Вы находитесь на странице: 1из 12

Paper No.

1536
RADIO SECTION

621.396.677.3

A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS


By A. BLOCH, Dr.-Ing., M.Sc, Member, R. G. MEDHURST, B.Sc, and
S. D. POOL, B.Sc.(Eng.), Associate Member.
{The paper was first received \lth October, 1952, and in revised form 23rd April, 1953.)
SUMMARY
The current distribution required for maximum directivity of an
array with a finite number of elements and any specified geometrical
configuration is shown to be completely defined by the self- and
mutual resistances of the elements and by a certain component of the
voltage (the "resistance voltage") across the terminals of each element.
This voltage component is required to vary from element to element
in the same way as the instantaneous local values of a sinusoidal
disturbance travelling across the array, in the direction under consideration, with the velocity of an electromagnetic wave.
As a consequence, the maximum gain of the array is expressible
either as a double sum containing only the mutual conductances
between the individual elements multiplied by trigonometrical factors
depending on their spacing, or as an expression identical (except for a
numerical factor) with that for the distant field of the array.
These theorems hold, slightly modified, for arrays of non-identical
elements.
The theory has been applied to the numerical calculation of certain
simple arrays. It appears that, for arrays of a given size, directivities
greater than those obtained by conventional design methods can be
achieved without excessive losses.
This has been substantially confirmed by an experimental array
of four elements operating at 75 Mc/s. The theoretical gain was
10-ldb, while 8-7 db was measured. Of the discrepancy, 0-6 db
was calculated to be due to losses in the feeder system and a further
0-2 db to losses in the dipoles. The bandwidth was about +$ Mc/s
for a drop in gain of \ db. The degree of super-directivity is indicated
by the fact that a physically identical array fed with equal-amplitude
currents phased for maximum field strength in the end-fire direction
would have a gain of 4- 6 db.

G = Power gain of the array with respect to


reference element.

2
"l=1

, = "Resistance" component of the total


voltage across the /th element.
As

Zm

As

= Normalized mutual impedance between


elements / and m.
= R + jXm = Driving-point impedance of mth element.

(1) INTRODUCTION
The work described in the present paper had its origin in the
so-called "super gain" theorem relating to directional aerials.
This states that it is possible to design an aerial of arbitrarily
small dimensions with a directivity as high as desired. The
theorem applies only when there is sufficient freedom to fill the
finite space occupied by the array with either a continuous
current distribution or an unlimited number of discrete radiators.
Closely connected is a second theorem which can be stated in
the following way:19 The bandwidth and radiation efficiency of
an aerial decrease if its gain is held constant while its size is
reduced below that of a conventional aerial having this gain.
The concept of super-gain arrays apparently contradicts the
common experience that an increase in the gain of an aerial
must be accompanied by an increase in its size, and is apt to be
considered an academic curiosity. It is not therefore surprising
LIST OF PRINCIPAL SYMBOLS
that the super-gain theorem was forgotten and rediscovered6-9
n Total number of elements in array.
twice during the 24 years following its initial discovery.1 Meanwhile,
intermittent attempts8-11'13 were being made to calculate
Zmm = Rmm + J^mm Self-impedance of element m.
Z/m = i?/m + jXlm = Mutual impedance between elements / "optimum" current distribution for systems where there is, in
fact, no finite upper limit to the directivity.
and m.
Im = l'm + //^ = Current in mth element.
The second of the two theorems stated above is of great
Vm = V'm + jV'n = Potential difference between the ter- practical importance to the aerial designer who wishes to know
minals of the mth element.
whether or not it is economical to use an aerial with a directivity
P = Radiated power.
higher than normal.* Recent work20-23-26 on this aspect of the
H = Received field strength at some distant problem has taken the form of computing current distributions
point in a direction of maximum for particular super-directive arrays, and pointing out that these
radiation.
are associated with very large current amplitudes, and hence
Is = Current in a reference element.
with low efficiencies.
Hs = Received field strength when the referThe methods used to find these current distributions do not
ence element is substituted for the consider the radiation efficiency of the array, nor do they lead
array.
to maximum directivity, where the aerial system is so restricted
Rs = Self-impedance of reference element.
that a maximum exists.
6m = Electrical angular distance from a
If one considers the current distribution required on an array
reference plane (perpendicular to the of a finite number of fixed elements to produce an arbitrarily
direction in which the array is to selected directivity, less than the maximum directivity, one may
have maximum gain) to the mth expect to find an infinity of such distributions, many of which
element. The positive direction is may be accompanied by low radiation efficiencies. This will be
from the reference plane towards the true whether the selected directivity is greater or less than
receiver.
normal. Consequently, to dismiss super-directive aerials as
impracticable23-26 merely because some extremely inefficient
Written contributions on papers published without being read at meetings are

invited for consideration with a view to publication.


The paper is a communication from the Staff of the Research Laboratories of The
General Electric Company, Limited, Wembley, England.

Normal directivity is taken to mean that associated with an array of equalamplitude currents, chosen so that the distant field components are in phase.

[303]

304

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

super-directive current distributions have been found does not


seem justifiable.
It is not yet known how rapidly the bandwidth and efficiency
of an array decrease when the directivity is increased above
normal, so that it is of interest to seek a current distribution
which, while having a usable bandwidth, combines superdirectivity with an efficiency great enough to give an improved
power gain over conventional designs.
To reduce the work to practicable proportions, the investigation was restricted to the unique current distribution giving
maximum directivity with a particular configuration of radiating
elements.
Previous calculations of super-directive current distributions
for discrete radiators have been based on the work of Schelkunoff,6 on linear equispaced arrays. However, Schelkunoff's
method does not, in general, lead to maximum gain; hence it is
not clear how much further improvement is possible by adjustment of the currents.
The only previous work concerned with maximum gains from
finite numbers of elements appears to be that of Uzkov,12 who
considers omnidirectional sources. Explicit solutions are given
in two special cases: (a) for a linear array of n elements spaced
A/2 apart, the maximum gain being shown to be n, and (b) for
an end-fire linear array of equispaced elements as the length
approaches zero, in which case the maximum gain is shown to
approach n2. However, Uzkov says nothing about the relative
excitations of the individual elements. Comparison of Uzkov's
results with those for Schelkunoff arrays of three and four
colinear equispaced elements (Fig. 22 of Reference 6) shows that
for these simple arrays Schelkunoff's method leads to directivities
which are less than 1 db below the maximum.
In the present paper the approach is rather different from that
of Uzkov, who obtains his results by a process involving the
resolution of the radiation patterns of the elements of the aerial
array into components which are orthogonal to each other in
an n-dimensional complex space. Instead, by a variational
process a theorem is developedthe "travelling voltage wave"
theoremfrom which the currents in the individual elements
required for maximum directivity can be derived simply by a
solution of a system of linear equations. The theorem, slightly
modified, also applies to arrays in which the elements are not of
the same kind or of the same orientation. It appears that useful
improvements in gain over conventional arrays can be obtained
without a prohibitive loss of bandwidth or efficiency. Some
experimental results have been obtained which, while not completely satisfactory, provide substantial confirmation of the
theoretical findings.
(2) THEORETICAL TREATMENT
(2.1) Definition of "Resistance Voltage"
In an aerial array having n elements the relation between the
currents and voltages at the terminals of the elements is given43
by a system of equations of the form:
S

Vm = 2 J/ <m

(i)
(2)

In eqn. (2) both components are complex, since the currents


are, in general, complex. They will be called respectively the
"resistance"* and "reactance" voltages. It will be shown in
* It should be noted that as the individual currents of the array can have all
possible phase angles, the resistance voltage is, in general, not in phase with the
current in the individual element and therefore bears no direct relationship to the
power supplied to this element. This power is determined by those parts of the
resistance and reactance voltage which are in phase with the element current; taken
over the whole of the array the contributions of the reactance voltage cancel.

Section 8.1.3 that it is sufficient to specify the resistance voltages,


Tm, in order to ensure that the directivity shall be maximum.
(2.2) The Condition for Maximum Gain of an Array of n Elements
The maximum gain obtainable from an array of a given
number of elements of fixed position, by adjustment of the
currents flowing in these elements, can be calculated by the
method outlined below. Various relations involved are derived
in Section 8.1. To simplify the presentation, it is at first assumed
that the elements are all alike, and that they have the same
orientation. The more general case requires little elaboration
(see Appendix 8.1.4).
For convenience, the terminology refers to a transmitting
array; by the reciprocity theorem, a receiving array will have
precisely the same directional characteristics. It is assumed that
the mutual impedances between the elements are known; the
values of these impedances are discussed in Section 3.1. For
convenience, normalized impedances, defined in the list of
symbols, are used.
Then using the definition of resistance voltage, the power
radiated by the (loss-less) array can be written as
n

P = E i*mrm

(3)

Since only relative values of field strength are of interest, the


expression for the received field at a distant point simplifies to
(4)
m=\

and consequently (see Section 8.1.3) the expression for the power
gain of the array with respect to a reference element identically
similar to the elements of the array also simplifies to
G=\H\*IP
(5)
If the current distribution is such as to give maximum gain, a
first-order variation of this current distribution must leave the
gain invariant (to the same order).
The results of this variational principle can be simply expressed
in terms of the resistance voltage defined in Section 2.1:
The resistance voltages across the terminals of the individual
elements must vary over the array in magnitude and phase as
though they were thefieldstrength values at each element position
of a plane electromagnetic wave travelling across the array in the
direction in which the gain is to be a maximum.
Since only relative values of currents and voltages are of
interest, the amplitude of the resistance-voltage wave can conveniently be made equal to unity. Thus
*" = *-*

(6)

and, by definition of Tm,


(66)

Eqn. (6b) fully specifies the current distribution over the whole
of the array.
The current in the mth element is given by

= 2
/ of the matrix
where the matrix glm is the inverse

(7)

rlm.

(2.3) Expression for Maximum Gain


It follows by substitution from eqn. (6a) into eqn. (4), that
(8)

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS
Comparing this equation with eqn. (3), it is clear that, since P
is real (see Section 8.1.1), the received field is real, and in the
present system of units they have the same numerical value.
Eqn. (5) then simplifies to

305

of an infinitely thin half-wave dipole. The proximity of one


similar parallel element at the spacing used in the array reduces
this value by 5 ohms. The mutual resistance between two
adjacent elements, ignoring effects of other elements, is about
4 ohms higher than the value for infinitely thin elements. Since
G=H=P . .
(9) we shall find that the magnitude of the smallest driving-point
resistance encountered (1-6 ohms) is much less than these
Substituting eqns. (7) and (9) into eqn. (4) gives
variations, it might be expected that the power relations in the
(10) array, and hence its maximum directivity, would be critically
affected by the element thicknesses. However, these effects
m=l /=1
should not be severe, at least as far as the maximum gain is
Since every term in this expression has a conjugate counterpart, concerned, for the following reasons:
this simplifies to
This gain is merely the result of the concentration of the radiated
energy into a small spatial angle, i.e. of the destructive interference
(11)
of the radiation patterns of the individual elements. For superdirective arrays we may expect this interference to be substantially
disturbed if, say, one of the individual patterns were changed.
If the array is made to have maximum gain in the opposite
However, if all the radiation patterns are changed in the same way,
direction the angles will be of the same magnitude but opposite
the overall effect will be merely a multiplication, by a corresponding
sign; hence the maximum gain of an array is the same in any two
factor, of the pattern previously produced. Now, if the array conopposite directions, although, apart from symmetrical arrays, the
sists of identically similar elements, thickening of the elements will
two current distributions are not just the reverse of each other.
produce changes in individual patterns which are to a first order
of approximation the same. It is only to the extent that the elements
In Section 8.2 eqns. (6b) and (9) are converted into forms
do not occupy equivalent positions in the array that this argument
which enable the values of the currents in the elements and the
is not conclusive.
gain of the array to be conveniently computed on a desk-type
It would, of course, be even more satisfactory if it were possible
calculating machine.
to use mutual impedances appropriate to thick aerials. Unfortunately, at present the necessary data do not appear to exist.
(2.4) The Broadside Array
Approximate values have been published for the case of two
The foregoing theory applies to arrays having the maximum aerials.36* However, mutual and self-impedances of thick rod
gain in any chosen direction. The broadside array is a specially aerials (unlike those of infinitely thin aerials) depend on the
simple case, as all the angles 6m are equal. It follows that this location of all other aerials in the vicinity, so that use of these
array will have maximum gain if the resistance voltages at the values for arrays containing more than two elements would
individual aerials are of equal magnitude and are in phase.
probably not give better results than one can obtain from the
simple thin-aerial approach.
(3) LIMITATIONS ON THE APPLICATION OF THE THEORY
It is interesting that the results of calculations37 on a pair of
TO PRACTICAL ARRAYS
elements (one fed and one parasitic), using the second-order
theory, show little difference in maximum-gain values, as the
(3.1) Values of Mutual Impedance
When the elements of an array are omnidirectional sources aerial thicknesses increase, from the results of the simple thin(the case studied by Schelkunoff6 and Uzkov12), the normalized dipole theory.
mutual resistances required, e.g. for eqn. (6b), take a very simple
(3.2) Tolerance on the Currents in the Elements of the Array
form. It is easy to show that for this hypothetical type of
Consider Ieff, the current required to flow in a single element,
element
to produce the same distant field as the array. If in eqn. (4) n
sin
r
is put equal to unity, it is apparent that Ieff has the same
lm
numerical value as H, and also as G and P [eqn. (9)]. A comparison of this current with the magnitude of the largest of the
where 6 is the angular distance between the elements.
In practical arrays designed for the v.h.f. band the elements currents flowing in the elements of the array is of interest, since
commonly consist of thin rod dipoles, half a wavelength long, a change of any current in the array by an amount Ieg- could
and the values of mutual impedance are not so unambiguous, reduce the distant field to zero. We are thus given a measure
of the practicability of constructing the array. For example,
since the current distribution is known only approximately.
23
It is usual, for lack of a better foundation, to assume that the Jordan has designed, by means of Tchebyscheff polynomials,
a
super-gain
array which has an Ie^ of 39 mA and an element
rods are infinitely thin and that the current distribution is sinusoidal. For the case of most interest, i.e. when the elements current of 17 MA. The tolerance on the element current for
are parallel, the mutual impedances can be calculated (see, for this design would8 consequently have to be much better than
example, References 29 and 32) with a reasonable expenditure 1 part in 4 x 10 , which, as Jordan points out, is quite impracof labour; numerical values have also been tabulated over some ticable. In the experimental array described in Section 4, Iejj is
about half the largest element current.
limited ranges.29"31
At first sight it might appear that calculations based on hypo(3.3) Heat Losses in the Aerial Elements
thetical arrays of infinitely thin dipoles could not be expected
to give a useful guide to the performance of practical arrays,
The theory given in Section 2 applies to loss-less aerials. In
since the changes in self- and mutual resistances when going practice, it is necessary to consider the effect of the resistance
from infinitely thin elements to elements having thicknesses of the elements.
encountered in practice may be much greater than the terminal
Neglecting any effects of such losses on the current distriresistances found in super-gain arrays. To illustrate: an iso* It should be noted that the earlier Tai values, 3 ' which are reproduced without
lated element of the experimental array described in Section 4 indication of their approximate nature in Reference 39 (p. 266), are first-order values
and according to King^s "cannot be in good agreement with experimental
has a theoretical self-resistance36 about 13 ohms higher than that only,
measurements."

306

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

bution along the individual elements, it is possible to represent


their effect fully by the addition of a series resistance RL to the
driving-point impedance of the loss-less element. Hence the
currents on the individual elements and the radiation pattern
can be made the same as in the loss-less case. This leaves the
directivity unchanged, although, of course, the gain of the array
is reduced.
For cylindrical half-wave dipoles, assuming sinusoidal current
distribution, it is easy to show that
.

(12)

where / = Frequency, Mc/s.


p = Resistivity, ohm-cm.
D Diameter of dipole rod, cm.
A = Wavelength, cm.
fi = Relative permeability.
The experimental array described in Section 4, which used
aluminium elements of |-in diameter, had an RL of 0 048 ohm.
The power lost in heating the dipoles can readily be calculated
from RL and the currents, and must be compared with the total
radiated power. In the experimental array the loss from this
cause was 0-2 db.
(4) DESIGN AND PERFORMANCE OF AN EXPERIMENTAL
SUPER-DIRECTIVE ARRAY
(4.1) General Considerations
Taylor19 and others have shown generally that the bandwidth
of an array decreases as the size is reduced if the gain is held
constant, and it appears that the decrease is likely to be very
rapid. In fact, it has frequently been suggested that for this
and other reasons it is not worth while to push a design into the
super-directive field. With this consideration in mind it was
decided to build an experimental array, which, with optimum
current distribution, would give about 5 db more gain than the
same system fed as a conventional equal-amplitude end-fire
aerial. It was felt that this represented a useful increase in gain,
while at the same time there was still hope of retaining a sufficient
bandwidth for some practical applications.
(4.2) Theoretical Performance Data of the Array Selected
The array chosen was designed to operate at 75 Mc/s. It was
an end-fire array of four parallel half-wave dipoles, whose
centres were equally spaced along a line 0 6\ long.
The principal results of the numerical calculations are given
in Table 1. The theoretical maximum gain of the array is
l O l d b over that of a single element. Some comparative
figures are offered to show the degree of super-directivity implied
by this gain value. A physically identical array fed with equalamplitude currents phased for maximum field strength in the
end-fire direction would have a gain of 4-6 db; a continuous
array of equal-amplitude doublets phased according to Reid's
"optimum" distribution13 should produce 6-5 db gain over a
single doublet. The gain of four omnidirectional sources having
the same spacing, and fed as described by Schelkunoff,6 is
given in Fig. 22 of his paper. Using the same method as
in Section 8.4, it is estimated that, if this array consisted of
half-wave dipoles the gain would be about 9 db. A 6-element
Yagi array about 2\ times as long as the experimental array
was found by Fishenden and Wiblin40 to have a gain of 9 db.
Fig. 1 shows the theoretical radiation pattern of the array, in
the plane at right angles to the elements. For comparison, the
main lobe of a physically identical array fed with equal amplitude
currents, having their respective distant field components in the

Table 1
RESULTS OF THEORETICAL COMPUTATION

Element number
1
2
3
-2-157 +1-473 +1-473

4
Real parts of currents..
-2-157
Imaginary parts of currents
-7-864 +18-327 -18-327 +7-864
Moduli of currents ..
8-155
18-386
18-386
8-155
Real parts of driving
point voltages
..
1-398 -12-694 +14-312 -2015
Imaginary parts of driving point voltages .. -0-976 +2-729 +1-553 -2-878
Driving-point resistances,
ohms
+5-1
+6-8
-1-6
-20-1
Driving-point reactances,
ohms
+14-4
+51-2
+57-2
+24-2
Input powers (relative) +340
+2288
-539
-1336
Power gain relative to A/2 dipole . . . 10-3 (10-1 db).
Current in A/2 dipole to give same field as array in same units as
above currents, 10-3.
Total radiated power (relative) . . . 73-1 x 10-3 = 753.
Power lost as heat in elements (relative) . . . 38 (0-2 db).
end-fire direction in phase (i.e. an "orthodox" end-fire arrangement), is plotted on the same diagram. The scales of the two
patterns are such that the maximum field strengths are the same.
(4.3) Feeder System
Reference to Table 1 shows that, in order to maintain the
specified current distribution, elements 3 and 4 must deliver
power to the feeder system. To obtain the full gain this power
must be returned to elements 1 and 2. A loss-less feeder network
supplying the required driving-point voltages automatically
takes care of this requirement. A discussion of the problem of
transmission lines terminated by impedances with negative
resistive parts will be found in Section 8.3.
Each element was fed through a network of lumped reactances
arranged as shown in Fig. 2. The networks were then connected
in parallel, using half-wavelengths of balanced twin feeder, and
the whole arrangement was fed with coaxial cable through a
balance-to-unbalance transformer.
The values of XP and Xs must be chosen for each of the four
networks so that, when the same input p.d. is applied to each
network across the points T, the currents in the respective
elements will be properly related to one another in magnitude
and phase. The number of ways in which this can be done is
theoretically unlimited. In practice, the reactance had to be
made from small components. Furthermore, it was convenient
to arrange that the admittance of the four elements in parallel
would approximately equal the characteristic admittance of the
feeder.
This method of feeding was selected for the ease with which
adjustments could be made. However, the values of the components chosen resulted in very high standing-wave ratios on
the half-wavelengths of feeder; in one case it was more than 80.
The components consisted of small coils or capacitors connected in parallel with trimming capacitors. Before the networks were assembled the reactance of each group of components was measured at the design frequency of the array, using
a slotted coaxial line. The trimming adjustment was provided
to allow for the difference between the actual driving-point
impedances of the elements of the array and those calculated
on the basis of infinitely thin half-wave dipoles.
(4.4) Gain and Bandwidth Measurements
The gain was measured by direct comparison between the
power delivered to the array and that delivered to a half-wave
dipole to produce the same field at a receiving aerial. The

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

150

160

170

180

170

160

150(

307

140c

130 c

4-element
super gain
array

Orthodox"end-fire array
having- elements in same
position as
\
the "super gain"array

50'

50'

Fig. 1.Theoretical radiation pattern of the experimental 4-element super-gain array.


Dipole arms
relative powers were measured with the slotted coaxial line and
due allowance was made for the losses in the coaxial cables
leading to the array and dipole respectively.
By systematic adjustments of the trimming capacitors of the
matching networks a gain of 8 7 db was obtained. This was
considered to be satisfactory, since the calculated losses in the
half-wavelength of twin feeder and in the dipoles themselves
Balanced
were respectively 0-6 and 0-2 db. It seems probable that the
feeder
remaining 0 6 db by which the measured performance falls
short of the theoretical prediction could be largely accounted for
Fig. 2.Matching arrangement for mth element.
by losses in the coils and capacitors of the matching sections.
Without making any further alterations to the matching net(4.5) Possible Developments
works, the gain was measured over a frequency band. The
Several
lines
suggest
themselves along which further developresults are shown graphically in Fig. 3. Although small, this
bandwidth is sufficient for some practical purposes, and may well ment could proceed. It is evident that for practical applications
the feeder system used in the experimental array is too cumberbe increased by changes in the feeder system.

308

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

|7

\\

x"
\

\
\

\
\

h
74
75
Frequency, Mc/s

Fig. 3.Variation of gain with frequency of 4-element


experimental array.

(5) S O M E NUMERICAL RESULTS


(5.1) Arrays of Three Parallel Half-Wave Dipoles: Length of
Array Varying
F o r three parallel equispaced half-wave dipoles arranged along
a line the m a x i m u m gain can b e written in quite a simple f o r m :
2TT

/2TT

\ ~|

- y d c o s i/r J + y\ cos (-ydcos


ipj
.
where d = Element spacing.
tjj = Angle between line of array a n d line of shoot.
a, jS, y = Algebraic functions of the m u t u a l impedances.

0-4 0-6 0-8 10 1-2 1-4 1-6


Total length of array, wavelengths

18

20

Fig. 4.Maximum gain of 3-element array of equispaced parallel


half-wave dipoles.

some. F o l d i n g the dipoles would facilitate the design of suitable


matching networks. Moreover, the lengths of the feeder sections
connecting successive dipoles could with advantage be reduced,
a n d this might be done, e.g. by dielectric loading.
A n o t h e r a p p r o a c h t o the p r o b l e m of improved design is suggested by Figs. 6 a n d 7, which show t h a t the m a x i m u m gains of
3-element a n d symmetrical 4-element linear arrays of constant
total length change very little as the positions of the inner
elements vary. A l t h o u g h n o c o m p u t a t i o n has been d o n e for
non-symmetrical 4-element arrays, one might expect that, in
general, the m a x i m u m gain of the present experimental array
will be substantially independent of the positions of the two inner
elements. If this is so, there m a y b e considerable advantage in
varying these positions, since the input powers a n d driving-point
impedances required for m a x i m u m gain should alter considerably
as the configuration of elements changes. It might, for example,
be possible to find a n arrangement in which one of the elements
is parasitic, or in which two elements carry powers of equal
magnitude a n d opposite sign (in which case they could, of course,
be connected together, forming a parasitic pair). If b o t h these
conditions could be obtained simultaneouslywhich is n o t out
of the question, since two degrees of freedom are involvedthe
feeding p r o b l e m would b e greatly simplified, because only one
feed point would remain.

0-2

(13)

Fig. 4 shows the variation of m a x i m u m gain with the length


of array, in the broadside a n d end-fire cases, a n d Fig. 5 shows
t h e variation against angle when d is 0-15A.
F o r the cases

End-fire.
Broadside.

covered by these Figures it is clear t h a t the m a x i m u m gain h a s


its largest value in t h e end-fire case, when the array length
becomes zero. It can, in fact, be shown by direct c o m p u t a t i o n
t h a t this is generally true for t h e 3-element equispaced linear
array.
Fig. 6 shows the variation in m a x i m u m gain of the array of
length 0-3A as the inner element is displaced towards either
end. Since t h e m a x i m u m gain of a n array remains unchanged
on reversal of direction of m a x i m u m radiation (see Section 2.3),
the resulting curve is symmetrical a b o u t t h e central position.
It is seen that the m a x i m u m gain is almost independent of
the position of the inner element, t h e extreme variation being
a b o u t 0 1 d b . T h e largest value of m a x i m u m gain occurs when
the inner element is central.
(5.2) Arrays of n Half-Wave Dipoles: Length of Array Fixed
A n o t h e r case t h a t has been studied numerically is t h a t of n
equispaced parallel half-wave dipoles, with the length of the
array equal to A. Values of t h e m a x i m u m gain are given in
Fig. 8. Also plotted in this F i g u r e a r e the m a x i m u m gains
in the broadside direction, a n d in addition the gains when
the array is fed in the o r t h o d o x end-fire m a n n e r , i.e. with equalamplitude currents, such t h a t the distant field c o m p o n e n t s in the
end-fire direction are in phase. I t is interesting to notice t h a t
the maximum-gain curve (end-fire case) in this Figure a p p r o a c h e s
the limiting curve of Fig. 9 (see Section 5.3) quite rapidly. F o r
seven elements the respective gains are a b o u t 14-8 a n d 14-9 d b .
(5.3) Limits of M a x i m u m Gain
It would b e useful to k n o w t h e m a x i m u m gain obtainable from
a particular n u m b e r of elements arranged in any way, as this
would set a limit to the possible i m p r o v e m e n t which might b e
made.
T h e trend of numerical results suggests t h a t this m a x i m u m is
obtained when the elements are arranged as a n end-fire array
a n d the total length tends to zero. T h e full-line curve in Fig. 9
shows this limit. This curve is derived from U z k o v ' s expression,
G n2, for the m a x i m u m gain of n equally spaced omnidirectional sources by a m e t h o d described in Section 8.4.

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS
140

150

160

170

180

170

160

150

309

140

130'

60'

50c

50

30

40

20

10

10

20

30

40

Fig. 5.Variation of maximum gain of a 3-element array with "angle of shoot."

0-6X
m

8-3

6 _

em

V-o"3 6
buojy

8-1

80

'3-s

0
005 01 015 0-2 0-25 0 3
Distance of centre dipole from one end of
array, wavelengths

Fig. 6.Variation of maximum gain of a 3-element end-fire array


with position of middle element.

005
010
015
0-20 0-25 0-30
Element spacing , 0, wavelengths
Fig. 7.Variation of maximum gain of a symmetrical 4-element
end-fire array with position of inner elements.

310

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

\
Y
/
/
/
/1
.
1
//

16

sive losses. An experimental array has substantially confirmed


these conclusions and has shown that the bandwidth of such an
array need not be prohibitively small.

^14

Il2
'a!

10
to
o

(7) BIBLIOGRAPHY
Super-Gain Arrays
(1) OSEEN, C. W.: "Die Einsteinsche Nadelstichstrahlung und die
Maxwellschen Gleichungen," Annalen der Physik, 1922, 69,
p. 202.
(2) HOWELL, W. T.: "Electromagnetic Waves from a Point Source,"
Philosophical Magazine, 1936, 21, p. 384.

(3) HANSEN, W. W., and WOODYARD, J. R.: "A New Principle in

(4)
(5)

/
9
10
4
5
6
7
Number of elements
Fig. 8.Array of equispaced parallel half-wave dipoles.

(6)
(7)
(8)

End-fire; maximum gain.


Broadside; maximum gain.
. . . . End-fire; field components in phase.

(9)
20

(10) BOUWKAMP, C. J., and DE BRUIJN, N. G.: "The Problem of

18

/
/

| 16

/
/

^14

/
/

Il2

/
/
/ /
/ /
y /

'
' A

/
/

/
/

/
/

,/
/

/
/

'I 2
C5

Directional Antenna Design," Proceedings of the Institute of


Radio Engineers, 26, p. 333.
FRANZ, K. "The Gain and the (Riidenberg) 'Absorption Surfaces'
of Large Directive Arrays," Hochfrequenztechnik und Elektroakustik, 1939, 54, p. 198.
FRANZ, K.: "The Improvement of the Transmission Efficiency
by Directive Aerials," ibid., 1941, 57, p. 117.
SCHELKUNOFF, S. A.: "A Mathematical Theory of Linear
Arrays," Bell System Technical Journal, 1943, 22, p. 80.
FRANZ, K.: "Remarks on the Absorption Surfaces of Directive
Aerials," Hochfrequenztechnik und Elektroakustik, 1943, 61,
p. 51.
LA PAZ, L., and MILLER, G. A.: "Optimum Current Distributions
on Vertical Antennas," Proceedings of the Institute of Radio
Engineers, 1943, 31, p. 214.
BOUWKAMP, C. J.: "Radiation Resistance of an Antenna with
Arbitrary Current Distribution," Philips Research Reports,
1946, 1, p. 65.

/ /
/

2
3
4 5 6 7 8 10
Number of elements
Fig. 9.Maximum end-fire gain of array of equispaced elements.
Array of half-wave dipoles.
Array of omnidirectional sources.

Although Uzkov's result applies only to equally spaced


radiators, it is possible to show, by the method of the present
paper, that for four symmetrically arranged end-fire elements
the limiting gain is independent of the ratio of the spacings.

Optimum Antenna Current Distribution," ibid., p. 135.


(11) DOLPH, C. L.: "A Current Distribution for Broadside Arrays
which Optimizes the Relationship between Beam Width and
Side-lobe Level," Proceedings of the Institute of Radio Engineers,
1946, 34, p. 335.
(12) UZKOV, A. L: "An Approach to the Problem of Optimum
Directive Antennae Design," Comptes Rendus de VAcademie
des Sciences de I'U.R.S.S., 1946, 53, p. 35.
(13) REID, D. G.: "The Gain of an Idealized Yagi Array," Journal
I.E.E., 1946, 93, Part IIIA, p. 564.
(14) RIBLET, H. J.: Discussion, Reference 9, Proceedings of the
Institute of Radio Engineers, 1947, 35, p. 489.
(15) GILLETT, G. D.: "Analysis of Effect of Circulating Currents on
the Radiation Efficiency of Broadcast Directive Antenna
Designs" (summary only), ibid., 1948, 36, p. 372.
(16) RIBLET, H. J.: "Note on the Maximum Directivity of an Antenna,"
ibid., p. 620.
(17) WiLMOTTE, R. M.: "Note on Practical Limitations in the
Directivity of Antennas," ibid., p. 878.
(18) BELL, D. A.: Discussion on Reference 14, ibid., p. 1134.
(19) TAYLOR, T. T.: Discussion on Reference 14, ibid., p. 1134.
(20) WOODWARD, P. M., and LAWSON, J. D.: "The Theoretical Pre-

cision with which an Arbitrary Radiation Pattern may be


obtained from a Source of Finite Size," Journal I.E.E., 1948,
95, Part III, p. 363.
(21) CHU, L. J.: "Physical Limitations of Omni-Directional Antennas,"
Journal of Applied Physics, 1948, 19, p. 1163.
(22) BELL, D. A.: "Gain of Aerial Systems," Wireless Engineer, 1949,
26, p. 306.
(23) JORDAN, E. C.: "Electromagnetic Waves and Radiating Systems"
(Prentice-Hall, New York, 1950), p. 445.
(24) TAYLOR, T. T., and WHINNERY, J. R.: "Applications of Potential

Theory to the Design of Linear Arrays," Journal of Applied


Physics, 1951, 22, p. 19.
(6) CONCLUSIONS
(25) FREEDMAN, J.: "Resolution in Radar Systems," Proceedings of
the Institution of Radio Engineers, 1951, 39, p. 813.
The problem of designing aerial arrays with increased gain
N.: "A Note on Super-Gain Antenna Arrays," ibid.,
has been approached afresh by an examination of the condition (26) YARU,
p. 1081.
under which an array of a finite number of elements and of (27) "Supergain Antennas," QST, 1951, 35, p. 46.
fixed geometry has to be operated in order to have maximum (28) AIGRAIN, P.: "Les antennes super-directives," L'Onde Electrique,
1952, 32, p. 51.
gain in some specified direction. This condition has been
formulated so that the current distribution in the array can be General References
readily calculated from a knowledge of the mutual resistances (29) BARZILAI, G.: "Mutual Impedance of Parallel Aerials," Wireless
Engineer, 1948, 25, p. 347.
between the elements. The theory has been applied to the
numerical calculation of certain simple arrays. It appears that (30) SMITH, R. A.: "Aerials for Metre and Decimetre Wavelengths"
(University Press, Cambridge, 1949), p. 43.
for arrays of a given size directivities greater than those obtained (31) MEDHURST, R. G., and POOL, S. D.: "Mutual Impedance of
by conventional design methods can be achieved without excesParallel Aerials," Wireless Engineer, 1951, 28, p. 67.

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

311

(32) CARTER, P. S.: "Circuit Relations in Radiating Systems and Appli- plane, the received field is, making the same simplification as
cations to Antenna Problems," Proceedings of the Institute of for eqn. (4),
Radio Engineers, 1932, 20, p. 1004 (see also 1948, 36, p. 1003).
(33) MILNE, W. E.: "Numerical Calculus" (University Press, Princeton,
HS = IS
(16)
1949), pp. 15-35.
(34) KING, R., and MIDDLETON, D.: "The Cylindrical Antenna;
The radiated power from the reference element is
Current and Impedance," Quarterly of Applied Mathematics,
1946, 3, p. 302; 1946, 4, p. 199; and 1948, 6, p. 192.
nr = I2
(35) TAI, C. T.: "Coupled Antennas," Proceedings of the Institute of
Radio Engineers, 1948, 36, p. 487.
since in the normalization adopted, rs is unity.
(36) KING, R.: "Self- and Mutual Impedance of Parallel Identical
Antennas," Cruft Laboratory, Harvard University, Technical
Report No. 118, 1950 (shortened version in Proceedings of the (8.1.3) The Voltage Theorem for Maximum Gain for n Equal and
Indentically Oriented Elements.
Institute of Radio Engineers, 1952, 40, p. 981).
(37) KING, R.: "A Dipole with a Tuned Parasitic Radiator," Pro-'
The power gain of the array with respect to the reference
ceedings I.E.E., 1952, 99, Part III, p. 6.
(38) CROUT, P. D.: "A Short Method for Evaluating Determinants element is
and Solving Systems of Linear Equations with Real or Complex
iHl 2 I1
\H\2
Coefficients," Transactions of the American I.E.E., 1941, 60,
p. 1235.
\HS\2P~ p
u/;
(39) KRAUS, J. D.: "Antennas" (McGraw-Hill, New York, 1950).
(40) FISHENDEN, R. M., and WIBLIN, E. R.: "Design of Yagi Aerials,"
Write
U=\H\2
(18)
Proceedings I.E.E., 1949, 96, Part III, p. 5.
(41) SAXTON, J. A.: "Determination of Aerial Gain from its Polar
Diagram," Wireless Engineer, 1948, 25, p. 110.
so that
GP= U
(19)
(42) LAMONT, H. R. L., ROBERTSHAW, R. G., and HAMMERTON, T. G.:
If the currents in each element are slightly perturbed from the
"Microwave Communication Link," Wireless Engineer, 1947,
values which makes G a maximum,
24, p. 323.
(43) TERMAN, F. E.: "Radio Engineers' Handbook" (McGraw-Hill,
New York, 1943).
(20)
8G = 0
(8) APPENDICES
(8.1) Derivation of the Condition for Maximum Gain
(8.1.1) The Power Radiated.

hence

G8P = 8U

(21)

From eqn. (18)

U = HH*

(22)

Therefore

8U = H&H* + H*8H

The radiated power is given by

P = [V'J'm

= 2&H8H*

+V ' ^ ]

. . . . (14)

Substituting from eqn. (1)

P=

(23)

Limiting the variation to the current Ir, it follows


eqn. (4) that SH = 8 l r e ^ ;
8 U = i M H h l U ' ^

hence

from

. (24)

Again from eqn. (15),

U'nJUirlm - I','xln) + /;; (V{rlm + /J

m = \ 1=1

P=

= [(/;/; + W H J

m= i

l*m%hr,m+

8l*mIirlm . (25)

1=1

1=1

m=l

771=1 1 = 1

= r?8it + rm8i*m

since xlm = xml

/=1

ltnhrlm

(15)

m=\ 1=1

= i*mrm

0)

m= l

(26)

m= l

If the variation is again limited to J r ,


8P=r*Sl

+ rrSl*

. (27)

= 2Tr8l*r

(28)

Eqn. (15) follows since rlm = rml and


Substituting eqns. (24) and (28) into eqn. (21),

i / [(W-h,] = 0
Note that the power input to the zwth terminal is the real part of
ImVm, and not of J j , ^ . The latter expression omits all those
contributions to the power input which are due to mutual
reactance and which cancel in the summation over the whole
array.
(8.1.2) The Received Field.
Select a reference plane perpendicular to the direction in
which the array is to have maximum gain.
The received field at a distant point in this direction is given
by eqn. (4).
If the array is replaced by a reference element similar in all
respects to the elements of the array and situated in the reference

GMTr8l*

= &H8l*~JOr

(29)

Since eqn. (29) must be valid for any 8lr let 8lr first be chosen
so that T*8lr is imaginary. The left-hand side of eqn. (29) will
then be zero, and hence also the right-hand side, i.e. both
expressions following the^? sign will be purely imaginary. Then,
advancing the argument of 8lr by 77/2 will make both expressions real, allowing t h e ^ signs to be removed from the equation.
It follows that
Tr = ~e-J0r

(30)

Since the gain of the array is independent of the power level,


the modulus of eqn. (30) may be made unity, and eqn. (6a)
follows.

312

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

(8.1.4) Generalization to Elements Different in Form or Orientation.


Eqn. (15) for the power radiated is still valid, and, by definition, so is eqn. (16). By introducing suitable factors, hm, the
distant field can still be expressed in the form
=

(3D

2'

where
'In

'21

cos 6{
c o s &)

'22

A= -

771=1

hm can always be made real by ascribing to each element an


appropriate position value 6m.
Now, introducing new currents by the relations
It

',,

cos 62

. . COS0rt

'11

'12

'17,

r2

n2

'T1

cos
C

(32)

AT

COS d{

'22

'21

eqn. (31) can be rewritten as

sin
sin

(33)
m=\

and eqn. (15) as


',,2

=S
771 =

(34)

2
1

s i n $>

(35)

The problem is now reduced to the same form as that discussed in Section 2. The maximum gain of the array, and the
modified currents Pm, can be calculated using the relations given
in that Section.

sin c
0

\n

n\

('ii = '22 =

(8.2.1) Separation of Equations into Real and Imaginary Parts.

2 rlrlm = cos 0n

i=i
n

m = l(l)n .

(36)

-sin

Then, from eqns. (4) and (9)

;;(-sin 0 j ]

(37)

For arrays of not more than about ten elements, eqns. (36)
can readily be solved on a desk-type calculating machine, using,
for example, Crout's method.33'38
If the array consists of elements symmetrically spaced along a
straight line, it is advantageous to choose the reference plane so
that it passes through the centre of the array. The values taken
by the cosine and sine functions in eqn. (36) are then respectively
symmetrical and skew symmetrical. It follows that the values
of /,' and 1'^ show the same symmetries, and the number of
simultaneous equations is halved.
(8.2.2) An Alternative Expression for Gain.
From eqns. (36) and (37) the gain could, of course, be written
explicitly in terms of the mutual resistances rlm and the angles
6m. In fact, with a little rearrangement, we obtain the following
form:
&
(38)

2n

and A =

In numerical work it will usually be convenient to solve the


set of equations (6b) directly for the currents, rather than to
carry out the intermediate operation of evaluating the inverted
matrix glm. The procedure for identical elements is as follows.
Separating eqn. (6b) into real and imaginary parts gives the
systems of equations

r22

(8.2) Computation Procedure

2 r,'rlm=

sin

l = \

r]m = rlm\hmh{

with

sin

n2

rt

= rnn = 1)

When the currents are required as well as the gain, eqn. (38)
has no particular merit, since the right-hand side of eqn. (37)
can be evaluated in a single operation on a calculating machine.
If, however, the gain only is required and the current magnitudes
are not of interest, e.g. if it is desired to examine the trend of
the gain as the element spacing or the number of elements
varies, there is some advantage in using eqn. (38). It is not
necessary in a particular case to evaluate the denominators
separately, since in Crout's method the values of the two terms
of G come out directly as the bottom right-hand terms of the
modified matrices obtained from the respective numerators.33
Eqn. (38) is particularly useful for dealing with limiting cases in
which elements coalesce, e.g. in the 3-element case shown in
Fig. 6, where the inner element coalesces with either of the outer
elements.
The approximations to the mutual resistances required for
this computation are reproduced here because they seem not to
be generally known:

m&2l-sinmp

sin [m<\/(p2 +12)] > ohms


(39)

to second order in S,
where i?p = Mutual resistance when spacing is p.
i? P+ s = Mutual resistance when spacing is p + 8.
m = 2TT/A

/=A/2

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

If Rl2 is the mutual resistance when the spacing approaches


zero,
i?12 = 2 9 - 9 7 9 ^ - ^ - + ^
where

Denoting the components of forward and backward travelling


waves by the subscripts / and b, we have
E=Ef

(40)

. . .)ohms

313

+ Eb

(44)

I=Ir-h

p Spacing

(45)

hence, defining the ratio Ejl as the load impedance Z,,

cj> = 2-nplX
r = 2-43765...

IZ,=

Ef + Eb

IZ0 = IfZ0-lbZ0=Ef-Eb

K= HL + A/16TT2 = 0 0271659 . . .

and
As far as the square term, eqn. (40) is given in Reference 39
(p. 267).

(46)

. (47)

/ = i / ( Z / + Z0)

(48)

Eb = y(Z,-ZQ)

(49)

The reflection factor is then

(8.2.3) Driving-Point Impedance.


Driving-point impedances are required for the design of a
?
(50)
feeder system and for the estimation of the heat losses in the
E
operation of the array.
Dividing both sides of eqn. (1) by Im we obtain the driving- with no restriction on the value of Zt. From Fig. 10 it follows
that, when Z, has a negative real part, |p| is greater than unity,
point impedance offered by the mth element as
1 "
m +JXm = 7 ~ 2 hZlm

. (41)

Then it is readily shown, using eqn. (36), that


n

Urn CS " f f l - 4 S I'l'Xim ~ Im Sin em + I'm 2 I'iX,m]


1

'

tn

. . . .

(42)

and
=

'"

[Im 2 I Mm ~ Im Sm 8m - lm COS dm + Im 2 11 Xim]


'-'
'='
(/'2 +1" 2 )
. . . .

(43)

Fig. 10.Vector diagram to show that impedances in the left-hand


plane are associated with reflection factors
Zi - Z 0 > 1

+Z0

In these equations the impedances are normalized and must


be multiplied by Rs to convert the units to ohms.

i.e. the reflected wave is larger than the incident wave, and thus
more energy is returned to the line than has arrived. This, of
course, is in agreement with the fact that for such values of Zl
(8.3) Transmission Lines Terminated by Negative Resistances
the energy consumption is negative.
Charts and formulae for transformation of impedances along
When optimum current distributions for arrays of aerials are
worked out it frequently happens that the driving-point resistance transmission lines are usually designed for use with values of
of one or more of the aerials is found to be negative, i.e. these \p\ less than unity, so that, at first sight, values of |p| greater
elements are drawing power from the array. For maximum than unity might appear inconvenient. But if a transmission line
gain this power has to be fed back to the other elements.* is cut at any point and it is found, say, looking to the right of
Suppose that each element is connected to a transmission line, the cut, that the line is here working into an impedance Zh
possibly through a network of some sort, and that these trans- there will be at the left of the cut an impedance Zt. This is
a simple consequence of the fact that impedance, in the sense
mission lines run to a common feed point.
When the standing-wave pattern on these transmission lines used here, is merely the ratio of voltage to current* and that
is considered a difficulty appears in connection with those lines both pairs of terminals have the same voltage across them but
which feed power from the array towards the feed point. It is the currents, while equal, are of opposite sign; if the current
usual to determine the standing-wave pattern from the "load" flows into one pair of terminals it must flow out of the other pair.
end of the line and work back towards the generator. It would If the impedance Z7 is terminating the line and has a reflection
therefore seem that in those cases just mentioned further analysis coefficient \p\ greater than unity, it is clear from eqn. (50) that
would be required in order to find the load presented to the line the impedance Zt to the left of the cut has a reflection coat this common feed point. However, this is not so; the standing efficient, \\\p\, less than unity.
wave pattern on all transmission lines is uniquely determined by
The transformation of the nominal load Z ; along a transthe load impedance (here, the aerial driving-point impedance) mission line towards the nominal generator (transformation
even if the real part of this quantity turns out to be negative, from loading-end impedance towards sending-end impedance)
and transmission-line equations can be applied in the usual way, can thus be solved by inserting the impedance Z/ in a formula
as can be seen from the following analysis of transmission line or chart intended for transformation from the actual generator
fundamentals.
towards the load (transformation from sending-end impedance
towards load impedance).
If the aerial is to be designed only for maximum directivity, without regard to the

losses incurred, it is, of course, much simpler to dissipate this energy by terminating
the elements concerned with impedances equal to the negative of their respective
driving-point impedances.

* This impedance must not be confused with the "internal" impedance which
informs us of the change in voltage associated with a change of current.

314

BLOCH, MEDHURST AND POOL: A NEW APPROACH TO THE DESIGN OF SUPER-DIRECTIVE AERIAL ARRAYS

(8.4) Method of Deducing the Gain of Zero-Length Arrays of


Half-Wave Dipoles from Uzkov's Result for Omnidirectional
Sources
An exact result found by Uzkov for linear end-fire arrays of
equispaced omnidirectional sources states that the limiting value
of maximum power gain is n2, with respect to a single omnidirectional source, where n is the number of elements. The
dotted line in Fig. 9 is based on this expression. For half-wave
dipoles the corresponding values can fairly easily be obtained
by the method developed in the present paper, for a small
number of elements. The limiting maximum power gains for
two and three elements (with respect to a half-wave dipole) are,
respectively,
1+ r
(51)
(1 (52)
1 - 2AKr
where r and K have the same significance as in eqn. (40).
Unfortunately, the algebra needed for a larger number of
r -

elements becomes very tedious. However, a good approximation for the half-wave dipole system can be obtained from
Uzkov's omnidirectional-source expression as follows: Saxton's
formula41 gives an approximate value for the half-power beam
width of the Uzkov array. On the assumption that the slope of
the pattern in this region is given closely enough by Lamont's
formula,42 the modification of this half-power beam width when
the omnidirectional pattern is multiplied by the half-wave dipole
pattern can be calculated. Were it assumed that in every plane
containing the direction of maximum radiation the pattern is
changed in the same fashion, renewed application of the Saxton
formula would give the gain of the array thus modified. However, since the pattern remains unchanged in the plane at right
angles to the plane of polarization, the final approximation is
obtained by taking the mean of this and Uzkov's result.
Since the error in the final result should decrease as the beam
width decreases, a stringent test of this device is to compare the
result so obtained for three elements with the exact value given
by eqn. (52). The figures are respectively 8-2 and 8-4 db.
Values so obtained are plotted in Fig. 9.

"RECENT PROGRESS IN RADAR DUPLEXERS, WITH SPECIAL


REFERENCE TO GAS-DISCHARGE TUBES"
RADIO SECTION INFORMAL LECTURE, 24TH NOVEMBER, 1952
Mr. P. O. Hawkins began his lecture by saying that the might be possible to derive improved performance with balanced
function of a microwave radar duplexing circuit was to permit duplexers.
the use of a single aerial for both transmission and reception.
Since the radar system performance would be impaired if energy
In the discussion which followed, some speakers confined
were lost into the receiver during transmission, the isolation themselves to seeking further elucidation of points raised in the
between transmitter and receiver had to be of a high order. It lecture, while the majority indicated the lines of their own
was possible to achieve this isolation even though the transmitter thought on the problems and their particular experience in the
and receiver were simultaneously connected to the aerial. development of various aspects of design. More time was
The duplexing action was then achieved by means of polariza- devoted to discussion of the improvement of conventional
tion-sensitive duplexers, which were passive circuits and did not arrangement in particular respects, such as in the choice of gasuse gas-filled tubes. The polarization-sensitive systems might in filling or of priming-electrode material, than to the more radical
special cases be used with advantage in pulsed radar systems, methods outlined in the latter part of the lecture; these were
even though the transmitter and receiver functioning was time- nevertheless generally regarded as being promising.
separated. In general, however, the use of microwave gas-filled
Much importance was attached to reliability of system operaswitches [transmit-receive (t.r.) switches] led to simpler duplexing tion and to satisfactory life of mixer crystals and t.r. switches.
systems. The lecture was concerned with advances in duplexers A variety of expedients were indicated for limiting the transmitter power from the transmitter which reached the mixer
using t.r. switches, i.e. active duplexers.
Although, in principle, for most applications a wide-band t.r. crystal of the receiver and for minimizing deterioration in the
switch used with a balanced duplexer was satisfactory, in practice discharge gaps, owing to sputtering and detuning, with conseit was difficult to maintain the requisite degree of isolation quent unreliability of operation.
For the order of transmitter power used in the more conbetween transmitter and receiver.
Some of the reasons for this were now understood, and it was ventional pulsed equipments, the problem of providing suitable
shown that the difficulties were due to compressing three distinct types of t.r. switch giving adequate protection of the crystal
duplexing functions into one switch. The operation of the t.r. mixer was not considered to present undue difficulty. Where
switch and the physical processes that led to imperfect operation higher powers were required, however, improvements became
were described. It was suggested that the passive t.r. switch extremely valuable, particularly where exceptional reliability
was not the best form of valve for duplexing, and that a system and continuity of operation were desired. Increasing interest
in which the three functions were separated and in which in broad-band operation and the consequent need for relatively
synchronous attenuation was introduced into the duplexer was low-Q-factor resonators meant that special attention needed to
be given to reliable means of initiating the discharge, which
superior.
A number of experimental solutions to this problem were being set up the reflection barrier against energy from the transtried. An experiment with switching tubes, in which the three mitter.
The improvement in performance and reliability offered by the
functions were physically isolated, was described. Attenuation
synchronized with the transmitter was introduced into the more complex arrangements envisaged in the lecture, however,
system with one of these tubes. This work had shown that it needed to be sufficient to justify the excess bulk and weight

Вам также может понравиться