30mA: better than 2mV
IC1 LM748
IC2 LM317L
RLA BT47-style 12V DC (energize for The requirements upstream of the regulator
Class AB1)
are a DC supply of about 420V from a suitable
TR1, TR2 IRF840P (500V 8A 125W or greater),
with heat sinks transformer, rectifier and a reservoir capacitor
TR3 BC107 of 16F or greater. Further smoothing is not
VDR1, VDR2, VDR3 V275LA40B (VDR, 275V RMS wkg,
140J capacity), 3 in parallel
required; the regulator itself takes care of that.
One option is to connect an overcurrent trip
11 10
SCREEN SUPPLIES
relay between the 420V supply and the floating way above ground potential, these
regulator; for more details, see the description bleeder resistors will make a conventional 0
of the comprehensive PSCU later in this 20mA meter display positive and negative
chapter. screen current on a scale of 10010mA per
The upper part of the regulator circuit (Fig. valve. This is just what you need, because
11.2) is a constant-current source. A constant positive or negative screen-current can then
45mA flows through TR1, and in the absence be read without the necessity for switching;
of screen current this passes to ground more on this subject later. Yet another option
through TR2 and the two 15V Zener diodes. is to make the supply switchable between
These two diodes also provide a 30V supply 350V for Class AB1 and 250V for Class C,
for the control circuitry. One input of op-amp although youll then need to have two scales
IC1 receives a sample of the screen voltage on the screen current meter because the
from the voltage divider; the other input of bleeder resistors will draw different currents
IC1 receives a stabilized reference voltage from and offset the zero.
IC2. The circuit, apart from the power MOSFETs,
To analyse the circuit, ask yourself: What can be built on a conventional single-sided
happens if the screen voltage starts to fall? If PCB [3] or on Veroboard, taking care to avoid
that happens, the voltage at the + input of high voltages between adjacent tracks. It is
IC1 falls, causing the output of IC1 to fall by a imperative that the MOSFETs are mounted on
much greater amount. This will tend to turn an adequate heatsink failure to do so will
off TR2, which will therefore allow the screen destroy them.
voltage to rise to its correct value. This Setting-up goes like this. Initially disable
regulating mechanism will operate until TR2 TR3 by adjusting the slider of RV3 to ground.
is fully cut off, allowing the full 45mA from Remove the link from the constant-current
TR1 to flow into the screen grid as positive source to the rest of the circuit and connect a
screen current; beyond that point, regulation milliammeter across the pins; adjust RV1 for
is lost and the screen-voltage will fall rapidly. 45mA indicated current. If the two 68k
Negative screen current is sunk through output load resistors are fitted, set the stand-
TR2, which would be turned on more and ing current to 45 + 10 = 55mA. (Do not then
more by IC1 to maintain a constant screen operate the regulator without the bleeder
voltage. To prevent excessive power dissipa- resistor, because the extra current would then
tion in TR2, an overcurrent circuit is provided: flow through TR2 and would add to its power
TR3 looks at a sample of the current being dissipation.) Next adjust RV2 to set the output
sunk though TR2, and deliberately throws the voltage to 350V for Class AB1. To adjust TR3
circuit out of regulation if the current is which prevents excessive sink current, briefly
excessive. This is not only to protect the increase the current through TR2 to 65mA
regulator itself: as you know from Chapter 6, using RV1, and adjust RV3 until the output
the valve shouldnt be running excessive voltage just starts to rise above its regulated
positive or negative screen current, so the loss value; then return RV1 to its proper setting.
of screen-voltage regulation can be used to In principle this circuit could be used for
generate an alarm signal. large single tetrodes if bigger MOSFETs are
VDR protection is provided against fitted and a few resistors changed it could
flashovers, in addition to the VDR connected easily do the business for a 4CX1000/1500 or
directly from screen to ground in the ampli- a 7213/DOD006, for instance.
fier. A further option is to fit a bleeder resistor
directly from screen to ground, to sink a SCREEN SUPPLY FOR A PSCU
permanent current of 10mA; two 68k 2W Unlike the stand-alone regulator Ive just
resistors will do the job for a 350V supply. As described, the screen supply in Fig. 11.3 forms
well as preventing the screen grid from ever part of a comprehensive power supply and
11 11
POWER SUPPLIES & CONTROL UNITS
11 12
SCREEN SUPPLIES
11 13
POWER SUPPLIES & CONTROL UNITS
11 14
CONTROL-GRID SUPPLIES
amp and drive flat-out into a dummy load. quite got total control of the anode current via
Then switch to receive, remove the load and the control grid near cutoff. To turn a
apply full drive again... 4CX250B right off you not only have to apply
One final thought is that the little plastic- a large negative voltage to the control grid,
cased meters which most of us use for grid- but you also have to take the positive volts off
and screen-current metering are not really the screen (which means you have to firmly
designed to be floated at high voltage above a ground it remember?). Only then will the
grounded metal front panel. There isnt much anode current fall to zero and stay there.
insulation between the scale plate and the Since the control-grid supply doesnt have
panel, and the insulation on the internal wires to source any current, a simple variable shunt-
is very thin. Be particularly careful if youre stabilizer will do nicely. You can reckon that a
using a meter with internal lighting youre 250B with 2kV on the anode and 350V on
bringing a ground lead right inside the meter. the screen will need about 55V to set a
If you want to use a small plastic-cased meter standing current of 100mA for Class-AB1 SSB
in any application where the internal work- use. Somewhere around 100V will do both
ings are at more than about 50V above (or for Class C and to cut off the valves on
below) ground, either dont use the internal receive. As long as the circuit will give you
lighting or replace the relevant wires with about 15V variation either side of those
something more substantial. There isnt much ballpark figures, you should be all right.
room between the movement and the casing, What sort of circuitry? Actually, you can get
so use thin PTFE-covered wire and avoid away with nothing more elaborate than a
scraping the insulation when you put it all transformer-rectifier-smoothing circuit to
back together. provide something like 150V, in conjunction
So thats about it for the screen supply. with a switchable potential divider including a
Obviously the choice you make will depend variable potentiometer. Hundreds of amplifi-
on what you want and how state-of-the-art ers have been built with grid supplies no more
you want to be. The main point of discussing complicated than that, and they work just
it at considerable length has been to convince fine. The only snag is that the working point
you that 4CX amplifiers in linear service really of the amplifier will move about when youre
do need better screen supplies than the vast using Class C which, to be honest, doesnt
majority of published circuits, and then to really matter. Its better to stabilize the supply,
show you what the options are. though, because otherwise the working point
of the amplifier in Class AB1 wanders around
CONTROL-GRID SUPPLY a fair bit as the mains wobbles about (or the
Happily theres a lot less to the control-grid generator voltage falls when your contest
supply (or grid bias as old-timers tend to call colleagues put the kettle on) so the power and
it) than there is to the screen supply. Apart intermod performance vary slightly. Here
from supplying a constant bias voltage at zero again, the G4AJW circuit [2] is fine, although
current for Class AB1, the control-grid supply its a bit old-tech now you could replace the
only needs to sink current (not source it) VR tube by a 150V Zener, for example.
when you over-drive into Class AB2, or when A slightly more elaborate Nelson-Noakes
you switch to a different voltage for Class C. circuit (Fig. 11.6) uses an op-amp and shunt
The voltages should of course be adjustable so stabilizer. The reason wasnt so much a desire
that you can set the standing anode current. for the last word in stability but a wish to
The control-grid supply also needs to eliminate the TX/RX changeover relay and so
produce a large negative voltage on receive, to remove any chance of the control grid ever
cut the valve off. Now heres another mild being allowed to float. In this design the grid
quirk of the 4CX250 family: unlike almost voltage is electronically commanded up and
every other tetrode ever made, you havent down between its transmit and receive
11 15
POWER SUPPLIES & CONTROL UNITS
11 16
HEATER SUPPLIES
11 17
POWER SUPPLIES & CONTROL UNITS
reduced in 01V steps. At some point a marked use a 6V overwind on a multi-purpose trans-
drop in output power (or, if two-tone drive is former with a large core.
used, a very pronounced deterioration in So some form of controlled soft-start system
intermodulation performance) will be ob- is a sensible addition to a heater supply, and
served. If the heater voltage is then increased this probably applies to other valves as well as
by 02V from that point, that corresponds to the 4CX series. Its fairly simple to incorporate
the optimum heater voltage for the valve in suitable soft-start circuitry into an electronic
that class of service. heater-voltage regulator system such as the
With the 4CX250B at frequencies between one described here.
120MHz and 300MHz the optimum voltage at In other words, theres actually a great deal
maximum ratings would usually be around to be said about valve heaters. Its not so
57V, and between 300MHz and 500MHz it much from the point of view of performance,
was about 55V. Maintaining that voltage but more in order to maximize the life of what
accurately and stably would lead to approxi- are when alls said and done very expen-
mately a threefold increase in life. The implica- sive devices. When we first published this
tion for amateur use is that one valve or pair of design [6] for a stabilized heater supply,
valves purchased or obtained from new would G4JZQ and I received a good deal of incredu-
probably last longer than the owner. lous comment from people who couldnt for
Another important factor when considering the life of them see the point of it. But if by
valve heater supplies is the surge current which spending about 20 you can triple the life of
flows in the heater when you initially switch other components costing anything up to
on. Like the filament of a lamp bulb, the heater 200, isnt that a bargain?
of a thermionic valve has considerably less Anyway, heres the design we came up with
resistance when it is cold than when it has and if you still cant see why we bothered,
reached normal operating temperature. If the skip this bit. The heater current of a 4CX250 is
source impedance of the heater supply is very specified as 26A at 6V, and the corresponding
low, the initial current flowing in the heater at figure for a 4CX350A is 29A. For a two-valve
switch-on will be several times its normal amplifier, this implies in round numbers a
running value. The resultant thermal cycling total heater current of some 6A at 6V. Assum-
and overload places considerable stresses on ing that a suitably rated 63V secondary
the heater. To obtain some idea of the magni- winding is available for use in the heater
tude of the problem, the heater of a 4CX250B supply, the design task comes down to
was fed for test purposes from a suitably deriving a 6V 6A stabilized DC supply from it.
metered 6V 30A DC supply. At switch-on the In the final design the value of output voltage
initial surge current took the supply into was to be finely adjustable, to permit use of
current limiting; the current being drawn was the running-in procedure outlined above for
still 11A five seconds after switch-on. Valves of extending valve life. Overvoltage and overload
the 4CX family are built to take this kind of protection were to be included, as was a soft-
treatment, but its reasonable to infer that it start system and optional remote sensing of
does nothing to lengthen valve life. Many 6V the voltage right at the heater pins.
transformer secondaries are quite able to The main problem in getting 60V DC from
supply very high peak currents for a short time, a 63V AC supply is that the trough of the
and this is probably a good reason not to use a output ripple voltage of the smoothed supply
generously-rated heater transformer. For many will be only about 62V at best. This is largely
of its small triodes and tetrodes (although not, due to the voltage drop in the rectifiers
as it happens, the 4CX250 family) Varian- which will be significant at 6A, even if you use
Eimac recommends a separate heater trans- Schottky devices. Conventional IC voltage
former which will saturate and limit the regulators for high current involve a voltage
switch-on surge. In particular, you shouldnt drop of at least 18V, so theyre out. Even
11 18
Fig. 11.7. Heater regulator
for the comprehensive PSCU.
See reference [6] for full
details.
11 19
POWER SUPPLIES & CONTROL UNITS
discrete regulators using low VCE(sat) bipolar some 4V positive with respect to the drain
transistors need more than the available (output). Since there is only 6V available, the
headroom of 200mV. A switching regulator supply has to be turned upside-down so that
is possible and indeed one was used at the FET can be driven from the existing rails.
GW4FRX for some time; it worked very well The valve heater doesnt mind which way
but there were rather a lot of components in round its supply voltage comes, of course, so
it! Also, some inductors involved were rather the supply polarity is academic.
difficult to make and did not prove easily Both long- and short-term output voltage
repeatable by other constructors, so for that stability of this supply are very good, since the
reason it isnt given in this book. reference Zener D1 has only a very small
The job became much more simple with the negative temperature coefficient and is fed
appearance of a new generation of power FETs from the stabilized output. Mounting one of
capable of handling very high currents. The the prototypes in a die-cast box and warming
FET equivalent of VCE(sat) is RDS(on), which is the box to about 40C produced a measured
simply the minimum resistance of the device decrease from 600V to 598V output.
when fully turned on. Modern power FETs Because the reference voltage is derived
such as those used in the present design have from the output voltage which obviously
RDS(on) figures of around 15 milliohms, isnt there at initial switch-on some form of
implying a voltage drop of no more than start-up circuit is required. This is formed by
about 90mV at 6A. The other nice thing about D2, R5 and R6, which at turn-on time provide
power FETs is that they can also be connected some 2V at the inverting input of IC1,
directly in parallel to achieve whatever low enabling the regulator to start. When the
level of RDS(on) is necessary, with no need for reference voltage is fully established, D2
ballast resistors and no danger of second becomes reverse-biased and isolates the
breakdown. This means that cheap devices unregulated start voltage from the regulator.
found at rallies can simply be paralleled to IC2 is a dual op-amp which is used for
achieve the required performance. The only overvoltage and overcurrent detection. IC2b
constraint is their quite high input capaci- compares the output voltage with an inde-
tance, and any driver circuitry for power FETs pendent reference voltage derived from the
must take account of this. Also, the majority 39V Zener diode D8. An output overvoltage
of low-RDS(on) devices available at present are fault drives a warning LED and also operates
N-channel, which means that this particular the fault relay RL1 via the diode OR-gate D5
design has a negative output voltage. and D6 and FET TR2. This relay can be
Fig. 11.7 shows the circuit of the final design connected to external circuitry as required;
evolved to meet the requirements outlined typically it should operate the main heater-
above. In essence theres an operational control relay to remove the heater voltage
amplifier, IC1, in which a sample of the output from the valve. The regulator circuitry itself is
voltage from the unit is compared with a directly interfaced with the heater control
reference voltage derived from a 39V Zener relay by means of TR3. The intention here is
diode D1. The sample voltage at the non- to ensure that the relay never switches high
inverting input is in fact slightly variable by current, either at switch-on or as a result of a
means of RV1, which forms the fine output- fault condition when the supply is switched
voltage control. The output voltage from IC1 is off by TR3 before the relay opens. In the
taken to the gate of TR1, which is the normal comprehensive PSCU setup, the amplifier
pass transistor. Note that the output of the unit airflow switch enables the heater relay via
is 6V, with the positive side of the supply contacts on RL1 but is also tied to the regula-
being at 0V (ground). The reason for this is the tor circuitry. Failure of cooling air turns off
high turn-on voltage of the FET, which for an the heater supply electronically in less than a
N-channel device implies that the gate must be millisecond, leaving the heater relay to catch
11 20
HEATER SUPPLIES
Fig. 11.8.
Connections to
the heater
regulator PC
board [6]
11 21
POWER SUPPLIES & CONTROL UNITS
using two packaged pairs is the next easiest topic. When the optimum heater voltage for
solution. Having said that, be advised that your particular application has been estab-
some of them are configured as common- lished, the overvoltage alarm should be reset
cathode devices, so you cant use them to so that it trips at some 01V higher than the
make a bridge rectifier. voltage in normal use.
The reservoir capacitor is an important
component in this design. Normally, of
course, the use of an active regulator circuit
implies that considerably less reservoir and
smoothing capacitance can be used to achieve
a particular level of output ripple. However, in
this application its essential to provide
enough capacitance to keep the trough of the
input ripple voltage above the input threshold
of the regulator during the worst case of low
mains voltage and full load. Calculating on
the basis of an imaginary transformer with 5%
regulation producing 63V at the required
current for 240V input and then assuming a
worst case of 220V input, full load current and
2V drop in the rectifier and associated wiring,
the minimum value of reservoir capacitor
which theoretically ought to be used is about
50,000F. The nearest preferred value above is
68,000F, and a working voltage of 16V would
be appropriate. The ripple-current rating of
the capacitor must also be adequate and
should be at least 10A for reliability. In
practice, large computer-grade components
with screw terminals such as those frequently
seen on offer at rallies should do the job very
well; many of these have their ripple-current
ratings marked on them. Remember, however,
that surplus electrolytic capacitors should
always be re-formed prior to use.
Finally, all wiring carrying the full input
and output currents should be substantial; we
used 30/025mm (15mm2) in the prototypes.
Fig. 11.8 shows the basic layout of the com-
pleted heater supply and the components
involved. Further details of construction,
commissioning and integration with the
comprehensive PSCU are given in the original
reference [6].
When running-in new or nearly new valves,
Id suggest you apply a heater voltage of 600V
(measured at the socket) for the first two
hundred hours use, and then carry out the
tests mentioned in the introduction to this
11 22
ALARMS & INTERLOCKS
11 23
POWER SUPPLIES & CONTROL UNITS
11 24
ALARMS & INTERLOCKS
IC1a/b and IC2a/b form over- and under- the raw DC feed to the screen regulator.
voltage detectors for the two channels of the IC2f is a spare fault input, used as the
screen supply. Typically these will be set to heater undervoltage/no-voltage alarm in
trigger the alarm if the voltage on either conjunction with the heater regulator de-
screen grid nominally 350V strays above scribed above and its easy to fan-in any
355V or below 340V. RLC1 and RLC2 alter the number of additional active-high alarm
voltage-divider values to cater for the different voltages to this point using a diode OR-gate.
voltages produced by the screen supply in Two further inputs to the common fault rail
Class AB1 and Class C. IC2c and IC2d form are derived from the airflow switch and from
under-voltage detectors for the two grid-bias the test switch, operation of which resets the
feeds. (Imagine what would happen if the timer. Other active-low alarms could be fed
grid-bias voltage disappeared and nothing was into the circuitry at this point via diodes if
done about it the anode-current meter required; in the GW4FRX amplifier the
needle would hit the end stop at about Mach antenna changeover relay supply is monitored
Three, and quite a lot of damage could be in this way.
done before the HV fuses blew.) IC1f, in All the outputs of the alarm detectors are
conjunction with the opto-isolator IC4a, tied to a common rail which is normally
forms a grid overdrive alarm whose function pulled up to +12V by R30. A failure condition
is discussed later. will force this line low and reset the one-
IC4b and associated circuitry forms a screen minute timer formed by IC1d, R17 and C15.
overcurrent detector. The diode part of the At initial switch-on C15 is discharged and the
opto-isolator is in series with the raw DC feed input to IC1d is consequently low. After C15
to the screen-stabilizer circuitry, and any fault has charged for one minute the output of
which causes excessive current to be drawn IC1d switches high, turning on TR1 and
will operate a resettable latch (IC1c) and energising the HV supply contactor. The
illuminate a LED. Incidentally, since the output of this timer also goes to one input of
screen supply is essentially a shunt stabilizer, the two-input AND gate IC3d. When the HV
this part of the circuit cannot be used for appears, a sample is fed via IC1e to the other
overcurrent protection of the valves them- input to IC3d. If there are no other alarm
selves. If the overcurrent persists for more conditions, the output of IC3d switches high,
than about five seconds, a relay disconnects turning on TR2 and energising RLD, the
ready relay.
The ready relay is the PSCUs message to
the outside world that the machinery is all
fired up and ready for action. There are several
ways of using this set of relay contacts, but in
essence it should be connected to the external
circuitry in such a way that the amplifier
cannot be placed in circuit unless it has
operated. This implies disabling the PTT line,
the antenna changeover relays, the screen
changeover relay and the bias-supply change-
over system unless the PSCU has closed the
ready relay. Have a look at Fig. 11.10 [4] for
one way of doing it. There is plenty of scope
for implementing your own requirements in
this area, and another possibility is presented
Fig. 11.10. Ready relay logic for the
comprehensive PSCU [4]. Fig. 11.15 shows later in Fig. 11.15, in the section on Tx/Rx
another option control.
11 25
POWER SUPPLIES & CONTROL UNITS
11 26
ALARMS & INTERLOCKS
11 27
POWER SUPPLIES & CONTROL UNITS
TRANSMIT/RECEIVE
CONTROL
11 28
TRANSMIT/RECEIVE CONTROL
11 29
POWER SUPPLIES & CONTROL UNITS
point as well, but you can transform this by The actual details of the preamp dont matter
adding a quarter-wave of coax line, to present here, although youll find a full description of a
a low impedance to the leakage from RLA. 432MHz design in Chapter 10.
This creates the worst possible impedance A GaAsFET typically requires 3V at 10mA, or
mismatch for the RF leakage, which will give a supply voltage of about 4V if youre using
the best possible isolation from the two relays source-resistor biasing. The best way to
together. arrange this from a 12V supply is to regulate
What would happen if you connected the the voltage down to about 5V using a 47V or
two relays directly together or used a half- 56V Zener diode (ZD1 in Fig. 11.12). Why not
wave of coax? In both cases the impedance at a 7805 IC regulator? Simply because Zeners
the RLA end would be high, so the leakage provide very fast clipping of voltage spikes,
through RLA would be transferred quite and active regulators dont. You can also turn
efficiently into RLB. In the worst case (with the dropping resistor for the Zener into a two-
two identical relays and a half-wave line pole lowpass filter, which will stretch and
between them) the extra isolation might be further reduce any spikes on the 12V rail.
only 6dB above that of a single relay! Youll be aiming to send about 10mA through
If the unused contacts of RLA and RLB are the GaAsFET plus about the same through the
shorted to ground (Fig. 11.13B) they both act Zener, so the 150 values shown for R1 and
as low-impedance magnetic pickup loops. R2 will generally be correct. RV1 and the
Once again youll need a quarter-wave line for source bias resistor of the GaAsFET should be
the worst mismatch and the best isolation adjusted to provide the necessary supply and
from the pair of relays. However, if one relay bias voltages, and can of course be optimized
has open contacts and the other one is for lowest noise figure. The second Zener
shorting, a quarter-wave line would provide diode ZD2 should have the same voltage
an impedance match for RF leakage and youd rating as ZD1; its normally inoperative but its
be in trouble. In that case the correct length there to save the GaAsFET if something else
would be a half-wave as shown in Fig. 11.13C. fails. The circuit is liberally sprinkled with
Fortunately the correct length isnt too 10F tantalum bypass capacitors and reverse-
critical. Most coax relays are only a little short voltage protection diodes (1N4148s or
of the desirable isolation, and two give you 1N4001s will do fine). Note that you must use
plenty in hand unless you make a serious tantalum capacitors ordinary aluminium-foil
error in calculating the length of the connect- electrolytics have poor HF performance and
ing line. Even so, its a Good Move to make wont even notice a microsecond spike. The
some effort to get the length right. Dont GaAsFET will fatally.
forget the velocity factor, and also dont forget
to include some estimate of the effective SIMPLE TX/RX SEQUENCER
lengths inside RLA and RLB. Fig. 11.14 shows a simple TX/RX sequencing
circuit suitable for controlling a transverter
DC PROTECTION OF GaAsFETS and a valve PA, which goes with the preampli-
Returning to Fig. 11.12, the DC connections to fier arrangements in Figure 11.12. This circuit
the GaAsFET are protected against voltage doesnt attempt to do anything clever, like
spikes on the supply rail. Modern GaAsFETs are verifying that the coax relays have changed
quite robust as far as RF leakage is concerned, over and stopped bouncing it simply leaves
but a single spike exceeding the voltage ratings enough time for those things to happen
will inevitably do them in. Some early pub- before applying 12V to the transmit stages of
lished advice has proved rather over-cautious; the transverter.
theres no problem about using the same 12V The main feature of this circuit is that the
supply for the GaAsFET preamp and the coax coaxial relays are energized on receive, which
relays, provided you take sensible precautions. has several advantages. When the entire
11 30
TRANSMIT/RECEIVE CONTROL
COMPONENTS
C1 12V electrolytic (see text)
D1 D6 1N4001 etc
R1 470R
R2 Equal to coil resistance of RLA (or a
little lower)
RLA DPDT relay, 6V DC coil
RLB DPDT relay, 12V DC coil
RV1 To suit reflected VSWR voltage and
sensitivity of SCR1
SCR1 TO92 thyristor, eg 2N5060
Fig. 11.14. A simple transmit/receive sequencing circuit for use with Fig. 11.12
system is switched off, the preamp is protected small thyristor, shorting out RLA and disa-
by two relays against things like static dis- bling the transverter and PA until the PTT is
charges. If the relay power fails, the system released and the system resets. Note that R2
fails safe. must have sufficient wattage rating to with-
All the timing is based on two relays; RLA stand being connected across the full 12V in
operates slowly, and RLB quickly. The contacts this condition.
of RLA and RLB are simply interlocked to This simple sequencing scheme can be
connect and remove 12V to the transverter adapted to many other systems. If you want
and the coax relays in the right sequence. RLA something a little more sophisticated than
is a 6V relay, fed from the 12V DC rail by R2 two relays, reference [8] describes a universal
which should, of course, be the same resist- TX/RX controller which changes over several
ance as the relay coil. When the PTT is contacts one after another, and then reverses
switched to ground, RLB switches quickly but the sequence.
the opening and closing of RLA is delayed for A potential problem with the circuit in Fig.
a few tenths of a second by C1. The value of 11.14 is that the initial changeover signal
C1 depends on the relay in question; experi- comes from the main HF transceiver, which
ment with values of a few hundred F to make means that low-level RF at 28MHz is already
the two relays switch in sequence with an present at the transverter input when the
audible ker-lick. transmit stages are energized. In this particular
The VSWR protection in Fig. 11.14 is case it doesnt matter, because nothing is
thrown in as a bonus. An excessive reflected being hot-switched with any significant RF
DC signal from the VSWR detector triggers the power. Even so, it would be nice to disable the
11 31
POWER SUPPLIES & CONTROL UNITS
prime mover until the transverter and PA are PSCU supervisory circuit votes in favour by
ready to accept a driving signal. The next closing its ready relay. All being well, this
circuit does just that and much more. should turn on the coax changeover relays.
TR6 is the next switch in the sequence, and it
ADVANCED TX/RX SEQUENCER will only turn on if the coax relays have
If you want to go the whole hog, heres a changed over correctly.
super-de-luxe bit of circuitry for interfacing Fig. 11.15 shows two alternative arrange-
practically anything to everything else, from ments for checking the coax relays. The
the fertile brain of Melvyn Noakes, G4JZQ. recommended method uses tell-back DC
This circuit really looks after you. It wont let contacts which are closed by the same mecha-
you change the PA over to transmit unless nism as the RF changeover contacts them-
everything is in order: the +12V and 12V selves. The HF-400 and HF-2000 relays provide
supplies must be present, the PA run/standby this option, which is obviously more impor-
switch must of course be in run mode, and tant on the high-power output relay. If both
the screen run/set switch(es) must also be set the input and output coaxial relays have tell-
to run. And then the pice de rsistance: the back contacts, the contacts should be wired in
circuit not only checks that the coaxial relay series as shown in Fig. 11.15, so that TR6 is
supply is on it also checks that the relays are only turned on after both sets of tell-back
there. Will it cope with different ways of contacts have closed. The relay coils could be
signalling a transmit condition, by pulling the connected in parallel if that is more conven-
PTT line low or by driving it to +8V or +12V? ient. The alternative method shown in Fig.
But of course... 11.15 is to sense the current drawn by the
At first glance Fig. 11.15 looks a bit horrific, coax relay coils, in the hope that the RF
but break it down into sections and all contacts will have changed over if their
becomes vividly clear. On receive, if the run/ electromagnets are energized. This sensing
standby switch is on standby, the PTT line method is more reliable if the two relay coils
has no effect on the PA, and the ALC output
to the transceiver is effectively zero. So the
transceiver runs barefoot at its full normal Fig. 11.15. Advanced transmit/receive
power. Switch to PA run and the controller sequencer by G4JZQ, which can be
takes over: the ALC output goes to 10V, incorporated into the comprehensive PSCU
holding off all RF drive until everythings instead of Figure 11.10. The circuit includes
good and ready. an alternative screen-grid switching
Receive-to-transmit changeover starts from a arrangement, incorporating SCR crowbar
protection which is activated by a
PTT input from the transceiver, on whichever rapid rise in anode current
PTT line is appropriate. This is the first link in
a chain of electronic and relay switches in COMPONENTS
series. Until all the switches have closed, the
screen and grid-bias relays cannot be ener- D1, D2 40A diodes (see text and Fig. 11.1)
gized and the ALC output remains firmly IC1 TIL117
negative. If the +12V rail is absent, this will M, M Screen current meters (see text)
RLA, RLB, RLC12V DC Coaxial relays As available
never happen; if the 12V rail has taken the
SCR1, SCR2 TIC126N (800V 12A or greater)
day off, the 82k from the +12V rail makes
TR1, TR4, TR7 ZVP0106A
TR2 clamp the PTT input to ground and so TR2 VN10KM
prevents any TX/RX switching. TR3 ZVN2110L
TR5 BCY70
If the PA run/standby switch is on run, and TR6 BFY50
+ and 12V are both present, a signal on the VDR1, VDR2 V250LA40B
appropriate PTT line will turn on TR3. How- Protection diodes across relay coils: 1N4002 etc.
All other unmarked diodes 1N4148 etc.
ever, nothing will happen unless the main
11 32
TRANSMIT/RECEIVE CONTROL
11 33
POWER SUPPLIES & CONTROL UNITS
are connected in series as shown, though the the voltage, but care would be required with
voltage-sensing potentiometer could be the PC board layout to isolate the HV+, screen
adjusted to sense if either of two parallel- and logic voltage levels.
connected coils had failed.
If all seems well with the coaxial relays, TR6 FINAL
turns on the screen relays (which cunningly By this stage of the book, you might think
double in this circuit as run/set relays), and youre beginning to detect an obsession with
enables the control-grid bias to bring up the reliability and doing the job as well as you
PA. Then and only then will the circuit release possibly can. Guilty, mLud. The reason is
the ALC and permit RF drive to flow. The final quite simple its the unashamed pursuit of
delight is that the ALC level on transmit is what that Zen Motor Cyclist called quality.
adjustable, to provide the variable-drive Quite apart from any aesthetic or creative
feature that I recommended in Chapter 6. If pleasures you may experience from building
the PSCU finds anything wrong in the PA and amplifiers and power supplies which for me
opens the ready relay, the antenna and are considerable and lasting the chief reason
screen changeover relays drop out just after for doing so is to work the DX more effec-
the grid-bias supply returns to the receive tively. In hard practical terms, the more you
condition, and the ALC voltage goes negative fret and worry about the amplifier and its
to turn off the drive from the transceiver. In power supplies when youre building it, the
other words, even Walter cant blow anything less youll have to think about it when you
up with this little lot! really need to use it for its intended purpose.
Fig. 11.15 also shows an improved way of Electronic components nowadays are
switching the screen-grid voltages and extremely reliable if theyre chosen with care
protecting the screens against flashover. This and run within their ratings, so the odds
circuitry improves upon the switching against something critical going bang just as
arrangements in Fig. 11.4 by making the the DX is starting to appear are quite remote. I
screen changeover relay double as the run/set dont mind too much if a random component
relay; in other words, you save two relays at failure once in a blue moon makes me miss
the expense of some modifications to the one prime contact. But I mind very much
earlier PSCU circuit board [5]. indeed if I miss some DX because I didnt do
The other feature is the optoisolator- my job properly when I built the PSCU.
coupled trip circuit which monitors the anode When the band opens, I want to switch on,
current via the 1 resistor on the HV-minus work the DX, have a wash-up with the locals
rail. If the circuit detects a fast sharp rise in and switch off to write out the QSL cards. I
current, characteristic of a flashover, it fires a dont want to spend half the opening trying
thyristor on each screen rail to provide to make the amplifier work properly, and half
positive crowbar protection to the valves and a week afterwards sorting out why it wouldnt.
bases, leaving the rest of the circuit to sort As the saying goes, train hard, fight easy.
itself out as before. This system is inspired by
REFERENCES
the pulse-transformer circuit in the G4AJW
PSCU [2]; the RC-coupled arrangement [1] Printed-circuit boards for bridge and
adopted here is rather more reliable. Ideally biphase rectifier stacks are available from
the crowbar thyristors should be in the Melvyn Noakes, G4JZQ*.
amplifier itself, though the HV-minus rail is
[2] Andy Wade G4AJW, Power supply and
more accessible in the PSCU. An alternative
control circuits for a 4CX250B amplifier. Radio
therefore would be to move the current
Communication, October 1977.
sensing into the amplifier, by monitoring the
voltage drop across a 1 resistor in the HV+ [3] PCBs for the screen regulator are available
supply. A suitable optoisolator will withstand from Melvyn Noakes, G4JZQ*.
11 34
TRANSMIT/RECEIVE CONTROL
[4] John Nelson GW4FRX and Melvyn Noakes Figs. 11.3 and 11.6-11.11 are reproduced by
G4JZQ, A power supply and control system for kind permission of the Radio Society of Great
tetrode amplifiers. Radio Communication, Britain.
December 1987 and January 1988. * G4JZQ can also supply PCBs completely
[5] Printed-circuit boards for the power supply built, tested and guaranteed.
and control unit are available from Melvyn QTH: 55 West Drive, Highfields, Caldecote,
Noakes, G4JZQ*. Cambridge CB3 7NY.
[6] John Nelson GW4FRX and Melvyn Noakes
G4JZQ, Recipe for longer life keep the
heaters under proper control. Radio Communi-
cation, July 1988.
[7] A printed-circuit board for the regulated
heater supply is available from Melvyn
Noakes, G4JZQ*.
[8] TR time-delay generator. ARRL Handbook
(eg 1986 edn, Chapter 31)
Its an evening in early November, and 144MHz has returned to normal after three days and
nights of a monster tropo opening to Eastern Europe. Youre having a natter with your
friendly rival down the road, exchanging congratulations and commiserations:
G3XBY from GW4FRX fine, Dale, all copied and very good indeed. Congratulations on
that OK in JN98. I heard you working him and I listened for all I was worth but couldnt hear
a thing I nearly fell off my chair when you gave him 539. Isnt it funny how tropo can vary
like that? Were not that far apart, our antennas are the same and our take-offs that way are
quite similar. I wish I understood tropo a bit better...
I wonder whether Fred, over your way whats his call, G0*** heard the OK too? Hes
got a super site to the east; I bet he worked a lot of good stuff. Have you spoken to him,
over?
From G3XBY. Well, its funny you should say that. I heard Fred on the first night, the
Tuesday, but I didnt hear him working anyone after that. I guess he must have gone off to
432, over.
Break, from G0***
Hello Fred, you must be clairvoyant we were just talking about you!
Yes, I know, Ive been listening to you for quite a while. Actually, I didnt do too well in the
opening because I was down to 10 watts from the prime mover. You probably wont believe
this well, you probably will, knowing me the amplifier blew up again. Thats three times
this year and to be honest Im kicking myself!
I heard all the SP stations same as you did, and that OK was quite strong too, but I just
couldnt get through the pile-ups. There was a very weak HG at one time too, but I couldnt
get through to him either. I really only called in to let you know that I wont be on the air
much for the next couple of months Ive decided to rebuild the whole thing, and Ill do it
properly this time!
Poor old Fred a dozen tasty squares to the bad and his street cred completely blown. You
resist the temptation to quote the parable of the Wise and Foolish Virgins at him, and just
offer to lend him your VHF/UHF DX Book. Maybe something in there might help...
11 35
POWER SUPPLIES & CONTROL UNITS
11 36
CHAPTER 12
TEST
EQUIPMENT
&
STATION
ACCESSORIES
12 1
TEST GEAR & STATION ACCESSORIES
TEST EQUIPMENT
& STATION ACCESSORIES
by Roger Blackwell G4PMK
12 2
RF POWER AND
VOLTAGE MEASUREMENT
12 3
TEST GEAR & STATION ACCESSORIES
Fig. 12.1. 50 RF
detector built into
a BNC plug
P=
( V - Vo ) the RF voltage detectors described above.
2R Almost any RF diode can be used as either a
where Vo is the turn-on voltage of the diode; square-law or a voltage detector, subject to
about 06V for an ordinary silicon RF diode voltage breakdown limitations at higher levels
such as the 1N914/916/4148 or about 03V for of RF. The transition between the square-law
a Schottky-barrier type such as the HP5082- and voltage-detector modes depends more
2800. Low-barrier and zero-barrier Schottky upon the RF signal level than on the diode
diodes have correspondingly lower turn-on itself. Some diodes are optimized for operation
voltages. as square-law detectors and possess low turn-
on voltages.
LOW-LEVEL SQUARE-LAW
DETECTORS SIMPLE PRACTICAL
At low RF voltage levels, of the order of a few RF DETECTORS
hundred millivolts, a detector diode is operat- There are many situations where a simple RF
ing within its turn-on region. The DC voltage detector is just about indispensable. All you
12 4
POWER & VOLTAGE MEASUREMENT
need is a Schottky diode and a 50 load fitted useful aid when setting LO drive levels or
into a type N or BNC plug as shown in Fig. testing the low-power stages of a transmitter.
12.1. This can serve as a simple milliwatt- The circuit shown in Fig. 12.2 offers excellent
meter, or as an RF sensing element in a VSWR performance with a flat frequency response of
meter for example. What deserves a wider 1MHz to at least 200MHz, and is usable up to
application is the increased sensitivity and 500MHz. Unlike many simple instruments it
power linearity which results from the also boasts a linear scale. The range is 10mV to
application of a little forward bias. You can 1V RMS, with a 50 input impedance. With
accomplish this by using a small constant- suitable calibration (or a squaring circuit as in
current source nothing more than a large [1]) it could also be used as a low-level power
resistor fed from a voltage of a few hundred meter.
millivolts into the diode from the load side. The circuit uses a pair of Schottky diodes
Its worth making the polarity of the diode (matched if possible), one of which is used in
such that it can easily be biased from the a DC feedback circuit to compensate for low-
associated piece of test equipment, as shown level non-linearity, a technique described for a
in the tuned AF level meter described later. slightly different application in [1]. Construc-
As a basic RF sniffer for probing around in tion of the RF detector circuitry should follow
transmitters, oscillators and frequency good VHF/UHF practice; the prototype units
multipliers, dont despise the simple pickup had a 50 microstripline input circuit with
loop connected to a diode detector. Use a chip capacitors and 100 chip resistors. The
sensitive meter with as large a scale as you can rest of the circuit operates at DC and its layout
find. For years I used a 5A meter with a 4- is not critical. I built several prototypes, and
inch scale, mounted in a hole in a cardboard when checked against precision commercial
box just temporary, you understand... RF millivoltmeters their accuracy was better
than 2% across the entire range. This is good
RF MILLIVOLTMETER enough to justify the use of a 3-digit LCD
An electronic RF millivoltmeter can be a panel meter module as an alternative to the
12 5
TEST GEAR & STATION ACCESSORIES
RF MILLIWATTMETER
Although not often seen in amateur stations,
an RF milliwattmeter is an indispensable item
in a professional RF lab once you have one,
youll wonder how you ever managed without
it! For example it is almost essential equip-
12 6
FREQUENCY
COUNTERS
There are several important points to consider upper limit of a few tens of MHz, owing to
about the measurement of frequency at VHF counting-speed limitations of the logic. If you
and beyond. The first is the ability to count want to count higher frequencies, you must
the frequency, which means often using a use faster (and more expensive) circuitry to
prescaler to divide the input frequency down divide the input frequency down to some-
to something within the counters range. This thing within your counters range. Usually
in turn affects the resolution of the counter, this means one or more specialized divider
because the prescaling means that a longer chips, possibly with input amplification.
gate time is needed to achieve the full resolu- Modern prescalers divide by 10 or 100,
tion of the counter. Finally, the requirements simplifying the alterations (mental or electri-
for the reference frequency standard of the cal) to the counter display. In general, avoid
counter become more severe at higher fre- the prescalers meant for use with frequency
quencies. Ill consider each point in turn. synthesizers, which divide by powers of 2 or
other non-decimal ratios. If your prescaler
PRESCALERS FOR VHF/UHF divides by 10, the counting time must be
Most basic digital frequency counters have an increased tenfold in order to retain the same
12 7
TEST GEAR & STATION ACCESSORIES
Fig. 12.5.
UHF
prescaler for
frequency
counters.
The
separate
VHF unit
begins at
the point
marked
VHF
resolution as you had with the basic counter; as add-on units or as an integral part of the
the decimal point must also be moved one counter.
place to the left. If you had not increased the Both prescalers have a 50 input impedance
counting (or gate) time, the resolution of and a 5V (TTL-compatible) output. The VHF
your counter would have deteriorated 10-fold. prescaler uses the same circuit as the final
For example, a counter with 100Hz resolu- stage of the 15GHz unit, ie everything to the
tion up to 150MHz would be more than right of the point marked VHF. If the 24V
adequate for setting up your transceiver to supply for the OM335 is a problem, you can
locate the 1443000MHz SSB calling frequency. use an OM361 which requires a 12V supply;
Provided that the frequency standard is also in that case, the output (pin 8) should have a
accurate see later you could hit the spot 1H choke to the 12V rail, with a 1nF chip
within 100Hz 1 count, ie within 200Hz. But capacitor providing a DC block before the
if you add a 10:1 prescaler to extend the input of the SP8668 divider. UHF prescaler
frequency range up to 15GHz without also chips are invariably low-impedance devices, so
increasing the gate time by a factor of 10, the dont hit them with the direct output from
accuracy would deteriorate to 2kHz at any the PA and expect them to survive! One of the
frequency which is probably not the kind of advantages of using a hybrid module pream-
improvement you were looking for. So theres plifier rather than the more obvious and
more to adding a prescaler than just bolting the cheaper MMIC is that the hybrid module will
chip on to the front-end of the counter. stand a few hundred milliwatts of RF input
With modern LSI frequency-counter chips, without overloading the prescaler chip.
its usually quite easy to make the necessary
changes when adding a VHF/UHF prescaler. FREQUENCY ACCURACY,
Often more than one prescaler is necessary to STABILITY AND ADJUSTMENT
achieve unbroken coverage from HF on up; The accuracy of a frequency counter depends
my own counter uses a single-chip divide-by- greatly on the accuracy of its internal fre-
100 prescaler to cover up to about 200MHz quency standard. This determines the gate
with good sensitivity, and a completely time during which the cycles of the test
separate divide-by-1000 unit plus a hybrid- signal are counted, so any error in the fre-
module input-amplifier block to continue the quency standard contributes directly to the
coverage up to about 15GHz. These prescalers errors in the displayed frequency count. Many
are shown in Fig. 12.5 and can be built either simple counters which use an inexpensive
12 8
FREQUENCY COUNTERS
crystal and an on-chip oscillator have Once youve achieved an adequately stable
accuracies of only 5 parts per million, at best. timebase, how do you check the accuracy? In
This corresponds to an error at 144MHz of the old days it could be very simple to zero-
something like 720Hz (plus the counting beat a harmonic of your oscillator against a
errors already mentioned). Use such a counter standard-frequency transmission. But thats
at 101MHz to set up the 404MHz LO of your no longer easy in the UK, following the QSY
432MHz transverter and you could miss the of BBC Radio 4 to 198kHz and the demise of
SSB calling channel completely. Even if you the HF standard-frequency transmissions from
can temporarily set the crystal accurately on MSF at Rugby. Personally Ive never had much
frequency, it wont stay there for long unless success in using HF standard-frequency
you are very lucky. transmissions from the other side of the
You can improve things considerably by world. The GB3BUX beacon on precisely
upgrading the crystal standard in your 50000MHz may be usable for zero-beating if
counter. Consider using a separate high- you can copy it strongly enough, but this too
quality crystal oscillator, with either a zero- has to compete with QRM from computers
temperature-coefficient crystal or one in an and other digital devices.
oven, or use a TCXO module (Temperature- There is a much better way, though, which
Compensated Crystal Oscillator) occasionally uses a VLF standard-frequency transmitter.
seen as a rally bargain. As a minimum, you This is MSF on 60kHz from Rugby, whose
could fit a lower-value trimmer in parallel frequency is accurate to 1 part in 1012. This
with the existing one (eg 03pF in parallel method avoids all the problems of phase-
with the typical 20pF trimmer) to make fine locking reference oscillators or whatever. It
adjustment a little easier. simply provides a visual comparison of the
12 9
TEST GEAR & STATION ACCESSORIES
12 10
IMPEDANCE AND
VSWR MEASUREMENT
LOADS
50 body. Now slip the modified nut over the long
The universal standard impedance in RF lead of the resistor and tighten the nut fully.
communications is 50, so the first essential Finally trim the projecting wire back and
in impedance measurement is an accurate 50 solder. While this isnt a precision load, it is
load to form the local reference against which perfectly adequate for a lot of applications.
your workaday 50 loads are judged. Its
worth obtaining one good standard load, eg ATTENUATORS
one in an N or SMA plug body, which can Attenuators have many uses, and a good
sometimes be found at rallies for a few attenuator which means one whose perform-
pounds. Keep this load in good condition by ance is accurately known and can be trusted
not using it as a dummy load for your trans- is almost worth its weight in gold. If you can,
mitter! A precision load will almost by try to acquire at least one commercially
definition be equipped with an N or SMA manufactured unit of good performance,
connector; devices with BNC plugs or sockets preferably with N-type connectors. These are
arent going to give the same performance, as available at rallies from time to time, and a few
the connector itself will introduce errors with pounds spent on one or two will amply repay
prolonged use. the modest investment. The most useful values
For many day-to-day applications, such as will be 10 or 20dB, but anything in this general
terminating the branches of directional range will be usable as a known value of
couplers or providing low-power loads for the attenuation. Having obtained your attenuation
50 detector I described earlier, you can standard, try to keep its accuracy by not using
manage with a load consisting of an ordinary it as a power attenuator beyond its dissipation,
resistor in a BNC plug. Although these are which for most semi-precision types will be a
relatively cheap to buy, theyre even cheaper watt or less.
to make. The BNC plug should be a captive- Youll need to build up a range of good-
contact type. Prepare it by soldering a small quality attenuators designed for use in a 50
disk of thin copper or brass sheet to the top system. These should include a switched set
(hexagonal end) of the clamp nut. When cool, giving any combination of 1, 2, 3, 4, 10 and
drill a 1mm hole in the centre of the disk. 10dB so that you can select 1 to 30dB in 1dB
Now take a 51 025W metal film (MF) steps, together with a selection of fixed
resistor, which is exactly what the cheaper attenuators of handy values such as 3dB, 6dB,
commercial loads use, and crop the lead at 10dB and 20dB. You can often pick these up as
one end to about 7mm. Slip the upper surplus, although if your budget for test gear is
insulator (the one that normally goes next to tight (and whose isnt?) by all means extend
the cable) over the lead, then seat and solder your selection of attenuators with a few home-
the pin. Fit the bottom insulator, then made devices. Youll find some information on
carefully push the assembly into the plug designing your own attenuators a little further
12 11
TEST GEAR & STATION ACCESSORIES
12 12
IMPEDANCE & VSWR MEASUREMENT
on. Home-made attenuators can be reasonably the VSWR would probably be rather higher,
accurate at VHF, and if built with care can even owing to impedance errors in the attenuator
work well into the UHF region. Excellent itself.
information for building your own attenuators If you make all your gain and loss measure-
can also be found in [6] and [7]. It isnt too ments using impedance-stabilizing attenuators
hard to measure their performance accurately at the input and output of the device under
with simple testgear, provided, of course, that test, youll pretty much know where you are.
you have one or more units of reasonable If you dont, youll get confusing and mislead-
accuracy to calibrate them against. ing results. For example, even your so-called
standard attenuators wont give the correct
USING ATTENUATORS TO attenuation unless they themselves see
STABILIZE IMPEDANCES source and load impedances of 50.
Almost all your measurements are going to be
made in a 50 system, ie feeding the device CUSTOM-MADE POWER
under test from a 50 source impedance and ATTENUATORS
terminating it in a 50 load. For low-level There are a number of occasions when you
measurements, such as those described in the need a particular attenuator, perhaps perma-
section on receiver measurements, you can nently built into equipment, of a value which
ensure this by placing the device under test isnt available off-the-shelf. For example, you
between two 50 attenuators of 1020dB. Any might wish to attenuate the output of your
power reflected from a mismatch at the far multi-mode transceiver to a sensible level for
side of an attenuator will itself be attenuated driving an external PA (Chapter 6). Its quite a
on the way back, so the impedance seen by simple job to make such attenuators using
the device under test cant stray far away from standard carbon- or metal-film (not wire-
50. For example, the return loss looking into wound!) resistors. Rather than present a table
a good 20dB attenuator cannot in theory be of values as most of the handbooks do, here
less than 40dB which is the same as saying are the necessary formulas to design your own
the VSWR cannot exceed 102 no matter 50 Pi and T attenuators as shown in Fig.
what is connected at the other end. In practice 12.7.
Fig. 12.7.
Pi and T
attenuators
12 13
TEST GEAR & STATION ACCESSORIES
12 14
IMPEDANCE & VSWR MEASUREMENT
Nominal 1 2 3 6 10 20
Fig. 12.9. Power measurement using a
Theory 098 207 302 616 1008 199
directional coupler Measured
1MHz 098 204 288 610 1006 200
this value into the Pi-attenuator formulas 10MHz 097 204 287 609 1006 200
30MHz 097 204 281 611 1010 200
above, we find that R1 = R3 = 163 and R2 = 50MHz 100 205 290 607 1003 199
34. For safety, design the attenuator to 70MHz 102 203 289 600 993 198
accept the full 25W of available drive, in 100MHz 100 206 292 619 1028 202
which case the powers dissipated in the 145MHz 100 207 293 609 1009 199
200MHz 101 210 290 615 1012 198
resistors R1, R2 and R3 are 77W, 81W and 300MHz 106 219 313 643 1045 199
21W respectively. Since neither you or I have 400MHz 134 225 330 673 1073 198
junk-boxes stuffed with 163 high-power 432MHz 127 232 345 670 1056 197
non-inductive resistors, they have to be made
from paralleled higher-value components.
Five 820 2W carbon resistors will give us a DIRECTIONAL COUPLERS AND
164 10W combination for R1; similarly R2 VSWR INDICATORS
can be made from ten 330 1W components
and R3 from five 820 05W. Since were not A directional coupler couples a known
interested in the last fraction of a dB in fraction of power flowing in one direction in
accuracy for this application, you can play the main line to a second transmission line, as
around endlessly with different combinations illustrated in Fig. 12.9. For the applications in
of stock resistor values. The attenuator could which were interested, both the lines are of
be built by soldering the resistors to wide PCB 50 impedance and both are terminated in
tracks, adding a box of soldered-together PC 50 loads. If the coupler has two secondary
board for shielding. lines, the power in both forward and reverse
directions can be measured simultaneously.
A SWITCHED ATTENUATOR SET Hence the VSWR can be calculated from the
References [6] and [7] mentioned earlier relationships given on page 12-12.
include descriptions of switched attenuator The coupling factor is defined as the ratio of
sets, capable of working into the VHF range if the incident power to that coupled from the
carefully made. Fig. 12.8 shows details of my main line to the auxiliary branch. Coupling is
own set. It is made with nothing more than usually expressed in dB, so a 20dB coupler is
common slide switches in a box made from one where the coupled power is 20dB smaller
soldered together PCB strips with internal than the power in the main line. Directivity is
screens between stages, and uses ordinary 1% a measure of the goodness of the coupler,
metal-film resistors with short leads. The table and is the ratio of the power appearing at the
12 15
TEST GEAR & STATION ACCESSORIES
isolated port to that at the coupled port. Fig. formed from parallel transmission lines, and
12.9 clearly shows all these relationships. The those such as the Bruene bridge based on
directivity sets limits on the accuracy of the current transformers. Both give outputs
coupler; for example, the directivity of 26dB proportional (in some fashion) to the forward
means that an indicated VSWR of 120 could and reflected power in the main line. Simple
in fact be anywhere between 108 and 133. parallel-line couplers have a frequency-
This is also a good moment to remind you dependent sensitivity, so using them on more
of the error introduced by the feeder loss than one band may not be practical. However,
when measuring VSWR at the transmitter end they will work at UHF and above, whereas the
of the feeder. For example, an indicated VSWR practical limit for current-transformer types is
of about 12:1 (400W forward, 3W reflected) at the 144MHz band.
the end of a feeder with 3dB loss is actually a
VSWR at the antenna of a little more than SIMPLE VSWR INDICATORS
14:1, since the forward power at the antenna A simple VSWR indicator has several uses
is only 200W, and the 3W of reflected power apart from the time-honoured one of check-
arriving back at the transmitter end was 6W at ing whether your antennas still there on a
the antenna. dark and windy night! For example, you can
There are two common types of VSWR insert a VSWR indicator in the line between
bridge those based on directional couplers your transverter and a valve linear amplifier to
ensure that the latter's input circuit always
presents something reasonably like a 50 load
Fig. 12.10. Details of the stripline directional
to the transceiver (Chapter 6). For that type of
coupler and detectors. The large layout is
reproduced at full scale monitoring job the VSWR indicator doesnt
need to be particularly accurate. For more
precise measurements, and for looking at the
VSWR of devices not intended for carrying
power, youll have to use something better
for example the return-loss bridge which Ill
describe later.
12 16
IMPEDANCE & VSWR MEASUREMENT
12 17
TEST GEAR & STATION ACCESSORIES
12 18
IMPEDANCE & VSWR MEASUREMENT
reference load to the REF port and a type N type N short-circuit before each measurement.
short-circuit to the TEST port. Apply a few mW Another application of the RLB is to prepare
from a modulated source at the BNC RF port, cables which are an accurate electrical half-
preferably at the highest frequency you intend wavelength long (or any multiple thereof). To
to use. Adjust the controls on the level meter so do this, connect a 50 load to the TEST port
the meter reads 0dB. Now replace the short- via a T-adaptor, check the bridge balance,
circuit with a second identical reference 50 then connect the cable to the other side of the
termination and adjust the level meter for an T-piece. When the open-circuited cable is a
on-scale reading, which will typically be multiple of a half-wavelength long at the
around 30dB. Since the two loads are suppos- frequency at which the bridge is energized, it
edly the same, any deviation from the ideal will present a very high shunt impedance so
(infinite) return loss is due to minor imbalance the bridge will again balance. This method
in the RLB. You can improve the balance by automatically takes into account variations in
either bending the bodies of the two capacitors velocity factor along the cable due to manu-
or experimenting with a small copper tab facturing tolerances. Accuracies of 1mm can
attached to one side of the diode. It should be be achieved at 432MHz regardless of the
possible to achieve a residual balance of 40dB overall length of the cable.
or so. To check the overall symmetry of the
bridge or possibly to expose differences THE SIMPLEST TDR
between your two supposedly-identical 50 How would you like to examine a suspect
loads exchange the two loads and see joint at the top of your feeder, even though
whether the bridge remains well balanced. its well after midnight, and blowing Force 8
In normal use, keep your best 50 load with hailstones the size of golfballs? Simple
permanently connected to the REF port as the use a Time-Domain Reflectometer (TDR) and
station impedance standard. Having set up the do it all without leaving your cosy shack. A
bridge, try connecting two 50 loads in TDR is a complex and expensive piece of
parallel to the TEST port by means of a T- testgear found usually only in the more
adapter. You should now see a return loss of fortunate professional labs. It allows the
6dB, which corresponds to a VSWR of 2 display on an oscilloscope of variations in
(reflection coefficient of 05) due to the 25 impedance along the length of a transmission
impedance youve presented to the bridge. line. Commercial units allow you to inspect,
Dont forget to set zero return-loss using the for example, the minute changes in imped-
12 19
TEST GEAR & STATION ACCESSORIES
ance at the joint between two connectors at pulse shows the length of time taken for the
the far end of a long cable. If youre prepared pulse to reach the end of the cable and return.
to lower your sights somewhat, and have a The final rise to full screen height shows that
scope with a reasonably fast timebase, the theres an open circuit at the far end. If the
simple circuit shown in Fig. 12.14 will do amplitude had instead dropped to zero it
quite a lot. You can measure cable impedance would have meant a short-circuit. So the
and length, and discover open- or short- scope gives you a picture of impedance along
circuits in feeders before tearing the tower the line.
down. Its based on an earlier design [9], but The circuit of the Simplest TDR (Fig. 12.14)
with an improved 50 line-driver following consists of a 555 multivibrator which generates
the pulse generator. square-wave pulses of about 6s duration at a
The operation of a TDR is not to difficult to frequency of around 10kHz. These are fed to
follow; its rather like an echo-sounder. the 2N4401 50 line driver, which is con-
Imagine what happens when a pulse from a nected to the line under test and the scope via
50 source is applied to a cable. As the pulse an attenuator. To use the Simplest TDR,
begins its journey down the line, it sees the connect it to one branch of a BNC T-piece via a
impedance of the line, so the instantaneous short length of 50 cable, and fit the T-piece
voltage on the line is that due to a potential directly to the front panel input of the scope Y
divider formed from the generator impedance amplifier. You need a scope with a DC ampli-
carefully arranged to be 50 and the cable fier usable up to about 30MHz and a triggered
impedance. Now imagine what happens when timebase of 1s per division. A sweep multi-
theres a change of impedance, say an open- plier also helps. Set the scope for DC input and
circuit. When the pulse arrives, the voltage at a sensitivity of around 100mV/cm. If your
that end of the line rises to something near scope has an external trigger input, connect it
the open-circuit output of the generator. This to the trigger output from the TDR. You should
higher-voltage reflected wave now propagates now see the pulse from the TDR on the screen.
back down the cable, and eventually it reaches Adjust the controls so that the timebase
the generator end where the rise in voltage is triggers on the rising edge of the pulse, and the
indicated on the scope. What you see in this pulse occupies the full height of the screen.
case is a short pulse of half-screen size fol- Now to begin testing. Check the operation
lowed by a rise to full amplitude. The half-size first by putting a 50 load on the other
12 20
IMPEDANCE & VSWR MEASUREMENT
12 21
TEST GEAR & STATION ACCESSORIES
RECEIVER
MEASUREMENTS
Fig. 12.16. Using a tuned audio-level meter to measure the gain of an amplifier
12 22
RECEIVER MEASUREMENTS
12 23
12 24
COMPONENTS
C1-6 10n polyester 5% tolerance
IC1-4 Dual op-amp, TL072 or TL082
T1 Transistor interstage transformer 1:4.5 approx. (eg Eagle LT44)
M and RM 100A meter and 22k, or similar combination to give about 25V full-scale
Fixed resistors 025W 1% or 2% metal film
level meter
This comes from the decibel formula, ments to the tuned circuits. To modulate
modified for a square-law detector: these or any other low-level RF sources, the
10 antilog10(dB below full scale) = % FSD simple circuit of Fig. 12.18 can be used. A
Note that the meter is scaled in terms of 1kHz oscillator using a 555 timer IC provides a
power, not voltage. If youre a perfectionist, square-wave drive to a simple RF chopper
you can use an accurate digital multimeter using a PIN diode, generating nearly 100% AF
with an external adjustable current source to modulation. The output impedance isnt
mark the meter scale directly. This compen- going to be 50 so the modulator must be
sates for any non-linearity in the meter followed by a normalizing attenuator; this
movement, which can be quite detectable. might as well be built-in permanently as
Setting up the instrument is simplicity itself. shown.
First feed a modulated RF signal into the diode
detector and peak each filter response. Then SENSITIVITY AND NOISE
short-circuit the input, switch the attenuator FIGURE MEASUREMENTS
to its most sensitive range (0dB) and adjust For modern VHF/UHF receivers, the best
the SET LEVEL control to about 50%. If all is methods of sensitivity measurement are those
well you should see about 20% deflection on based on noise figure or better still, the
the meter due to circuit noise. fundamental concept of noise temperature.
The traditional method using a signal
MODULATED RF SOURCES generator cannot sensibly be used with a
To use the AF level-measuring technique just modern VHF/UHF front-end which can
described, youll need some amplitude- probably detect signals as small as 20nV, as
modulated RF sources. You may have a explained in Chapter 4 and reference [12].
suitable signal source already the modulat- Most signal generators simply cant produce
ing tone of most RF signal generators is 1kHz, such low levels with the required accuracy, so
which is also the standard modulating its extremely difficult to make measurements
frequency for these measurements. The local based on the signal level required to give a
oscillators already described in the earlier 3dB increase in receiver output. If you have a
chapters of this book can be adapted as in- reasonably screened generator and good
band RF signal sources with appropriate attenuators, you can use it to measure the
substitutions of crystals and minor adjust- sensitivity of an HF transceiver used with a
COMPONENTS
RV1 1k 10-turn cermet trimmer
C1 100n polyester
D1 BA479 or similar PIN diode
Fixed resistors 1% or 2% 0.25W metal film
RFC 47H miniature choke
12 25
TEST GEAR & STATION ACCESSORIES
transverter, but thats about all. measurement, its best to use a bandwidth of
Absolute noise-figure or noise-temperature at least several kHz. One genuinely worth-
measurements are not an easy task for the while improvement to the original G4COM
amateur. In the absence of a calibrated noise design is to increase the square-wave voltage
source, the only viable solution is to measure driving the noise source to about 12V.
noise temperature using the hot/cold The traditional noise-source using a micro-
method. This involves measuring the change wave mixer diode is not satisfactory for
in thermal noise from a 50 resistor at the modern FET front-ends. The noise source
receiver input, first when the resistor is hot must present an accurate 50 impedance to
and then when its cold. Typical combinations the front-end in both its on and off states.
of temperatures for amateur measurements are Otherwise the change in the match can
boiling water and melting ice, or room considerably alter the gain of a FET stage
temperature and melting dry ice or liquid whose own input impedance is not close to
nitrogen. The details of this method are 50, and this will render any NF measure-
described in [12]. The thermionic noise diode ment completely invalid. Worse than that,
can no longer be considered a reliable absolute you cant even tweak the front-end for lowest
noise-source it was good enough in the days indicated NF because the main effect of your
of valve front-ends (which those of us old adjustments will be not be upon the front-end
enough to remember are still trying desper- performance but upon the measurement error
ately to forget) but its no good in the GaAs- itself! A well-matched noise source avoids
FET era. these problems, so a front-end optimized by
So where does that leave us? If youre not an amateur methods using a well-matched source
adept with cryogenics, relative measurements will actually be in better shape than one thats
are about all thats left. Quite good enough been measured using the older generation of
too, for most purposes. If your front-end has professional noise sources.
been designed to meet a noise-figure target as A good noise source [15] which is eminently
described in Chapter 4, tweaking for maxi- suitable for use with the G4COM alignment
mum signal-to-noise ratio is probably all you aid is shown in Fig. 12.19. It uses the reverse
need to do. This kind of relative measurement base-emitter breakdown characteristics of a
is fairly simple if you watch what youre BFW92 transistor to generate a generous
doing. If absolute numbers still worry you, amount of noise well up into the microwave
take your pet preamp to the measurement region, allowing sufficient attenuation to be
stand at a rally, or find someone who boasts placed on the output to swamp the on-off
of a professional noise-figure meter at work. changes in source impedance. Ive shown a
But even relative NF measurements have 16dB attenuator built-in, though you may
their pitfalls. Unless youre prepared to spend want to include some extra external attenua-
hours on repeated measurements of tion. Build the noise head in a shielded box,
signal+noise and noise levels, youd better with the shortest possible lead lengths for the
build an automatic relative NF indicator. The output circuit. For those of you who can read
almost universal choice is the G4COM the original reference in Dutch [15], or are
alignment aid [13], which uses the difference brave enough to build from the circuit
between the receiver outputs when a noise diagram alone, that article also described an
source is switched on and off to generate a improved NF meter which can have its scale
signal proportional to NF. A number of directly calibrated in NF. That might be worth
modifications to the basic design have been trying, assuming you can check the calibra-
published [14], though the main cause of tion against a reliable professional instrument,
erratic performance seems to be the behaviour although the long-term stability of any noise
of the SSB IF filter through which measure- source is problematical.
ments are usually made; in any noise-based A noise-figure measuring instrument of
12 26
RECEIVER MEASUREMENTS
STRONG-SIGNAL SOURCES
AND PERFORMANCE
Good sensitivity isnt the only characteristic
that the VHF/UHF DXer requires from a
receiver. Strong-signal handling is vitally
important too. If youre not a convert yet, go
back and re-read Chapter 5. To measure the
strong-signal performance of your system you
will need some strong-signal sources. The
requirements for these are pretty stringent,
since the shortcomings of the receiver are
often not easily distinguishable from those of
a poor signal source. Ill simply outline the
bits of specialized testgear youll need for
making the measurements; for the theory
behind it all, refer back to Chapter 5.
Measurements of the strong-signal handling
performance of a receiver require sources
12 27
TEST GEAR & STATION ACCESSORIES
TRANSMITTER
MEASUREMENTS
12 28
TRANSMITTER MEASUREMENTS
12 29
TEST GEAR & STATION ACCESSORIES
it pays to check rather than trust the calibra- directional couplers, with some practical
tion! Within an amateur band your home- designs, see [6] and [21]. Directional couplers
produced power measurements can be at least are usually rather frequency-conscious so
as accurate as a Bird 43, and decidedly more youll need to calibrate them for each band.
accurate than many power meters made for Generally a coupler which uses parallel lines
the amateur market. has an upper frequency limit where the length
of the branch line approaches a quarter-
DIRECTIONAL COUPLER AND wavelength. The lowest useful frequency is set
MILLIWATTMETER by the falling coupling factor.
The customary way to measure high power Although making a good high-power
(watts rather than milliwatts) in professional directional coupler is a rather difficult task
circles is to use a directional coupler to obtain without some machining facilities, you can
a known fraction of the power in the main measure its performance quite accurately
line. A low-power instrument, such as the using the return-loss bridge, modulated signal
milliwattmeter I described earlier, can then be source and tuned AF level meter which were
used to measure the power in the branch line; described earlier. Before the days of network
hence the power in the main line can be easily analysers, thats exactly what the professional
calculated. A typical setup was shown earlier RF labs did use.
in Fig. 12.9. For accurate measurements it is A directional coupler using semi-rigid coax is
vital that the main line is correctly terminated shown in Fig. 12.21. Based on a design de-
in 50, or else the coupler will not work as it scribed in [22], it offers coupling of about 55
should. So thats yet another reason for to 45dB over the range 100500MHz, is usable
needing a good dummy load capable of up to at least 23GHz and can be simply made
dissipating the full legal power limit and then with hand tools. This coupler can form the
some. basis of a power monitor for a high-power
For most applications Id suggest a coupling 432MHz transmitter for example. You first
factor of about 20 to 30dB. With a 30dB need to find some -inch semi-rigid cable, since
coupler, 500W in the main line results in the more common UT141 is too small. To
500mW appearing at the coupled port on the make the coupler, first file identical slots in two
branch line (Fig. 12.9). You can of course pieces of semi-rigid cable as shown in the
extend the range of a 20dB coupler by adding diagram. Then slightly bend one of the pieces
a suitably-rated attenuator at the input of the so that there will be enough room for all the
milliwattmeter. For a good treatment of connectors when the two slots are offered up to
each other. When everything is positioned
Fig. 12.21. Directional coupler made from correctly, bind the cables together with wire.
semi-rigid coaxial cable
Then solder the joint with a big iron or hot-air
gun, or very carefully with a blowtorch. Finally
fit the connectors and calibrate the coupler.
Another alternative to a directional coupler
is to use a power attenuator (eg 2030dB) with
a milliwattmeter. If you can come across a
large power attenuator at a good price, it
forms the basis of an excellent power meter
and dummy load.
12 30
TRANSMITTER MEASUREMENTS
12 31
TEST GEAR & STATION ACCESSORIES
order that the attenuated and combined two- COMPONENTS FOR TWO-TONE SOURCE
frequency signal emerges at the correct drive (FIG. 12.23)
700 Hz oscillator 1900 Hz oscillator
level for the power amplifier. If youre R1, R2 11k 470R
prepared to contemplate all that, I reckon you Adjust on test for exact frequency
C1, C2 1n0 + 1n0 1n0 + 1n0
wont need any more guidance from me! C3 2n7 + 2n7 1n0 + 1n0
For most of us, a two-tone intermodulation C4 22n + 12n 12n
C5 100p + 39p 47p
test means injecting a two-tone audio signal Where two values are shown these should be paralleled.
All the above capacitors should be polystyrene, 5% tolerance
into the microphone input of the SSB exciter. or better.
This has the virtue of testing the entire B1, B2 Tandy (Radio Shack) 272-1141 12V
25mA bulb
transmitting setup, though individual sources IC1-3 TL072 or TL082 dual op-amps
RV1, RV2 470R or 500R 10-turn cermet trimmer
of intermodulation can become very difficult RV3 470R or 500R panel-mounting
to unravel. For amateurs, however, the main potentiometer
RV4 1k panel-mounting potentiometer
consideration is that a two-tone audio source
is far easier to build than one based on two
CW transmitters. The audio two-tone source I
use (Fig. 12.23) is based on the design in the
1986 ARRL Handbook [23]; it works just fine,
so I make no apology for recommending it
here. As you see, it consists of two separate
Wien-bridge oscillators and active low-pass
filters, with a combining amplifier. Particu-
larly good features are the inclusion of a step
attenuator (calibrated in dB of course) and the
ability to generate a single tone which allows
static linearity testing as well.
The particular bugbear of Wien-bridge
oscillators is the amplitude-stabilizing element
frequently either an expensive thermistor or
an incredibly obscure low-current bulb. In this
design, the necessary bulbs are readily avail-
able in many countries from Radio Shack or
Tandy stores. Its useful to make the output
socket a common DIN type, and to have a
number of short adapter leads with micro-
phone plugs to suit your various SSB rigs.
Putting a PTT switch in the box is a good idea
too.
12 32
Fig. 12.23.
Two-tone
source
based on
ARRL design
12 33
TEST GEAR & STATION ACCESSORIES
In this section Im going to describe a number little to be gained by building your own filters
of different applications of filters, with since off-the-shelf Toko helical filters are
examples for each band. The accent will be on available from several UK component suppli-
modern, reproducible designs. Most have been ers, so Ive concentrated here on bandpass
designed or analysed using computer pro- filters (BPFs) for 50MHz and 70MHz. Ive also
grams; all have been built and tested to make included a design for 28MHz since equipment
sure that their performance is up to expecta- reviews show that the transverter outputs of
tions. many HF rigs will benefit from the extra
selectivity to clean up spurious signals. All the
BANDPASS FILTERS FOR BPFs shown in Fig. 12.24 are two-resonator
LOW-POWER STAGES Butterworth designs and have 50 input and
The filters in this section are intended either output impedances so that they can be
for use in the receiver chain for example, to cascaded if necessary. Reference [24] describes
provide front-end selectivity after a low-noise a technique for tuning multi-resonator filters,
preamp or after the mixer in a transmit which is particularly easy using the modu-
converter. For 144MHz and 432MHz theres lated-source testgear described earlier in this
chapter.
12 34
FILTERS FOR VHF & UHF
COMPONENTS
50MHz:
C1 68p
C2 120p
L 160nH 10 turns 05mm enamelled
wire on T37-12 toroid; or 4 turns
15mm enamelled wire on 95mm
diam. former, close-wound
50MHz with 100MHz notch:
As above, but replace one C2 with a
series-tuned circuit of a 110nH
inductor (see 70MHz details) and a 5-
35pF trimmer. Tune for best
suppression at 100MHz.
70 MHz:
C1 47p
C2 82p
L 110nH 8 turns 05mm enamelled
wire on T37-12 toroid; or 3 turns
15mm enameled wire on 95mm
Fig. 12.25. Lowpass filters diam. former, close-wound
144 MHz:
C1 22p
C2 39p
L 55nH 3 turns 1mm wire on 5mm
diam former, spaced wire diameter
432 MHz:
C1 5p6
COAXIAL TRAP DETAILS C2 8p2
All dimensions assume the use of URM43 cable L 12nH As Fig. 12.25
Band /4 /8 C(max) Printed-circuit capacitors can be made
50MHz 10m 05m 45p from double-sided board as shown.
70MHz 071m 035m 35p C1 and C2 require areas of about 300
144MHz 035m 017m 12p and 550mm2 respectively for G10
(glass-epoxy) board; for Duroid 5870
(glass-PTFE) board these areas would
Fig. 12.26. Notch filters using coaxial-cable be 630 and 1150mm2
stubs
12 35
TEST GEAR & STATION ACCESSORIES
the large circulating currents and if they components list also give details for this
dont survive, they werent good quality! For alternative, together with suggested values for
the 432MHz LPF, Fig. 12.25 gives an alterna- 50, 70 and 144MHz. To adjust any of these
tive layout using PCB capacitors. stubs, a test setup using equipment Ive
Lowpass filters for the 50MHz band require already described is also shown.
extra suppression of the second harmonic,
which falls within the FM broadcast bands.
The component list in Fig. 12.25 includes a
variation of the standard 7-pole LPF in which
a 100MHz series tuned trap is substituted for
one of the capacitors, giving an extra 20dB of
rejection.
12 36
STATION ACCESSORIES
Here, to finish with, are a few ideas to make burst is replayed. Modifying a standard
operating your thoroughly tweaked and tested cassette recorder for this purpose is a usually a
station more pleasant and successful. simple job, provided the motor has an
electronically-regulated power supply. You
KEYERS AND ACCESSORIES merely have to doctor the potential divider
FOR METEOR SCATTER which sets the speed, usually identified by a
There are plenty of good memory keyers trimmer potentiometer. If the motor has a
described in recent books and magazines, so I fixed mechanical governor, by the way, forget
wont add to that list here. All I ask is that you it and buy another tape recorder.
spend some time analysing your operating Since the necessary change in playback
habits in other words, what you really do speed may be a factor of ten or more, a meteor
send time after time before buying or burst recorded at normal pitch is replayed at a
building the memory keyer of your dreams. burbling 100200Hz which doesnt make for
Unless youre a meteor-scatter or moonbounce easy copy. Unless you permanently shift the
addict, youll probably find that the most BFO for MS reception, or your receiver has an
common use for a keyer is merely to grind out unusually wide IF shift, you cant pitch the
your own version of CQ CQ CQ DE G4PMK incoming signal much higher than 2kHz
G4PMK G4PMK, or CQ CQ CQ TEST for without risk of losing it over the top edge of
contests. If so, a very simple keyer [26] with the IF filter. The most popular way around
an EPROM memory will satisfy over 90% of this problem is to use an audio up-converter.
your operating requirements. Most keyers will The audio signal fed to the tape recorder is
adapt quite easily to the addition of an shifted up in frequency by mixing it with a
EPROM message facility. It only remains to tone of about 7kHz, so the MS bursts are
find your local friendly computer hacker with recorded at an audio frequency of 89kHz and
an EPROM burner. played back at a comfortable 800900Hz.
If you are keen on meteor scatter, an The popularity of the idea is due in no small
EPROM memory is still useful for the un- measure to LA8AK, whose elegant implemen-
changing messages such as CQ MS G4PMK tation of this idea using a readily available
and RRRRRRRR G4PMK, in addition to the consumer IC is shown in Fig. 12.27; this and
main read/write memory. earlier versions of his idea have appeared in
To read ultra-high-speed CW for MS several publications [27]. The TBA120 IC
operation, you naturally have to slow it down contains both the double-balanced mixer and
first (Chapter 3). The accepted way of doing the active elements for the 7kHz oscillator,
this is to use a variable-speed tape or cassette which is pretty good going because this IC
recorder. MS signals are recorded at a faster was originally meant to be an FM IF amplifier
tape speed than normal, and the motor is run and discriminator! Setting-up merely consists
more slowly than normal when the captured of confirming oscillator operation by means of
12 37
TEST GEAR & STATION ACCESSORIES
a scope connected to the test point and circuit shown in Fig. 12.28 is suitable for the
adjusting the balance preset for minimum majority of rigs, which have a push-to-talk
7kHz feedthrough with no AF input. Although (PTT) line sitting at about +12V on receive and
the circuit specifies a TBA120, other suitable needing to be grounded to transmit. The unit
devices include the TBA120A, TBA120S, monitors the state of the PTT line and detects
SO41P and SN76660N. the rising edge when the PTT is released. It
One unusual idea for MS recording is to then pulls the PTT line back down to ground,
turn the memory keyer idea on its head, so to holding the rig on transmit, injects a 1kHz
speak. G4ASR converts the output from the tone into the microphone input line for about
receiver into a logic signal, which is clocked 30ms and finally releases the PTT. In many
into the memory at a high rate. When a burst rigs the necessary connections to the PTT line,
is heard, the loading is stopped, the clock-rate microphone input line and +12V power
slowed down, and the output from the supply are all available at an accessory socket,
memory switched to a tone generator. so the pip-tone unit can simply be plugged
Still on the subject of keyers, its now into the back of the rig.
possible to digitize speech and store it in a The unit is based on two CMOS logic ICs; a
memory for repeated replay with good audio 4011 quad 2-input NAND gate (IC1a-d) and
fidelity [28]. Soon the SSB MS throat will be a four elements of a 4049 hex inverter (IC2a-d).
thing of the past, and the digital voice IC1a, IC1b and IC2a form a detector which
recorder can also take the strain of those monitors the state of the PTT line. On receiv-
endlessly-repeated CQ contest calls. ing a trigger pulse from IC1b, the monostable
IC1c-IC2b delivers a 30ms pulse via IC2c,
PIP-TONE UNIT which pulls down the PTT line through TR1
Used with common sense, a pip-tone at the and also enables the tone generator formed by
end of a transmission can be a very positive IC1d and IC2d. The PTT sensing circuit is a
aid to weak-signal SSB communication. It is little complex because it has to respond only
most useful in conditions of deep fading or in to a deliberate release of the PTT switch. This
quick-break meteor-scatter operation. The is achieved by gating the rising-edge detector
12 38
STATION ACCESSORIES
12 39
TEST GEAR & STATION ACCESSORIES
microphone gain control, and this level is Products which fall outside the normal audio
suitable for direct clipping by the long-tailed passband are removed by the lowpass filter
pair TR1 and TR2 which are biased as a (TR3 and TR4) but those which fall within the
symmetrical peak limiter. Ideally the transis- audio passband cannot be removed in this
tors should be a matched pair or a dual manner; too much audio clipping will thus
transistor to provide symmetrical clipping of produce a heavily distorted signal with
positive and negative peaks. Clipping is a reduced intelligibility. If you want more talk
form of distortion and produces copious power, youll have to use a more advanced
intermodulation products and harmonics. and complex speech processor such as the
Datong this simple AF unit is more for our
benefit than for yours.
Speech clippers always need to be set up
with great care and Chapter 6 gives full
instructions on how to do this. The only
preliminary alignment required for this
module is to is to adjust its overall gain to
unity, so that the following stages of the rig
are not over- or under-driven. Feed a 1kHz
tone into the module, at a level too low to
suffer any clipping, and with the aid of a
scope adjust the preset control to give equal
signal levels at the input and output of the
module. Then follow the general setting-up
instructions in Chapter 6. In this particular
arrangement the CLIPPING LEVEL control
referred to in Chapter 6 is the rigs own
microphone gain control which comes before
the clipper module; the DRIVE LEVEL control
is the rigs own drive control.
12 40
STATION ACCESSORIES
FINIS
I hope youve found some ideas in this
chapter to help you obtain the best perform-
ance from your station, and that you get as
much satisfaction in making the measure-
ments as you do from the results. Just keep a
sense of proportion, and dont lose sight of
your objectives. The ultimate aim is not to
achieve perfection in the art of measurement,
but to make your station as effective as you
can and then to use it for chasing DX!
Fig. 12.30. Rotator-speed controller. The original rotator components are shown inside the
dashed line
12 41
TEST GEAR & STATION ACCESSORIES
12 42
INDEX
vii
INDEX
INDEX
viii
INDEX
Bold numerals indicate a reference to a whole chapter
ix
INDEX
Bold numerals indicate a reference to a whole chapter
D E
D region, ionosphere 2-14, 2-38, 2-54 E region, ionosphere 2-15, 2-31, 2-39, 2-48, 2-56
dB, dBm, dBW see Decibels Earth currents 2-39
dBc 5-9 Echos
dBd, dBi 7-2 moon 1-2, 2-59, 4-12
Decibels 12-3 long-delayed 2-63
Delta match 7-39 Effective radiated power 4-10, 6-5, 6-12, 7-19
Design Effective receiver sensitivity 4-10
antenna 7 Efficiency, antenna 7-7, 7-30
for DX 1-4 Electric field 2-4
power supply 11 Electromagnetic compatibility 6-42
receiver system 5 Electromagnetic waves 2-2
RF system 4 Electrons 2-5, 2-31
test equipment 12 Electron oscillators 2-5
transmitter 6 Elephants 4-13
50 and 70MHz 9 EM waves see Electromagnetic waves
144MHz 8 EMC see Electromagnetic compatibility
432MHz 10 EME see Moonbounce
Detectors Emission, secondary 6-26
50 RF 12-3 Endfire arrays 7-11
square-law 12-4 Equipment
Diffraction 2-7, 2-12 Test 12
Diodes 50 and 70MHz 9
mixer see Mixers 144MHz 8
Schottky 12-4 432MHz 10
varicap 5-26 Equivalent noise temperature 4-4
Zener 11-8, 11-30 ERP, EIRP see Effective radiated power
Directional couplers ERS, EIRS see Effective receiver sensitivity
calibration 12-30 Es See Sporadic-E
semi-rigid coax 12-30
UHF PCB 12-15 F
Directivity F region, ionosphere 2-15, 2-49
x
INDEX
Bold numerals indicate a reference to a whole chapter
xi
INDEX
Bold numerals indicate a reference to a whole chapter
xii
INDEX
Bold numerals indicate a reference to a whole chapter
N noise 5-8
Neighbours, psychology 6-42 synthesized 5-9
Nets VCO 5-10
HF liaison 3-28 VFO 5-2, 5-9
telephone warning 3-26 Oval, auroral 2-31, 2-33
Neutralization 6-29 Overdense 2-8, 2-57
Newsletters 3-30 Overdriving, transmitter 6-4, 6-7
Noise Overload, receiver 5
digital 5-10
factor 4-6 P
figure (NF) 4, 5, 8-3, 8-24, 9-2, 9-20, 10-3, Path loss 2-18, 4-9, 4-10
10-25, 12-25 capability 2-18, 4-9
floor 4-9, 5-12, 5-14, 5-19 free-space 2-18
galactic 2-60, 4-3, 4-7, 4-15 moonbounce 2-59, 4-12
generator 12-25 with reflection 2-19
ground 4-7, 4-15 troposcatter 2-29
man-made 4-7 Pattern, antenna 7-3
oscillator 5-8, 5-17 Peak envelope power 6-5
pedestal 5-9, 5-11 meter converter 12-30
phase 5-9 PEP see Peak envelope power
power 4-3 Phase, EM waves 2-4
sidebands 5-9, 5-11, 5-17 coherence see Coherence
thermal 4-2 Phase-locked loops 5-10
sky 2-60, 4-7, 4-15 Phase velocity 2-4, 2-6
sun 4-7, 4-15 Phasing, antennas 7-34, 7-36
Noise figure (NF) 4, 5, 8-3, 8-24, 9-2, 9-20, 10-3, Phonetics 3-7
10-25 Pile-ups 3-3, 3-5, 3-9, 3-11, 3-14, 6-12
measurement 12-25 PIN diodes 8-7, 9-5, 9-23, 12-25
Noise/gain analysis 4-4, 5-3, 5-18 Ping see Meteor-scatter
Noise temperature 4, 5 Pip-tone 3-16
antenna 4-7, 4-8, 7-6 accessory 12-38
measurement 12-25 Plasma 2-14, 2-32
sky 2-60, 4-7, 4-15 PLL see Phase-locked loops
system 4-4 Polarization 2-4, 2-10
target 4-8, 5-3, 8-3, 8-24, 9-2, 9-20, 10-6 circular 2-4, 7-12, 7-21, 7-39
Norton amplifiers 10-7 horizontal 2-10, 7-2, 7-8, 7-9, 7-12
Notch filters, RF 12-36 linear 2-2, 2-4
rotation 2-62
O Power, RF
Obstacles, reflection 2-16 EM waves 2-4
Obstacle gain 2-16 gains and losses 4-9, 12-3
Open-wire feedline 7-36 low (QRP) 4-13, 6-12
Operating 3 high 4-13, 6-12
Oscillators legal-limit 6-12, 6-38
AF two-tone 12-31 measurement 12-29
crystal 5-2, 5-9, 8-6, 9-5, 9-21, 10-11 peak envelope 6-5, 12-30
electron 2-5 Power amplifiers 6
xiii
INDEX
Bold numerals indicate a reference to a whole chapter
xiv
INDEX
Bold numerals indicate a reference to a whole chapter
xv
INDEX
Bold numerals indicate a reference to a whole chapter
U
Universal Coordinated Time (UTC) 3-16
Unbalance, antenna 7-37
Underdense 2-8, 2-57
V
Valves, transmitting 6-14
Varistors see VDRs
VDRs 6-30, 8-32, 10-32, 11-5, 11-9
VCO see Oscillators
Velocity
radio waves 2-4
group and phase 2-6
VFO see Oscillators
VHF nets (on HF) 3-28
Virtual height 2-15
Voltage-controlled oscillator (VCO) see Oscillators
Voltage-dependent resistors see VDRs
Voltmeters, RF 12-3, 12-5
VSWR
definition 12-12
bridge, 28144MHz 12-16
bridge, 50/70MHz 12-17
bridge, UHF 12-16
W
Water vapour 2-13, 2-27
Wattmeters, RF 12-3, 12-6, 12-16, 12-29
Waves
radio, electromagnetic 2-3
gravity 2-45
standing 12-12
xvi