Академический Документы
Профессиональный Документы
Культура Документы
Compone nt
Testing
Edited by W F Waller
Macmillan Education
Published by
THE MACMILLAN PRESS LTD
London and Basingstoke
Associated companies in New York Toronto
Dublin Melbourne Johannesburg and Madras
Preface
This book shows the student how to test the
working parameters of electronic components. It
is, of course, appreciated that manufacturers carry
out exhaustive testing of their products and in so
doing employ a range of sophisticated equipment
which is not generally available to the component
user; the contributors to this book therefore confine themselves to examining component evaluation exercises which are within the powers of the
component user and should provide him with
sensible and accurate data.
Chapters 1, 2 and 3 deal with conventional
electronic components-resistors, capacitors and
coils and transformers-and their treatment centres
mainly upon these devices being used within the
normal frequency ranges. The following three
chapters deal with three main classes of semiconductor devices before the book moves on to
integrated circuitry, which has been divided into
analogue and digital operations. Chapters on
gridded values and relays and switches complete
the component coverage.
A companion volume in the same format,
Electronic Measurements, shows how electrical
values may be accurately measured.
Conte nts
Chapter
9
Resistors
Page
9
2 Capacitors
R J Everitt BSc
Chief Chemist
Dubilier Limited
3 Wound Components
18
4 General Purpose
Semiconduct ors
26
34
R D Eveleigh
Ratings and Evaluations Engineer
AEI Semiconductors Limited
43
A Hale BSc
Group Leader
Mullard Limited
56
A R Townsend
Quality Assurance Engineer
G EC Semiconductors Limited
63
R S Whiskin BSc(Eng)
Product Marketing Manager
G EC Semiconductors Limited
9 Electronic Tubes
71
83
J J Langridge
Quality Control Manager
Keyswitch Relays Limited
Glossary of Terms
89
Chapter 1
Resistors
approved to it. It is evident that whilst this philosophy suffices for most 'general purpose' requirements, there are other cases when every aspect of the
resistor is of interest and may affect the ultimate
choice of type.
Table 1 shows the parameters of a resistor to be
assessed for a typical BS9000 approved type, and the
way in which the assessment is made by testing.
Certain tests can be made by customers as a 'goods-in'
check and these are indicated in the second column,
whilst the manufacturer will make other tests periodically as shown. The test parameters are given in order
of importance for general purpose use.
RESISTANCE VALUE
It is usual for a resistor to be specified by a 'Nominal
Value' and a tolerance on that value, which may be
wide in the case of certain composition types at
20%, or as tight as 0.01% for precision types. This
tolerance has been known as a 'selection tolerance'
which was appropriate for certain types of resistor
that could not be manufactured to a tight tolerance,
and therefore a precise initial value could only be
achieved by selection after a resistance measurement.
However, many resistors are now manufactured to
the required tolerance and a selection process is not
used. Furthermore, resistors which fall outside of this
tolerance may be abnormal in some respects and
therefore may be rejected as being potentially hazard-
By Customer
Test
Parameter
Resistance value
Solderability of leads
Solvent resistance
Dimensions
Visual appearance & marking
Voltage proof
Temperature characteristic
Robustness
Goods-In
lot-by-lot
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
Climatic
3-Monthly
*
*
*
3 Yearly
or Type Test
Weekly
Noise
Non-linearity
R F characteristics
Effect of soldering heat
*
*
*
Thermal emf's
Temperature cycling
Vibration
"
Lead
Wire
dia.
Resistance
of 1 em
of Wire
0.4mm
1.4
milliohms
0.6mm
10!2
100!2
1.4%
0.14%
0.014%
0.0014%
0.61
milliohms
0.61%
0.06%
0.006%
0.0006%
0.8mm
0.35
milliohms
0.35%
0.035%
0.0035%
0.00035%
1.0mm
0.22
milliohms
0.22%
0.022%
0.0022",.{,
0.00022",.{,
0.1!2
Ideal preferred
value
10 X 10
10
X 10 116
= 10 ohms
Rounded off
value
Actual value
used
10
10
= 14.68
15
15
22
22
32
33
10 X 10416 = 46.42
46
47
10 X 10516 = 68.13
68
68
Convec- Conduction
Rating Hot Spot
at 70C Temp. Radiation
tion
via leads
Oxide Film
859111-N-002
Style 1 X-56-E
'hW
115C
7%
18%
75%
Vitreous Wirewound
859114-F-001
Style 2E -56-12
12 W
320C
57%
37%
5%
Resistor
. 859111-N-002
Style 1X-56-E
859111-N-002
Style 1 X-56-C
Point on
Derating
Curve
Hot
Spot
Temp.
Stability
Requirement
Remarks
0.176W@70C
81C
1%
Easy to meet
requirement
0.044W@125C
129C
1%
More difficult
to meet and
marginal
failures occur
2W@70C
170C
3%
Difficult to meet
161C
3%
Easier to meet
than above
o.35W@125c
Temperature coefficient
The change in resistance with temperature is not
normally a linear function and this has led in recent
years to the adoption of other terms to describe the
effect without implying that a linear function applies.
In BS9100 the term temperature characteristic is used
and tests are made at certain specified temperatures
covering the whole range of application, and the
change in resistance value at that temperature is
measured and compared with the reference temperature of 20oc. Another useful concept is to derive
from such measurements a "chord" slope given by
Change in Resistance
Over the normal range of
Change in Temperature
.
temperatures experienced in practice it is only necessary to measure the resistance value at four temperatures to define the characteristic, because none of the
resistor materials currently used have such unusual
characteristics that a cubic curve could not be fitted
to the measurements to give adequate accuracy for all
applications. As an example consider the characteristic of a precision flat bulk metal film resistor which
is measured at four temperatures as shown in Table 6.
The whole characteristic can be fitted to a cubic
curve giving the change in resistance value relative to
20oc hence:
6R (ppm)
-25.41 +1.9591 T -0.035065T 2
+0.00003172T 3 (where T is ambient temperature in
oc.)
Using this expression the resistance value can be
computed for any temperature in the range -55C to
- 2.5ppm/C - 3.3ppm/C
Resistance Change
INSULATION
TESTER
500
OR
aov
MECHANICAL CONSIDERATIONS
Solderability of leads
The globule method of assessing solderability is a
quantitative test which correlates very well with
users' requirements. The use of printed circuits and
automatic flow-soldering techniques requires components which solder in about 2 seconds or less. The
solder bath method of assessing solderability is also
used but it is more difficult to assess whether the
soldering is satisfactory or not. However, recent
machines which can give varying immersion times,
can be very successful. Wherever possible the globule
method should be used for assessment and most
wire-ended components can be tested on the 'Multicore' machine. This machine is provided with a
thermostatically controlled anvil, on which is placed a
solder pellet of specified size. The component lead is
lowered into the globule of molten solder, thus
dividing the globule into two parts. When wetting
occurs the two halves join over the lead arid the time
for this to occur is noted. In the as-new condition the
component will usually solder in less than 2 seconds.
The effect of storage and ageing may reduce the
solderability and methods of accelerating this process
are being developed using either dry or steam heating.
For tags or other terminations which cannot be tested
on the Globule machine, a solder bath may be used.
The immersion time is varied to determine when
wetting and de-wetting occurs.
''60
W oxide film
DURATION Of TEST
~
30
20
24
48
168
500 1000
~
05
+ ~
aT
01_
=#:
.
dE
-2:.2....
5
I
I
3
22 106
4
136 605 1097
8
104 328 1069 1508 1547
340 580 825 570 278
2768 223( 1108 513 185
42
66 24 47 42
4
19
12
21
63
I
4
I
2
I
I
I
2
I
200050001/I~&>VJ7!
I
I
83
6
76
5
32
460 226
18
104 280 507 748 752
325 707 888 969 822 1090
12n 1391 '155 1!62 825 873
1278 867 678 531 328 312
206 141 232 23
16
9
174 66
14
II
5
5
I
2
I
Bx + T + C
I
I
= 0.0093 X 10 -6
STABILITY
CLIMATIC TESTS
HOURS
0 92
= 3282 X 30,000
where
and
----6R%
25
03
AMBIENT TEMPERATURE
35
025
30
20
25
02
15
20
015
15
0
005
05
tI
2000hr STABILITY
I YEAR STABILITY
ANNUAL STABILITY (6R% PER ANNUM) AFTER lsi YEAR
NOISE MEASUREMENTS
lmV
CARBON
COMPOS! TION
V 2 = 4kTBR
where k, Boltzmann's canst.= 1.38 X 10- 23 joules
T, absolute temperature, degrees Kelvin
B, Bandwidth
R, Resistance
l~~E
MAXIMUM
NOISE
LEVELS
IOO~V
,.
lI
RESISTOR
OXIDE
FILM
METAL
FILM
>
,-'JoHNSON
,NOISE
(})
6
z
(})
IJ.J.V-
0 ~ Yoo~.ll-___:,1..k---,O.l.k---,o-'=oc-k--::L,M:----:,-:!:o:-:-M-,:-;;oo~M
RESISTANCE VALUE
NON-LINEARITY
A perfect resistor has a linear voltage current characteristic. Resistors made from bulk metals or alloys do
behave in this way, but all film resistors and also
other types, such as carbon composition and silicon
carbide, have a non-linear volt&ge - current characteristic. The amount of non-linearity may be very
small and it is quite difficult to measure the effect on
metal film or oxide film resistors.
g,.,:~f~---~--=x-x--~k~
-~
x
x-x--x
-x-x--X--- xX
<r <r
9oa
x-
IOk .
lOOk
X--X~
06
04
" 'X
02
RF MEASUREMENTS
Every resistor has stray inductance and capacitance
and when measurements were difficult to make a
great deal of effort was made using sophisticated
bridge techniques to assess the steady state performance of resistors. In many applications a knowledge
of the value of the lumped inductance or capacitance
will provide enough information to the user to enable
him to make a sensible choice of component. In
recent years the existence of good pulse generators
and fast oscilloscopes has made it easier to evaluate
the RF performance by direct experiment. For
instance in terminating a 50 ohm line by a single film
resistor the helixing of the track creates an inductance which predominates over the capacitance and
can be observed easily on the oscilloscope. The use of
two 100 ohm resistors in parallel will invariably
improve the performance and the optimum number
of resistors is easily found.
The ultimate geometry for this kind of application is
probably the disc and unhelixed rod forms which are
easily incorporated into coaxial systems and have
been successfully used at frequencies up to lOGHz.
At these frequencies the geometry is critical and the
materials used must be selected with great care to
minimise frequency dependent loss properties.
Capacitance of resistors is largely a function of size,
and small resistors are obviously better in this respect
than large ones. The actual capacitance may be of the
order 0.1 pF for 1/.t-watt sizes increasing to 0.5 pF for
a 2-watt oxide film type. Inductance is also
8
I
0 ~~----~~~~~~~~o~--~--~L-~~~~oo
FREQUENCY MHz
CONCLUSIONS
The advent of BS9000 is expected to have farreaching consequences and the extension in Europe
of similar systems of specification and Qualification
Approval should ensure a degree of uniformity in
components that has not been achieved previously.
As far as fixed resistors are concerned the generic
document BS9110 gives full details of the tests which
are appropriate to these components. Other useful
documents that are referred to in BS9110 are:
BS9000 - General Requirements for Electronic Parts
of Assessed Quality BS9001- Sampling Procedures.
BS2488 - Preferred Numbers. BS2011 - Test
Methods. BS1852- Marking Codes.
Chapter 2
Capacitors
R J Everitt BSc
Dubilier Limited
During the production and testing of capacitors, the
manufacturer carries out a large number of specialised
tests on the capacitor parameters. In most cases the
equipment used has been purpose built and in
consequence is not available to the average user. In
consequence many engineers find themselves in the
position of using components which they are not able
to adequately test or specify. The test methods
described in this chapter are generally suitable for use
in the average electronic engineering workshop.
Before describing test methods it will be useful to
examine briefly the implication of some of the
parameters to be examined.
In Fig.l a very much simplified equivalent circuit is
shown for a capacitor. The resistance in parallel with
the capacitor represents the leakage resistance and
will normally be measured in megohms. This resistance has the characteristic of increasing with time
and is also very temperature dependant. The series
resistance Rs represents the losses in the capacitor
and is normally referred to as the equivalent series
resistance or ESR. This can be divided into two parts:
cos
~---.0----~ct---.---~~-----o
</>~tan
where w
5 = Rs Cw
(1)
27Tf
f = applied frequency in Hz
Fig.l. Equivalent circuit for capacitor.
we
tan
8'_!f
;;;c
'RCw
-c:
~ 0
1-
10
~<t
15
<t
u
0 01 pF TUBULAR
J:
;!. 20
10
a:
~
01
a:
w
a.. 001
100
IKz
FREQUENCY 1Hz)
ol
IOKz
~.
--~F
-.........:::. ~-------iD,E
- - x........_
......._
' x~
en
':i
~
x---x-x~-
~. ........__._./~
-40
-20
40
20
0
TEMPERATURE tC)
80
IMPEDANCE MEASUREMENTS
Although accurate measurement of capacitance is
useful in tuned circuits and a knowledge of power
factor essential for power loss calculations, the
majority of applications for capacitors in electronic
circuitry require a value for the impedance rather
than the separhte capacitance and power factor
vectors. The measurement of impedance is normally
well within the scope of the average engineer and can
be accomplished with equipment available in most
workshops or laboratories.
The circuit used consists of a resistance R, connected
in series with the capacitor and both housed in a
screened test box. The connections should be made
through separate coaxial sockets on the box. The
supply from an oscillator is made with a screened
lead. The resistor must be non-inductive over the
frequency range required and have an accurately
known value of 50 to 100!2. The oscillator should
have an output of at least 1 volt with pure sine wave
output. The presence of harmonics in the oscillator
output will result in erroneous impedanc~ values and
make the measurements non-reproducible.
The detector used can be either a calibrated oscilloscope or an electronic a.c. voltmeter suitable for the
frequency range to be covered.
The test is carried out as follows:
(1) Connect the capacitor in the circuit (preferably
with soldered connections).
(2) Set the oscillator to the required frequency and
adjust the outp'ut to approximately 1 V.
rih
32k
z=Y2R
Vl
CAPACITANCE RANGE
500pF to 0 lfLF
001 to IOfLF
60
24
Rl
795k
795k
R2
630k
63k
M
50fLA
lmA
12
LEAD
LENGTH
2in
4
w
~ 3
<!
~ 2
xiOOOO
-------------~SEPARATORS
components of the equivalent circuit on the impedance characteristics. At the low frequency end the
impedance is controlled by the capacitance almost
entirely. As the frequency increases the resistive and
inductive components become increasingly important. If the resistive element is dominant the
impedance curve will lie above the theoretical value
for capacitance only, while if the inductive element
dominates the curve will dip below the theoretical. At
the minimum in the curve, which is the self resonant
point, the impedance is equal to the ESR and above
this the unit progressively takes on the character of
an inductor. The large deviation of the electrolytic
types results from the effect of the rather complex
resistance and capacitance combinations in the unit.
The capacitor is made up of an oxide film on the
surface of the positive electrode with a layer of
conducting electrolyte filling the space between the
oxide surface and the negative electrode. This is
shown diagrammatically in Fig.10. The true capacitor
is made up of a layer of negative ions on the
electrolytic-oxide interface and a positively charged
metal on the opposite face of the oxide. The
equivalent circuit for this can be expressed as shown
in Fig.ll, where the resistances represent the varying
electrolyte path lengths to small capacitance elements
on the oxide. There is therefore a series of elements
with varying time constants, thus when the frequency
is increased a smaller charge movement occurs in the
elements with long time constants with a consequent
apparent loss of capacitance. When the electrolyte
resistivity is increased, either at low temperatures or
by exhaustion of the electrolyte in service, the
impedance values will increase, particularly in the
region of 100 kHz. At frequencies above 10 MHz
these changes are generally much less pronounced.
Coupling impedance measurement
An important characteristic of capacitors having more
than one capacitance element in a single container is
the coupling impedance. This is a measure of the
transfer of signal from one element to another. The
test procedure consists of feeding current into one
element and the measurement of the induced voltage
in the other elements.
~-r-----
__oc. .t_o_c!~oc
.
. .J_o_cf_oc
.
. .. . ___ _
13
A*
VARIAC
RESERVOIR
SECTION
voltage out
.
.
.
The ratio current in has the dtmenswns of Impedance and is referred to as the coupling impedance.
Unlike a normal impedance, however, the required
condition of minimum signal transfer is obtained
when the coupling impedance is low.
Two measurement methods are in general use with
the choice between them being made on the basis of
the type of application and convenience of testing.
A typical circuit used in the first of these methods is
shown in Fig.12. The resistance R is used to simulate
the normal load condition of the capacitor reservoir
section. For a capacitor element of 200 iJ.F working
at 300 V, a value of around soon is reasonable. This
gives a ripple with the correct wave shape to be
passed through the reservoir section. The variac is
adjusted until the ammeter reads 500 rnA. The
induced signal on the remaining elements is then
examined on the oscilloscope and the value of the
peak to peak voltage is noted. A typical limit for this
is 40 to 50 mV P/P.
While this test is relatively simple it often requires a
screened transformer to isolate noise from the mains
supply and it suffers from the disadvantage of rather
high voltages on the capacitor. This means that either
the test can be carried out only by qualified engineers
or that elaborate safety precautions have to be taken.
The second method makes use of an oscillator to
provide the driving signal which can be set to any
desired frequency. The circuit used is shown in
Fig.13. For 50 Hz measurements a low voltage
transformer can be used to replace the oscillator.
R
OSCILLATOR
SUPPLY
Cl
0
14
(
I
----------I
7k
2k
I
I
700
200
70.0.
20.0.
I
I
1-,!,-
7Sl
1::::
I
OFF o
I
1 SWITCH CLOSED I
5k.ll
1 BY SHUNT ARM)
.I
II I.Q
2.0.
1.0.
500
~------------------------1
Fig.14. Circuit for insulation resistance tester. Galvano2500 mm/p.A. Shunt S. 10 OOOf2
meter sensitivity
total resistance tapped at 1, 3, 10, 30, 100, 300, 1000,
>
3ooo, 1o ooon.
Measuring Voltage
< 10volts
10 Volts 1
>soo volts
100 15
500 50
15
Capacitor type
Tan
Insulation
Resistance
>1o,.;;; 25 pF o.oo3
> 25 ,.;;; 1000 pF 0.002
>1000pF
0.001
Mica
Small
paper
Grade
Grade
Small
paper
25,000 Ms-2
or 1,000 ri.F
whichever is less
<0.01
>5,000 r!F or
25,000 Ms-2
whichever is less
<0.01
>2,000nF or
10,000 Ms-2
whichever is less
<0.01
>2o,ooonF
H6
tubulars
tubulars
H5
Polyester film
Metallised
Polyester
<0.008 at 1 kHz
<0.015 at 10kHz
>30,000 Ms-2
below 0.33 #-( F
> 10,000 r2 F for
0.33 #-{F and above
Metallised
Paper
<0.02 at 1 kHz
>soonF
Tantalum
Foil (Plain)
v ,.;;; 16 v < 0. 15 )
(Etched)
v >16 v<o.1 l
v <so v <o~ 2 )
Leakage current
at 120Hz
v ~5.0 v <0.15)
Tantalum
Solid
Aluminium
Electrolytics
Polystyrene
<
0.5
0.35)
0.25)
0.2
at 100 to 120Hz
0.0005
250,000 r!F or
750,000 Ms-2
whichever is less
The position with regard to non-electrolytic constructions is much more standardised and in almost all
cases a test period of 1 minute is specified for
insulation resistance measurements. The voltage used
for insulation resistance should be as in Table 1.
SUMMARY
BIBLIOGRAPHY
Fundamentals of Electrical Measurements. C. T.
Baldwin. Harrap.
Higher Electrical Engineering. J. Shepherd, A. H.
Morton, L. F. Spence. Pitman.
Alternating current Bridge methods. B. Hague.
Pitman.
Power Capacitors. R. E. Marbury. McGraw-Hill.
Insulating Materials for design and engineering
practice. F. M. Clarke. John Wiley & Sons.
17
Chapter3
Wound
Components
cores can be made to possess quite high permeabilities, and are suitable for use up to 15 MHz.
Applications include LF and RF coils and rod
antennae.
ACCEPTANCE TESTING
Acceptance testing on components received from a
manufacturer should ideally establish identification,
electrical functioning and mechanical fit. In many
instances the extent of acceptance testing and checks
will be defined either in a procurement specification
or 'in-house' document. In the absence of such
information and to assist in the compilation of a
suitable document the following factors should be
considered.
Test sample size: Depending on batch quantity, tests
could be 100% or on a sample quantity based on
batch size and Acceptable Quality Level. Reference
should be made to BS9001 and from the tables a
sample size determined.
Acceptance testing should include all or some of the
following checks.
LF Components
Identification
Robustness of terminations
d.c. Resistance at +25oc
Solderability
Resistance to soldering heat
Inductance
(with or without d.c.)
Insulation resistance
Voltage proof (on parts intended to be electrically
insulated).
On RF Components the following additional tests
should be included:
Q- Min value
Self-capacitance
Tuning range (if fitted with adjuster)
Self-resonant frequency
To enable the above tests to be made the following
instrumP.nts would be required: calibrated weights or
test fixture for termination test; ohmmeter - preferably digital reading; inductance bridge - commercial
grade of 0.25% accuracy; insulation resistance test set
to read to 106 M.Q; voltage proof test set to 5kV
a.c./d.c.; 'Q' meter; grid dip oscillator
MEASUREMENTS ON LF INDUCTORS
Inductance, a.c. resistance and impedance
A widely used precision bridge for series measurement of inductance, a.c. resistance and impedance at
low frequencies is shown in Fig. I. It is basically a
Maxwell comparison method. This bridge consists of
Fig.l. Basic schematic of a bridge for the measurement
of inductance, a.c. resistance and impedance at low
frequency.
~---~
Rz
AMP/
DETECTOR
D.C.
A.C.
INPUT
D.C.
A.C.
AC
INPUT
1.
SUPERIMPOSED
D.C.
SUPPLY
II
22
' '
RANGE
RESISTORS
i
NULL
o-----~(o)
0
L-----~l
....
0ARSE
NDUCTANCE
FINE
INDUCTANCE
~--------~
(b)
RH-RC
RC (234.5 +TC)
WINDING UNDER
TEST
DC BRIDGE
~------~
o-----~
~--~~c23________~
Cz
L---~------------;~c~'--------~
Fig.4. Measuring the temperature rise of a winding
energised at working voltage.
~---------------------------------------~
MEASUREMENT OF H F INDUCTORS
As the frequency of measurement on air cored, dust
cored or pot core components increases, direct
measurement by bridge type instruments becomes
impractical. Whilst it is acknowledged that certain
measuring instruments with direct read out of the
inductance are available up to about 10 MHz, an
instrument most suited to this measurement is the 'Q'
meter. The 'Q' meter is widely used where measurements involving coil wound items are to be found and
is a useful instrument for routine testing.
INDUCTOR TOLERANCES
The tolerances to be expected are to a large extent
governed by the materials used and me.thod of
construction. Considering an air cored coil wound on
a bobbin or former, a realistic production spread
would yield tolerances of about 2% on inductance
and up to 10% on d.c. resistance. Self capacitance is
dependent largely on the method of winding and
could give values of between 5 and 20 pF.
Precision
air
cored
25
Chapter 4
General
Purpose
Semiconductors
M R McCann CEng MIERE
ITT Semiconductors
There are many different types of semiconductors
available on the market and used in a multitude of
applications. It is difficult to state categorically what
constitutes a general purpose device but these devices
have been selected as being most commonly used:
bipolar transistors, small signal diodes and zener
diodes. Since technology is advancing so rapidly it is
now necessary to include integrated circuits of both
digital and linear types.
With semiconductors three testing procedures are
common:
(1) Testing the devices as individual components
before insertion in the circuit.
(2) Testing a circuit board as a functional element
containing one or more semiconductors.
(3) Testing the complete system.
All or some of these tests might be made depending
upon the economic/quality compromise. Testing of
devices in a board is rare and is often rather complex
compared with items (1) or (2) due to the interaction
of other components on the measurements.
If item (1) is required then there are various factors
to consider before testing a device to a given
specification: (a) What parameters should be tested?
(b) The size of the sample on which to test any given
parameter (c) What test limits to apply compared
with the specification limits (d) How many samples
should fail the test specification limit before a device
is rejected. (e) The system of measurement.
What parameters should be tested? It is a lengthy and
costly business to test all the parameters of any
semiconductor except for the application where
extreme reliability is needed as for devices in submerged repeaters. At the other end of the scale a
device might need no testing since it is used well
26
Parameter variations with temperature are predetermined by measurements on extreme values of product. In the following tests the temperature depen
dence is indicated for each parameter where
applicable.
Test systems. Equipments are available for tf'sting and
measuring semiconductors. These range from the
simplest which test semiconductor junctions to automatic digital printout systems for propagation delays;
also systems range from measuring a small sample to
computer controlled systems handling at least 5
million devices a year with a separate tape programme
for each specification. Whichever system is employed
it is essential to have a knowledge of the parameters
of a device for the reasons already stated if one is to
make the most efficient use of the test equipment
employed. This also applies when testing devices on a
board or a complete system in that tests can be
tailored to reveal all the major faults liable to occur
due to 'out of specification' products.
Assuming testing is to be made on individual semiconductor devices then testing methods are described in
detail in the following publications: British Standard
Publication 1 , The USA Department of Defence
Standards 2 , Electronic Industries Association and
National
Electrical
Manufacturers
Association
Standards 3 , and the Commission Electrotechnique
Intemationale
International Electrotechnical
Commission Recommendations 4 . Within these publications there is considerable detail on circuits for
measuring most parameters, certainly those for
general purpose semiconductors.
SYMBOL
GIVEN
CONDITIONS
APPROXIMATE
TEMPERATURE
COEFFICIENT
hFE
'c VeE
+ 1%/C
VcE(SAT)
lc. Is (usually
ratio of 10: 1)
Very small
VsE(SAT)
lc. Is (usually
ratio of 10: 1)
VsE
'c VeE
Vcso
lc(IE=O)
Very small
VcEo
lcOs=Ol
Very small
VESO
IE Oc = 0)
Very small
1cso
Vcs (IE= Ol
Varies
I cEO
VeE (Is= 0)
Varies
1ESO
VEs Oc = 0)
Varies
PARAMETER
Common emitter forward current
gain~~]
27
}150
... 100
'
a so
lr
,u
l(){ftA
lmA
IOOmA
IOmA
COLLECTOR CURRENT lc
lA
TESTING TRANSISTORS
General purpose transistors include both germanium
and silicon each available as p-n-p and the majority of
n-p-n are in silicon. The circuits for testing all types
remain essentially the same except to reverse the
polarity of the d.c. supplies depending upon whether
it is p-n-p or n-p-n.
Static characteristics
Most of these include current gain, leakage current
and saturation voltages under various conditions. The
ratings of transistors rarely need be checked since
they are inherent in the design of the device. The
exception to this is the breakdown voltage which can
be very low in a faulty transistor but these need not
be measured if tests are made of leakage current,
which should be specified on the data sheet.
For completeness the common parameters are shown
in Table 1. Most equipment suitable for the measurement of silicon transistors can be used for measuring
germanium transistors.
Current gain (hFE)
The d.c. common emitter current gain is one of the
most important parameters of a transistor since it
determines the static operating point in an amplifier
and the base current drive in a switching application
(if it is quoted at a low Vce> The hFE is dependent
upon collector current and voltage (see Figs.1 and 2)
and ambient temperature.
The gain increases with junction temperature at a rate
of about 1 %/OC. If the transistor is not mounted on a
heat sink then the increase in hFE is dependent in the
same manner with ambient temperature. There are
many equipments available and many circuits recommended for measuring hFE An approximate method
(if it is necessary to construct a test circuit) which is
simple to use is shown in Fig.3.
This circuit directly measures the base current IB
required under given conditions of IE and V CE
28
02
2V
20V
VeE
CURRENT
LIMITING
RESISTANCE
Fig.4. Measurement
of leakage current.
CURRENT DEFINING
RESISTANCE
29
c'
r bb' b"
'cc
~~
v ,
b'c
9,\/sc
'co
Cbe
'b'e
e
r cr'
e
'bc=
9m-
h~ '
r. =..!!.!r...!s_
r
ce
cb'e
JL
be
= ....!L
JL
I
7e
=c junct;on+
(bose)
ColfFUSION
(collector)
..
b'e
=--~
w-rh~
'
parameters in the hybrid 7T depend upon the operating conditions. Since the hybrid 1r parameters can be
directly related to the 4 terminal parameters the same
can be said of them, their spreads and typical values.
SENSITIVE
VliLVE
VOLTMETER
It can be seen that given hFE Ic, VCE #J. and the
extrinsic resistances there are very few unknowns.
GENERATOR
I VOLT
AT IOOMHz
oe
re
=-
#l
Where hfe
>>1 andre=
All the 'h' parameters can be approximately determined if hfe J.~ and rbb are known at the given Ic
and VCE Similar approximations can be made on the
other 'h' parameters. With 'Y' parameters at high
frequencies it is impractical to make approximations
of this kind.
The measurement of the small signal parameters is
best done using the many equipments available on the
market specially designed for this purpose. Alternatively the circuits provided by BSI are adequate but
eircuits have to be constructed and addition:al test
equipment is needed.
TANGENT AT A
c
hFE AT A
AC
= Be
= TAN 8
AC
hte AT A = DC
= TAN
A.
't'
Switching characteristics
The switching times are very often given on transistors. The measurement of these parameters is slow
and needs specially constructed test jigs made to the
switching test circuit shown in the manufacturers
31
tf
tr
ts
Medium/fast switching
10
30
Amplifier transistor
(non-gold doped)
10
30
500
32
~UBR.f:NJ
CURRENT
/;J-_
_ _ _
LEAKAGE
CURRENT
_!
i~ENT
SLOPE IS
R,
u,
REVERSE VOLTAGE
ZENER DIODES
The high resolution digital voltmeter is one of the
best instruments for the accurate measurements of
zener diodes. The zener voltage is specified at some
reverse voltage often to an accuracy of 5%. Also
zener diodes can have a very low slope resistance ( 5
ohms on an 8.2 volt planar zener diode at a current of
2 rnA) and a resolution of at least 0.1 mV is essential
for an accurate measurement. Fig.10 shows the
specified parameters. The other parameters present
no special problems.
INTEGRATEO CIRCUITS
Digital integrated circuits are specified for having
defined limits of logic level, propagation delays and
operating to a given truth table. A check of the
operation of a simple logic gate to its truth table is
easy to perform. However, with a complex logic
function it can be very difficult to generate all the
sequence of events necessary to cover every logic
transition especially when signals have to be applied
in a particular order within certain times, ( i~ a 4 bit
clock left clock right presettable shift register). A
preferred method of measuring integrated circuits is
when they are wired together to perform a logic
function or the circuit used on one circuit board, or a
complete system.
Often logic checks and propagation delays can be
performed on circuit boards which might contain at
least ten packages, and then a similar check made on
REFERENCES
(1)
(2)
(3)
(4)
33
Chapter 5
Power
Semiconductors
R D Eveleigh
AEI Semiconductors Limited
Semiconductor manufacturers of power diodes and
thyristors, like other component manufacturers put
their products through rigorous tests before the
customer receives them. It is essential that these tests
are understood before discussing any tests that
semiconductor users may apply. The nomenclature
and symbols used in the test are those given in
BS9300:1969
v.RSM ----
ACCEPTANCE TESTING
The manufacturer will have carried out comprehensive tests on the product before shipping and therefore the acceptance test should only be necessary to
check any damage occuring in transit. When making
any test it is essential that all precautions are taken
not to inadvertantly damage the devices or exceed the
stated rating. There are two main acceptance checks
for thyristors and one for diodes.
Voltage testing
As previously discussed there are up to four voltage
ratings given for diodes and eight for thyristors. If
each rating is taken in turn, it will become evident for
which it is possible to test.
VasM and VnSM non-repetitive peak voltages,
require only a single pulse of voltage for testing and if
the voltage peak is exceeded the test devices will be
damaged. To test for this limit is not very satisfactory.
VaRM and VnRM repetitive peak voltages, can be
tested but the voltage could be close to the nonrepetitive value, therefore, if transients occur while
measuring the device could be damaged. The current
limits laM and InM are frequently quoted at this
voltage.
VawM and VnWM crest (peak) working voltages are
the maximum recommended peak of the voltage
under normal working conditions. These are the
safest values to test to as there is usually sufficient
safety margin between these and the non-repetitive
values.
FORWARD
CURRENT
FORWARD
CONDUCTION
ON-STATE
CONDUCTION
HOLDING
CURRENT~
REVERSE
VOLTAGE
REVERSE
VOLTAGE
/ REVERSE
AVALANCHE
BREAKDOWN
FORWARD
VOLTAGE
REVERSE
CURRENT
tal
REVERSE
/' AVALANCHE
BREAKDOWN
~ ~-
FORWARD
BREAKOVER
_v2i-,!AEJE
FORWARD
VOLTAGE
OFF-STATE
BLOCKING
VOLTAGE
REVERSE
CURRENT
ibl
37
DEVICE
VARIABLE
UNDER
a. c.
SUPPLY 1----='c:.._--'---...J TEST
(a)
VARIABLE
de.
SUPPLY
DEVICE
UNDER
!---""'---......_-_. TEST
(b)
SAFETY
INTERLOCK
Test procedure
Before commencing any test it is essential to ensure
that the conditions are as stated in the manufacturers
data sheet. The temperature should be noted and, if
required at elevated temperatures, the device should
be raised to the temperature by external means and
allowed to stabilise within the normal temperature
tolerance of 2oc. To commence the test ensure the
voltage supply is at zero, insert the device into the
test position and turn the switch to 'reverse'. The
voltage is increased and the reading observed on
either the meter or, the oscilloscope if used. When the
specified voltage is reached or when the leakage
current has reached the maximum limit quoted, the
voltage reading should be noted. The voltage should
not he held for more than five seconds. The voltage is
then reduced to zero. For a thyristor, the switch is
changed to 'forward' and the test repeated. The test
device is then removed from the test position. If the
meter measuring leakage current is a mean reading
instrument then the reading obtained should be
multiplied by 1r to convert it to the peak value or the
reading may be taken directly from the oscilloscope.
A similar procedure is adopted when using the d.c.
method but the starting temperature should be noted
from the data sheet as this may be lower than the
.maximum junction temperature.
Gate characteristic testing
The gate current and voltage to fire a thyristor can be
measured very simply. The circuit in Fig.5 consists of a
low voltage half wave d.c. supply capable of giving
about 3 to 5 amps and another variable low current
d.c. supply, to provide the gate requirements.
The test procedure is to increase the variable d.c.
supply until the ammeter I 1 just reads or lamp L
lights. The switch SW 1 is opened and readings of gate
voltage and current are read fron1 the meters V and A
respectively. Points to note are the inclusion of a gate
to cathode resistor if specified in the data sheets and
the temperature at which the test is performed, for
acceptance tests this is normally room temperature.
3XSJ
1403F
47k.U
---a'o250Vno~JI
38
OSCILLOSCOPE
HORIZONTAL-VOLTS
.___ _ _ _ _ _ _ _ _ OSCILLOSCOPE
VERTICAL-CURRENT
SLI03~
iv
v
CONSTANT
CURRENT
CALIBRATION
SUPPLY
~II
39
~BEFORE
BALANCE
--'--.,....---"--.--AT BALANCE
4 c~(. T
CALIBRATION
2~5~ty;
HOLE
STORAGE
SPIKE
(a)
AT
BALANCE
ON LOAD
(b)
Fig.7a) this reading, V 1 , is noted from the potentiometer dial, as is the temperature T 1 , from the
electronic thermometer. Current is then passed
through the device by increasing the variable transformer until a current appropriate to the device is
passed. The temperature is allowed to settle and the
potentiometer reset to balance at a point after the
hole storage spike, (approximately 2 to 5 ms after
current zero, see Fig. 7b ). Readings of the potentiometer dial V 2 , temperature TB and power W, are
noted. If a wattmeter is not available readings of
mean current, as indicated on the meter across the
shunt, and peak volts across the device are recorded.
The latter can be displayed either on a separate
oscilloscope or the second trace of a double beam
oscilloscope. Thus mean power dissipated is equivalent to mean current times peak voltage, to an
accuracy of approximately 5%. The property of
semiconductors of having a negative voltage temperature coefficient at low forward currents is useful in
determining the junction temperature without having
to put thermocouples or other temperature measuring
devices inside the capsule. The temperature coefficient is approximately -1.8 m V jC. This can be
checked by passing the calibrating current through
the device and measuring the voltage drop at various
temperatures. To evaluate the thermal resistance;
(V 1 -V 2 ) X Temp.Coeff.+T 1 =TJ,
then
TJ-TB
W
= Rth
When testing certain thyristors the waveforms displayed on the oscilloscope may jump. This is caused by
different areas of the silicon slice turning on due to
temperature effects and reducing the voltage drop.
This may upset the readings but may be overcome by
passing a constant current 0. 5 to 5A, anode to
cathode, through the device in addition to the gate to
cathode calibrating current.
Points to note for this test are that at the commencement, the device and heatsink are at ambient
temperature, since the test is based on this assumption and that the device is damped down to the
recommended torque. The readings of thermal resistance should be within the manufacturers specification, if not the mount down should be suspected.
40
IO.UIOW
250V
a. c.
...- - - . S J
203F
Jkc;o:F
~
TEST
SWITCH
lk.U
A ANO B MERCURY
WETTED RELAYS
250V
oc.
42
~TRACE
-----rf\1 CURRENT
TRACE
AT BALANCE
\Y--SPOT
(a)
AT BALANCE
(b)
BIBLIOGRAPHY
For further details of test methods and procedures
covering diode and thyristor testing refer to
BS9300:1969 Specification for semiconductor
devices of assessed quality: Generic data and method
of test.
For further reading SCR Manual, General Electric
Company, Editor F. W. Gutzwiller
Chapter 6
Microwave
Semiconductors
A Hale BSc
Mullard Limited
VARACTOR DIODES
Varactor diodes are a broad class of semiconductor.
diodes useful for electronic tuning, harmonic generation and parametric amplification. Diodes have been
developed specifically for each of these applications.
Silicon is the material most commonly used and
offers the advantages of an advanced technology.
These include low thermal impedance, high breakdown voltage, high operating temperatures, reliability
and low cost. The other material in use is gallium
arsenide which offers high efficiency and operation at.
higher frequencies than silicon. Silicon varactors are
normally used at J-band (12-18 GHz) and below, and
gallium arsenide devices are more appropriate for
operation at Q-band (26-40 GHz).
For silicon, a diffused planar epitaxial construction is
used as shown in Fig.1 whilst the gallium arsenide
varactors employ a diffused mesa construction as in
Fig.2.
43
E.,, :g
. . - - - - - - - METALLISED CONTACT
:
OXIDE
n- TYPE EPITAXIAL LAYER
~~~i~r~:igs~~~~2T
44
I
Cs
~=
-'
gls
tfci
~Rs
5
4
a.
u-
10
-70
FORWARD~
VOLTAGE
45
16
1-4
12
04
02
10
FORWARD-+- REVERSE
VOLTAGE
VOLTAGE
DEVICE HOLDER
Using the sweep oscillator the series resonant frequency of the diode in its holder can be found.
Removal of the diode and the insertion of a suitable
amount of attenuation allows the transmission loss of
the diode at resonance to be measured. At series
resonance the diode shunts the transmission line with
its series resistance only.
46
Ti!~+1]2
where Z0 is the characteristic impedance of the
reduced height waveguide.
Repeating the experiment at different bias levels
results in a series resistance/voltage relationship as
shown in Fig.S.
At high frequencies the equivalent circuit is reduced
to that shown in Fig.9. The series resonant frequency
is fres
and fres
= 2n .;c~
(1)
where CT = Cj + Cs
Re-arranging equation (1)
CT =
1
2
2
47T Ls fres
(2)
Q
Another 'figure of merit' often quoted by manufacturers and used extensively to grade the quality of
varactors used in parametric amplifiers is the term
"'Y fc,
6
5f-
4f-
=
2
10
FORWARD
VOLTAGE
-+-
-20
-10
-30
-40
-50
-60
eT(V.)- eT(V 2 )
2[eT(V I)+ eT(V 2 )]
(6)
-70
REVERSE
VOLTAGE
Step-recovery diodes
Another group of varactors specifically designed for
harmonic generation are called 'snap-off' or 'steprecovery' diodes. For this group the capacitance
variation is confined almost entirely to the forward
bias region, as shown in Fig.lO.
211' ej
(3)
Rs
(4)
(5)
ZfRQ
~R.f!E~T- _l
__ _
TIME
with graded junction varactors. This may be attributed to the lower effective series resistance coupled
with the maintenance of a high breakdown voltage.
Transition time (tt). The test circuit for the measurement of the transition time is shown in Fig.13. The
input pulse should have a rise time of less than 1
nanosecond and an amplitude of at least 10 volts. The
forward current is adjusted to about 1 rnA. The
output pulse is attenuated and displayed on the
sampling oscilloscope. (See Fig.14.) The transition
time is defined as the time between 20% and 80%
points of the observed waveform.
GUNN DEVICES
In 1963 Gunn reported that when a d.c. voltage was
applied to n-type gallium arsenide the current increased linearly and then oscillated when the average
electric field increased beyond a few thousand volts
per centimetre. The device is fabricated using
epitaxially grown gallium arsenide in the form of an
n+-n-n+ structure. Metallised contacts are made to the
n+layers allowing the chips to be mounted in standard
microwave encapsulations. The basic structure is
shown in Fig.15.
SAMPLING
OSCILLO-
TRIGGER
SCOPE
SYNC
son
200
.....
z
!
-------20%
o,L-------~5~--------~~o--------_jls
BIAS VOLTS
48
IPEVICE ON
HEATSINK
.I
Olll
X-Y
X RECORDER
OR
y OSCILLOSCOPE
POWER
SUPPLY
OR RAMP
GENERATOR
RF. CHOKE
OUTPUT
COAXIAL CAVITY
WAVEGUIDE CAVITY
49
Fig.25 and 26. Typical variations of power and frequency caused by mismatch. The shaded regions represent the possible effects caused by variation of the phase
of the mismatch.
=
=
=
=
=
=
where Is
e
T
k
n
v
= Is [ e
To
B
G
nkt
ev -1
(1)
dB.
NR =
where N 1
where Fo =
F1F
Fo
(FJF -1) + NR
=
=
Hence N 2 = 1 + NJN
N1
FK T 0 B
and F
NJN
KT B[~~-1]
(2)
(3)
_... 2
-5
-4
-3
VOLTS
-2
-I
<t
:ot..
01
-"' 4
6
52
02
03
VOLTS
04
05
f=~o -~
(4)
F 0 = (1 + r)[T
l!o
_J
J
[_L]
-!]
aif
15.0.
53
FILTER
1---iBANDWIDTH
0-1 MHZ
= FJF (Y - 1) +1
PULSED SIGNAL
/111~11//~l !MJII/l~l/t~~(
AT RECEIVER
OUTPUT
~~lllf~W~
Fig.33. Relationship between signal frequency, tangential sensitivity and amplifier bandwidth.
-70
100kHZ
I MHZ
E
to
"'
10
SIGNAL FREQUENCY GHz
100
ACKNOWLEDGMENTS
BIBLIOGRAPHY
REFERENCES
1. B C Deloach. 'A New Microwave Measurement
Technique to Characterise Diodes'. IEEE Trans.
Microwave Theory Tech. MTT-12(1): 15-20 (1964).
2. Penfield P and Rafuse R P 'Varactor Applications'.
M.I. T. Press 1962.
3. Wei-Ching TSAI, F J Rosenbaum, L A Mackenzie.
'Circuit Analysis of Waveguide-Cavity Gunn Effect
Oscillator'. IEEE Trans. MTT Vol. MTT 18 No. 11
November 1970.
55
Chapter 7
Linear
Integrated
Circuits
A R Townsend
GEC Semiconductors Limited
The Linear Integrated Circuit (LIC) is, in widest
terms, a large number of different circuit elements
conveniently encapsulated into a small package to
provide an output which is some linear function of
the input. This does not preclude the possibility of
some circuits realising simple boolean expressions. In
theory a complete complex circuit can be reduced in
size to one multiterminal component; complex but
convenient; simply add a mains supply and switch on.
However, in practice, perfection is seldom achieved.
Input/output interfaces are required, control of circuit parameters may be necessary, interfaces with
other parts of a system, and power output requirements that exceed the potential of the LIC all
demand that it is supplemented by discrete components. Designer'> can seldom afford or achieve
complete integration of their circuits into one
package, notable exceptions being in the radio receiver industry, where all but the aerial, loudspeaker,
tuning and volume controls can be fully integrated.
LICs currently available may be d.c., AF, RF, or
wide band, single or multipurpose 1 Most amplifiers
aimed at the AF market are d.c. coupled, with useful
gain into the RF regions; care must therefore be
taken in the design and layout to prevent high
frequency oscillation. Multipurpose circuits may be
considered as 'gain-blocks' with the ability to realise
many circuit functions with the aid of different
feedback configurations. By far the most widely
known of these "gain-blocks" is the operational
amplifier or op-amp. This chapter will tend to refer in
particular to the ubiquitous '709' as it is the
operational amplifier that most engineers will run
into first, and consequently the one with which they
may have the most problems.
CONSTANT
FAILURE
RATE
PERIOD
EARLY
FAILURE
PERIOD
'
'
II
I
'
:
I
i~l\:
I
I
The customer may wish to perform his own acceptance testing, be it batch inspection by the firms
Incoming Goods department or, on a smaller scale, on
an individual basis by the engineer. This chapter
should provide the means to that end.
High reliability, higher even than that achieved by the
methods discussed above, coupled with the probability of receiving fewer rejects, may be realised by
extra factory screening. One of the most widely used
screening operations is the 'bum-in' screen 2 During
burn-in each circuit is biased in a 'use' condition at
maximum voltage levels and at the maximum
guaranteed temperature condition, generally 125C.
This condition is maintained usually for one week
(150 hours). This process removes any potential
early life failures and by so doing re<Jults in a more
reliable batch. Data that has been accumulated on
burned-in circuits has shown that elevated temperature provides a means by which failure occurrence
can be accelerated. The same data has revealed that
reliability failure rates for microelectronic circuits
improve with operating time. Thus the failure rate
(see Fig.1) for any given group of circuits can be
improved by the application of a burn-in screen.
The standard Acceptable Quality Level (AQL) for
off-the-shelf parts is 1%. Extensive screening improves
the AQL by approximately one order of magnitude.
II
,_+0-111_ _ _--r'l
I
R,. =250,000.0.
A,.=50,000
Rout= 150.0.
NON-INVERTING
INPUT
R,n =CX)
A,.=oo
RourO
METHODS OF TESTING
WEAR
OUT
FAILURE
PERIOD
I
II
EQUIVALENT CIRCUIT
(PRACTICAL)
TIME
EFFECTIVE
BURN-IN
PERIOD
TIME
57
Acceptable figures
commonly found
90 dB (x 50,000)
250kf2
150f2
Infinite band-width
2.0mV
90dB
50p.V/V
a
GENERAT
. _ , o - - - - - - - - - . - I X OSCILLOSCOPE
2MSl
TO'SCOPE
HORIZONTAL
INPUT (X)
R,
:!:5V
1KHz
N
TRIANGULAR
58
500kSl
son.
IOpF
c
-NV--.-ct---,
15Vac
( SIGNAL GENERATOR)
OR MAINS 50Hz
---o
----+--4--~~~~--~
0/P
>-----+---IY
SIJW-TOO
GAIN (A l:~
VJ' BC
- - I N P U T VOLTAGE----+
e
f-
~ ~+--f--#-~+-~
5
INPUT
INPUT
:!:mv
~
b
+15V
E
~
Fig.4. (a) An ideal circuit for measuring input offset
voltages. (b) A practical circuit. 5
Rz
properties of the LIC accurate measurement can be
obtained. Consider Fig.4b. The amplifier is given a
closed loop gain of 100, and instead of the inputs
being adjusted for zero output they are set to the
same voltage (0.0 volts) via low value source resistors.
Any input offset voltage previously required to null
the output now appears at the output conveniently
multiplied by 100, the overall designed gain of the
circuit.
Measurement is now possible without disturbing any
of the voltages due to the output having low
resistance. In extreme cases where a DVM is not
available, a multimeter would give sufficient indication to determine the state of the input offset
voltage.
+V
OOIJLF
50.Q.
15k
::!:5V
200pF
-15V
0006JLF
59
c
d
e
f
g
15
15
15
15
9
9
9
9
15
15
15
15
9
9
9
9
+5
-5
+5
-5
+5
-5
+5
-5
10kU
10k
10k
10k
2k
2k
2k
2k
+ 12v
-12
+ 6.5
-6.5
+10
-10
+6.0
-6.0
SIGNIFICANT PARAMETERS
From the six checks examined above only two are of
immediate interest to the engineer involved in trouble
shooting. These are transfer function, generally a
go/no-go test, and input offset voltage. Having
established that an LIC is operational and complies
with the primary requirements, other parameters may
be neglected unless further evidence suggests otherwise. This is due to the initial design and fabrication
of integrated circuits which provides, as implied, a
full integration of the combined parameters. The same
manufacturing processes which result in low input
offset voltages and high amplification are also responsible for input offset currents, input resistance,
output resistance, and output swing. Thus from a
possible 30 or more tests which may be performed on
an LIC only two need be performed to give a
reasonable assurance of circuit integrity.
6. Vout
or expressed in dB's:
CMRR = 20 log (5 + 5) x 100
6. vout
100 = 20 mV
1 Correct orientation
2 Soldering irons
3 Power supplies
Resistors used by the manufacturer in the measurement of input offset voltage will be selected to 0.02%
accuracy but the engineer will have available standard
2% .components. In the circuits given these may be
used with confidence and still only result in 10%
errors. For the situations described this will prove
completely adequate. Measuring instruments, their
calibration and general accuracy, coupled with any
loading effect that they may have, give rise to more
errors. For example; a multimeter (Zin = 20 OOOQ
per volt) used to measure output voltage will have
considerably more loading effect than a digital volt
meter (Zin = 5 000 MQ). However, inexpensive
portable digital multimeters are currently available
and exhibit much higher input impedances, generally
>10MQ.
General
4 Spikes
5 Oscillation
INTEGRATED
INTERPRETATION OF SPECIFICATIONS
Table 3. Expected values for an operational amplifier
under bench tests.
Parameter
Expected values
50,000
5mV
50nA
10V
100dB
50JJ,V/V
10% Dependent on
measuring
20% instrument
5%
20%
20%
6
7
36V
300mW
5.0V
10.0 v
5sec
-65C - + 150C
-55C-+ 125C
300c
Conditions
Rs~10kf2
350
0.6
10
100
700
150
2.5
2.0
50
200
mV
nA
nA
k!"l
3.6
mA
16
Average temperature
co-efficient of input
offset voltage
Rs; 5of2,
T A ; +25C
toTA ;+125C
Average temperature
co-efficient of input
offset current
TA; +25C to
TA; +125C
1.8
10 IJ.V/C
0.08
0.5 nA/C
18
CMRR
19
SVRR
Rs~ 10kf2
40
20
15V, Rs 25,000
2 2kf2, Vout
15V
21
62
V5
REFERENCES
-
17
Rs~ 10kf2
8.0
80
110
12 14
dB
100 IJ.VIV
70,000
Chapter 8
Digital
Integrated
Circuits
R 5 Whiskin BSc(Eng)
GEC Semiconductors Limited
Digital circuits form the great majority of integrated
circuits manufactured today. There are many ranges
available from various manufacturers and these tend
to be classified by family type. They include Resistor
Transistor Logic (RTL) of Fig.l, Transistor Transistor
Logic (TTL) and Diode Transistor Logic (DTL)
examples of which are shown in Figs.2 and 3, and
Emitter Coupled Logic (ECL) shown in Fig.4. RTL
was one of the earliest types to be introduced, but is
now obsolete and rarely used in new designs. Both
DTL and TTL are forms of Saturating Logic, so called
because a transistor in the output stage saturates
when the device operates. ECL is a form of Current
Mode Logic which is nonsaturating and so achieves a
faster switching speed. MOS transistors can be used
with certain advantages in integrated circuits and
these form a separate category.
Over the years certain types of logic have emerged as
being more popular than others and because of this
have become "second sourced" by several manufacturers. The prime example of this is the 7400
series of TTL which is now made by at least a dozen
manufacturers throughout the world. This chapter
will detail only the types which are popular and
therefore most likely to be wcountered by the
reader, however, the various test methods and procedures discussed can be applied to all logic families.
Vcc+5V
Vee+ 36V
175K
650
OUTPUT
OUTPUT
~
A
II
.-------......---~~-------.......... 11
Vcc+5V
----~~--~-----0
NOR
OUTPUT
INPUTS
-5V
Because of the variation of test methods and procedures that can occur between different manufacturers, national standards have been established to
ensure conformity in the quality level of integrated
circuits irrespective of manufa<;turer. The BS9400
specification for integrated circuil:sapplies in the UK
and MIL-STD-883 is the American standard. This
chapter makes reference to the BS9400 specification
for the basic parameters to be tested and the methods
of test.
64
OUTPUT
H
L
H
L
H
H
L
L
L
H
H
H
TRUTH TABLE
r-----------.---~
Vcc!+5V)
OUTPUT
INPUT
CURRENT
(INPUT IN
LOW STATE)
~=2AVMINKNOISE
V (0 8V) ------l[ )
lEI--~:......,..;,...-.,......
IL .
~
V1H=20V
MARGIN
~8------
65
r,
-1 )-)--...
HIGH STATE
CURRENTS
LOW STATE
CURRENTS
Fig.9. High and Low state currents {or DTL and 1TL.
High State
Low State
66
OUTPUT
CURRENTS
INPUT
OUTPUT
loH
loL
0 Vee
r - - - Vee
GATE I
,;R4
Vour
GATE2
Cs
GATE 3
DTL GATE
OUTPUT
DTL GATE
OUTPUT
(a)
(b)
Fig.JO. (a) Capacitative load on DTL gate. (b) Fanout improves rise times.
COMPLEX DEVICES
The basic parameters of current, voltage and delay
times can also be applied to devices more complex
than individual gate circuits. The simplest of these
include such functions as JK Clocked Flip-Flops,
which are classified as Sequential Bistable Circuits in
BS9400. The operation of these devices is controlled
by a clock signal sometimes called a trigger input, but
apart from the fact that there are more input and
outputs to be considered, which makes the functional
testing more difficult, the measurements of the
parameters can still be made in the same way.
This also applies to Medium Scale Integration or MSI
as it is commonly referred to. The more complex
devices such as binary counters, shift registers,
decoders etc. are usually fabricated using TTL technology and as a resuit have inputs and outputs which
may be regarded as being identical with the inputs
and outputs .of simple TTL gates. The same test
methods and limits can therefore be applied. The
major problem with MSI devices is the functional
testing, which is clearly related to the complexity of
the device. For very complex functions this can
present a major problem.
MOS CIRCUITS
MOS integrated circuits may be tested using basically
similar test methods to those used for bipolar circuits.
The main difference is that the devices require much
higher supply voltages (20 to 30 V) which are
negative, as against the 5 volts positive VCC generally
used for bipolar circuits. The signal input voltages are
related to the MOS gate threshold voltage which can
vary depending upon the actual manufacturing process used. These processes include thick oxide, nitride
and silicon gate technologies. Typically the input
threshold voltages range from 1.5 V to 5.0 V. Output
voltage swings are generally greater being related to
the higher supply voltages. Control or clock inputs
also require large voltage swings.
Unlike bipolar circuits there is a large variation in the
basic parameters of MOS circuits, due to the variety
of types available and as a result it is difficult to be
specific. However, the test methods described are
generally applicable to MOS circuits, once the
magnitude and limits of the various parameters have
been established.
TEST METHODS
( 1) Low state output voltage VoL (BS9400 Method
2001). In this test (Fig.ll) the gate output is Low
and the voltage is measured while an output current
IoL flows into the output terminal to guarantee the
fan-out capability. RL is varied to set IoL VIH is
applied to the inputs of a NAND gate to guarantee
noise immunity. This also applies to a positive NOR
gate, except that each input in turn has VIH applied,
the others being held low.
67
74TTL
930DTL
VOL
04V(022V)
045V (02V)
IoL
16mA
VrH
VrL
ECL
(-1 6V) TYPICAL
VALUES IN
BRACKETS
133mA
20V
19V
oav
II V
74TTL
930DTL
16mA
15mA
0-4V
045V
ECL
-16V
Fig. 11. Test circuit for Low state output voltage (VOL)
and device limits.
o---+---+-----.L---<J
74 TTL
V0 H
24V(33V)
930 DTL
26V(40V)
ECL
(-0 8V)
180fLA
V1H OR V1L ARE AS FOR V0 L TEST
68
REFERENCE
TYPICAL VALUES
IN BRACKETS
Vee MAX
9)
OPEN CIRCUIT
v,N
REFERENCE
74TTL
r,H
IOmA AT V1N=55V
40p.A AT V1N=2-4V
930 DTL
5p.A AT 26V
ECL
15Cf'A AT -08V
v.
PROPAGATION TIME
RL
~g~Db~t ~~gu~TTL
I
69
from the
terminals.
functions
tween the
Temperature testing
Digital integrated circuits are designed to operate over
a certain temperature range. Data sheets give the d.c.
parameter limits for the extremes of the operating
temperature range (eg 0C and 75C). The test
methods described can be applied to measurements at
the temperature extremes but it is usually only
necessary to do this on a sample basis to assess the
quality of the whole batch. This is the procedure
employed by manufacturers.
PRECAUTIONS
A large proportion of digital integrated circuits
returned to suppliers as having failed, , prove on
examination to have been subjected to an excessive
electrical overload. This accidental damage by the
customer can occur as a result of carelessness during
his test procedures or due to his lack of understanding of the limitations of the circuit.
Electrical overloads at the output of a DTL or TTL
circuit usually occur if the output is accidentally
shorted to V CC If the output is in the Low state
excessive current will flow damaging the output
transistor. This may occur when probing on a board
full of circuits during commissioning or fault finding
procedures. There is also a maximum limit on supply
70
CONCLUSION
The test procedures outlined in this chapter can be
used to ensure that devices meet their electrical
specification with the aid of very simple test equipment. Several manufacturers market integrated circuit
testers which range from simple manually operated
equipment up to fully automatic computer controlled
testers, similar to those used by integrated circuit
suppliers. The user can, therefore, easily build or
select the equipment which exactly suits his requirements.
Chapter 9
Electronic
Tubes
GRIDDED VALVES
A gridded valve can be defined by either its characteristics or its parameters, which are the slopes of these
characteristics. The most important characteristics of
the tube show the behaviour of the output electrode,
the anode, with respect to the input electrode, the
control grid. In a triode these are the only input/
output characteristics, but the same information can
be presented in three different ways. Figs.l, 2 and 3
show this for the ITT triode type 5924.
When the tube includes an extra grid, as in a tetrode,
it is necessary to show the effect of this grid upon the
anode current as it modifies the effect of the control
grid. This is achieved by showing several sets of
characteristics at differing screen grid voltages. It is
71
15
450
400
I Va=Vg
I
I
300
<(
~10
a:
a:
:::>
u
200
0
0
z<(
100
0
-100
7
-250
-150
-100
>
~
~
~
~
0
!
2
I
-100
05
-200
-300
-400
-500~0-~-*-~~~--~--~--~~
2
3
4
56
ANODE VOLTAGE (kV)
72
1-
z
.....
lE:::>
u
Vg,= 400V
300V
.....
ll!u
<J)
~ov
0
200V
14
10
8
6
ANODE VOLTAGE (kV)
Vg 1=400
300
200
100
10
8
6
ANODE VOLTAGE (kV)
12
14
(b)
Vgt , Va, Vg 2 , otherwise excessive screen grid dissipation (Vg 2 X lg 2 ) can occur. The reverse order should
be empfoyed when switching off.
GRID
VOLTAGE
ANODE
WAVEFORM
GENERATOR
GRID
WAVEFORM
GENERATOR
73
gm ra
74
oj~E,f] ~E,
pE
GRID
BIAS
HT
SUPPLY
(A)
(B)
When a.c. methods are employed, however, it becomes necessary to ensure that the various test
voltages are accurately in phase and that stray
reactances, such as those of the interelectrode
capacitances of the tube, do not introduce serious
errors. To obtain the best null it is desirable to make
the detector immune to noise generated within the
tube under test. As reasonable sensitivity is desirable
a detector offering gain to a narrow band of
frequencies is required, tuned to the test oscillator
frequency.
(A)
(B)
E 2 /ra = E 1 /Rs
ra = E 2 .Rs/E 1
(A)
(8)
75
Precautions
1. In circuits for the measurement of both amplification factor and mutual conductance it is necessary to
ensure that the resistance of the voltage source is low
enough so that voltage E 1 is actually applied to the
grid without change of magnitude or phase.
(A)
(8)
Thus:
and, since n is usually large, 10 or more, it is possible
to neglect the last term. Thus:
78
-20d8
Pour
dBm
DuAL
DIRECTIONAL
COUPLER
79
c,
TRANSMITTING APPARATUS
In many ways the checking of transmitting apparatus is
similar to the checking of IH apparatus. It can be shown
that the 'knee region' of a valve's characteristic has a
considerable effect on its performance in transmitting
apparatus and in order to make measurements to
explore the characteristic in this region it is necessary
to know the grid drive accurately. In transmitters the
drive depends on the condition of the penultimate
stage and not the feedback as in IH apparatus. A very
rough approximation to the conditions at the grid can
be obtained from the grid current (if grid current is
drawn) but to make measurements with any accuracy it
is necessary to measure grid drive by means of a peak
reading voltmeter as previously described.
The second important parameter that needs investigation is the anode loading. Here the transmitter differs
from the IH apparatus. The IH output will receive a
varying loading in most cases, but the transmitteris
designed to work into a constant load presented by
an antenna system.
A tetrode or pentode will give a good indication that
the anode is swinging correctly by the sharp rise in
screen grid current that occurs when the anode swings
below the screen grid voltage. In some ways this is a
better indication than any other, but it only operates
when the valve is fully driven and reveals nothing in,
say, a half drive condition that may be met in an SSB
equipment.
c2 ;c 1 + 1
Since C 1
X peak voltage at A
~ C2
= C 1/C 2 X peak
voltage at A
BOX CONTAINING
REMAINDER OF
CIRCUIT
(A)
(B)
81
WATER
OUT
= 4.18
F
X 60 X 1.06 X 240 X T
= 17.7 X F X T watts
Where Fin m 3 /min
Tin C
82
Panode- 4.18
X F X 10--"3
60
X 10 X 10 T
= 70 X F X T watts
Where F in litre/min.
and Panode = 314 X F X T watts
Where F in gal/min.
Cooling
Panode
or
= 4.18
BIBLIOGRAPHY
Class C Amplifier and Oscillator design - a simplified
version. L.T. Apps. Electronic Engineering Jan. 1955.
Calculation and Design of Class C Amplifiers. F .E.
Terman and W.C. Roake. Proc IRE 1936.
Linearity Testing Techniques for Sideband Equipment. P.J. Icenbice and H.E. Fellhaver. Proc IRE Dec.
1956.
Linear Power Amplifier Design. Warren B. Bruene.
Proc IRE Dec. 1956.
Chapter 10
Relays
and
Switches
J J Langridge
1.
2.
4.
3.
5.
X3
Fig.l. Types of relays. From left to right: telephone type relay; general purpose industrial relay; microswitch relay; reed
relay; delay/timing relay.
TESTING
The first aspect of testing must be the level of
inspection and the user must determine this level
based on his knowledge of the product, the standing
of the manufacturer and the in-house facilities which
are available. He will have to decide whether the units
need 100% viewing, a sample inspection, or even no
incoming operation, ihe company relying solely on
the final test of the equipment manufacturer to find
any defects. To decide on this, the probable defect
rate has to be known. In general a figure of not more
than 1% is accepted for the relays within the cate-
2.
3.
4.
5.
Relay types 1, 2, 3, 4, 5,
1 (Visual), 2
1, 2, 3, 5
3, 5
3
84
r------,
COIL
' - - - - - - - . , - - - - - - SUPPLY
MICROSWITCH
OPERATE
DELAY
CONTACT
CURRENT
COIL
VOLTAGE
_4,_:
11
I I
I I
1 1
RELEASE
~DELAY
------~j~r'----~~----
t
-TIME
85
well or not. The common specification for all contacts is that they should not be more than 1 /3 of
their diameter out of alignment, and if this is adhered
to, the capacity of the device will not be impaired.
High voltage insulation testing is best carried out after
the relay has been assembled into the equipment,
since socket or tag wiring is most prone to leave sharp
solder points which can provide unforseen flash-over
points. The easiest type of equipment with which to
carry out such a test is the non-destructive flash tester
which has a mains driven transformer giving a variable
high voltage output, the current from which is limited
by a capacitor, and indicated by a meter. This will
provide several different levels of test to cover the
various figures quoted by relay and switch suppliers.
Breakdown testing is most important when either
inductive circuits are to be switched and hence high
back EMFs are probable, and also where contact sets
on the same relay are to switch separate circuits at
widely differing levels.
Many types of modern equipment need delays built
into their operation so that mechanical. movements
are not 'beaten' by the electrical control circuits.
Hence relays which have to delay an output signal
need to be checked out with care before they are
mounted. The normal way to effect this check is by
means of an electronic timer, the digital type being
the easiest to use. In general terms the principle is to
start the timer with a pulse from the relay coil start
current, and to stop it with the opening or closure of
a set of contacts on the relay.
It is most important to realise that most timers will
r------,
OS
l ITrf
-------,
~
~
'---i--:--ll_j
, ~~,~rr----+~
I RELAY UNDER I ",!L___- 7 - - - - - - - e COIL
LTS,.SI_ - - - - J
Y,
SUPPLY
MICROSWITCH
COPE
PLATE
y
IOOmA
FS.D.
IRELAY UNDER
.Jf2T_--IOOmA SUPPLY
d. c.
86
PLATE
RELAY SELECTION
The selection of the correct relay for a particular
application can provide a difficult problem to an
engineer if the performance needed is a complicated
one. The following factors should be considered.
1. Size.
2. Operating voltage.
3. Maximum coil dissipation.
4. Number of circuits to be switched.
5. Contact current handling capacity.
6. Insulation breakdown voltage.
7. Type of contact load (resistive, inductive, etc.)
8. Coil voltage range.
9. Environmental immunity.
10. Operate/release times.
11. Method of mounting.
12. Method of connection.
GOLD PLATED
SPRING WIRES
MEASURING
CIRCUIT
1
IOOmA
SUPPLY
TEST
COIL
COIL
SUPPLY
d.c.
INSULATING WIRE
OPENING CAM
may require regular replacement. In counting equipment, reed switches are often used because of their
high speed, and because they can be tripped by a
magnet on the unit to be counted, eg a coil winding
machine. Here the limitation may come solely by
virtue of the large number of operations involved, and
the mechanical life of the device must be considered
when specifying for such an application.
In choosing the operating voltage for the relay coil, it
is better to err on the low side for the supply so that
the relay is not too heavily energised. A high operate
87
88
MANUFACTURERS' SPECIFICATIONS
Although each manufacturer will specify his relays in
a different way, the main points are common and can
be correlated easily. Firstly, the coil voltage is given
either as a continous rating or for intermittent operation. A continuous rating only means that the coil
can remain energised at its rated voltage indefinitely
if the surrounding air is sensibly at room temperature.
This could restrict the use of the relay if it is to be
used in high temperature conditions such as furnace
control or oil burners. In such cases, it is best to
reduce the drive voltage after the relay has operated
so that the dissipation is reduced.
For intermittent operation, the manufacturer usually
states a duty cycle which again must be reduced if the
temperature is excessive.
In the specification of contact sets, some light duty
contacts have twin contacts. This indicates th~;~.t each
circuit is connected by two contacts on the same
spring but separated by a slot in the spring so that
both contacts can be adjusted by the manufacturer to
make contact at the same moment. This provides a
higher degree of reliability in operation, and also a
lower contact resistance in general. This type of
contact is often used for low level audio switching
and especially in the telephone industry.
The life of the relay is often quoted as a mechanical
life, and this must not be taken as the life to be
expected if the current switched by the relay is near
the stated maximum. Here a manufacturer can
usually indicate a probable life, but a life test will
provide a result with a higher confidence level.
Approval by such bodies as BSI or the Post Office
will always indicate that the specification meets the
basic requirements for the type of relay involved, and
can be used as a guarantee of the design approval of
the relay.
Glossary of Terms
(electronics,
voltage:
Accelerating
oscilloscopes) the cathode-to-viewingscreen voltage applied to a cathoderay tube for the purpose of accelerating
the electron beam.
Air cored cons: (electronics) coils with
no magnetic material used in their
construction; include coils wound on
non-magnetic formers as well as
airspaced coils.
Alpba cut-oft': (electronics, semiconductors) that frequency at the
point at which the current gain of a
transistor has fallen to 07 of its low
frequency value.
Alternate mode: (electronics, oscilloscopes) a means of displaying output
signals of two or more channels by
switching the channels, in sequence,
after each sweep.
Anderson bridge: (electronics) a modified
form of Maxwell bridge used for
measuring inductance in terms of
capacitance and resistance.
A.Q.L.: (electronics) a point on the
operating characteristic of a sampling
scheme which defines the quality level
at which there is a reasonable
probability (usually 095) that the lot
will be passed as acceptable by the
sampling scheme.
Audio frequency choke: (electronics) an
iron-cored inductor designed to offer
high resistance to audio frequency
signals.
triggering: (electronics,
Automatic
oscilloscopes) a mode of triggering in
which one or more of the triggering
circuit controls are preset to conditions suitable for automatically
displaying repetitive wave-forms. The
automatic mode may also provide a
recurrent sweep in the absence of
triggering signals.
Auto transformer: (electronics) a wound
transformer in which part of the
winding is common to both the
primary and secondary windings.
Bandwidth: (electronics) a statement of
the frequencies defining the upper and
lower limits of a frequency spectrum
where the amplitude response of an
amplifier to a sinusoidal waveform
89
90
91