A/VwrM&d

Measuring and Analyzing

Intermodulation
VJ
C. j. LE BEL"

The oscillographic method is developed to give quantitative results, based upon the
experimentally found relation betwegn total notch depth and IM percentage.

the amplitude of the fith harmonic

would be multiplied by nf2, which
leaves the second harmonic unchanged.
A more recent proposal was made by
Fig. 1. Typical inter- Shorter: who proposed a weighting in
modulation analysis which the amplitude of the nth harmonic
circuit. would be multiplied by ne/q, and in
which harmonics weaker than .03 per
cent would be, neglected.
A fundamental problem of the har-
monic method is that of achieving a pure
waveform at the input to the equip-
HE INTERMODULATION MEASURING modulation and the aural effect rise merit under test, for. the harmonic

T DEVICE to be discussed is a quanti-

tative development of an oscillo-
graphic method hitherto considered only
greatly.
The harmonic method may be satis-
factory if we measure the relative am-
measuring device cannot distinguish be-
tween an imperfection of the source
waveform and one produced by the unit
being tested- At any one frequency, Pure
qualitative in nature. I n the course of the plitude of the individual harmonics, then
research, gross inaccuracy was found in waveform can be achieved by an inex-
the traditionally accepted 4 : 1 relation- pensive filter, but when testing over a
ship between intermodulation and har- wide frequency range the pure wave-
monic distortion. d
form source becomes more expensive
Intermodulation meters built accord- - than the dis'tortion meter itself.
ing to Hilliard's design1 have had pro- The intermodulation method presents
vision for connecting an oscilloscope no such waveform problem, for the wave-
into the circuit, but the first published form of ordinary laboratory oscillators
(A) (6)
study of oscilloscope images in distortion [Continued on page 201
measurement may be credited to Mc-
Pro~d.~ Fig. 2 Envelope of oscilloscope images without
- intermodulation (A), and "ith intermodulation E. L. Shorter: The Influence of
(B). High Order Products in Non Linear pis-
Harmonic vs. lntermodulation Diatortion- tortion. Electronic Engineering, vol. 22,
A Review multiply each by a weighting factor. 'no. 266, pp. 152-153, April 1950.
I n typical audio applications, distor- This requires a wave analyzer, whose
tion measurement must yield a number use is tedious and time consuming; and
which bears some relation to the offen- little has been published on the weight-
siveness of the sound to the ear-for ing factors which would equate the or-
audio equipment is usually made to be der of a harmonic and its offensiveness -.
listened to. to the ear. An early step was the 1937
By this test, the harmonic method of RMA proposal3 according to which
measuring distortion may be unsatis- 3 Radio Manufacturers Association : Spe-
factory. I t fails to indicate the bad effect cification for testing and expressing overall Notch Depth- #(In Percent)
of polishing a disc master, while the in- performance of radio broadcast receivers-
termodulation method succeeds. The Part. Z-Acoustic tests ;-P. 5 ; tDee. 1937
writer has found that poor tracking (of reVIS1On. Fig. 3. Definition of notch depth.
a phonograph pickup stylus in its
groove) may produce hardly any effect 100
on harmonic distortion, while the inter-
Z

*Audio Instrument Co., 133 W. 14th St., 2

I-
so
N. Y . 11, N . Y . a
-I
1John K. Hilliard: Distortion tests by
3
o 60
the intermodulation method. Proc. I.R.E., Fig. 4. Relation be- 0
vol. 29, no. 12, pp. 614-620, Dec. 1941. tween notch depth B
2 C. G. McProud: Simplified intermodu- qnd per cent inter- 40
lation measurement. Am10 ENGINEERING, modulation. z
vol. 31, no. 3, pp. 21-23, May 1947. I-
5 20
0

paper rests upon the author, and state- 2 0

0 50 100 150 200 250 300
ments contained herein are not binding
PER CENT NOTCH DEPTH

18 AUDIO ENGINEERING JULY, 1951

 of the two frequencies. A typical dircuit stage either two or four notches in the
is shown in Fig. I. pattern will ordinarily appear, depending
4. Observe the filter output on an on conditions. If. the bias is too low,
oscillosco~e whose sweep is s~nchron- two narrow notches may appear toward
ized to the lowfrequency tone. one side of the patterns as in Fig. 5. If
A. If the high-frequency tone is not the grid is driven heavily toward cutoff,
modulated by the low-frequenc~ broader notches will appear toward the
(i.e., in- opposite side of the pattern. W e have not
termodulation), the oscilloscope used the words and "left" to
Fig. 5. Envelope of notch pattern with in- screen will show a smooth rec-
sufficient bias, for single-ended stage. tangle of light, like (A) of pig. refer to notch position because the posi-
2. tion depends on the number of amplifier
is satisfactory, and it is easy to mix two B. If one tone the other (by stages and on the oscilloscope circuit.
definition an intermodulation con- In a simple single-ended circuit with the
tones without creating the intermodu- dition), the rectangle of light most common oscilloscope character-
lated source tone that would be a cause will be marred by one or more istic, bias notches will be at the left side;
of error. notches, as in (B) of Fig. 2. and notches due to grid cutoff at the
The intermodulation method was first 5. There is a quantitative relation be- right side. If an amplifier is heavily
used extensively in the film industry, tween the size of the notches and the in- overdriven, both bias and cutoff notches
after the harmonic method had failed to termodulation percentage. Each notch has may appear-four in all.
serve as a satisfactory guide to the ear's a notch defith, which is defined in Fig. 3, a push-pull stage, four notches are
opinion of film distortion. Roys has made and which is best expressed in per cent. standard, as in ~ i6; and ~ in. a perfectly
An image of this type will generally have
extensive use of intermodulation in vari- more than one notch, each with its own balanced stage, the notches are equal in
ous problems of disc recording and re- depth. The following relation is then size. If the push-pull stage is driven by
producing. H e has shown that it is a re- used : a single-ended stage which overloads
liable index to processing faults5 where ~ - 1 notch depth= ~~~~h~ depth + easily, one of the notches will diminish
the harmonic method is valueless, and is in size and may even disappear, as in
also very useful in judging the tracking
of disc reproducerse and in studying The fact that these notches occur on
tracing distortion. top or bottom of the pattern indicates
that intermodulation can occur on either
W e may theorize that the intermodu- positive or negative half cycle of the
lation method more nearly correlates
with the ear's opinion because the result high-frequency wave. Since the notches
is automatically weighted by the order are not necessarily symmetrical, it be-
of the distortion. There has been little comes clear that the rectifier in an
attention to this, and it would be a real Fig. 6. Normal notch patfern for push-pull stage. intermodulation meter must be wave
contribution to the art if someone were -else it may ignore intermodulation
to study the action of high order distor- effects occurring in the half cycle which
tions. From Frayne and Scoville's work is not rectified.
it is evident that the weighting for This is sufficient to indicate the gen-
higher distortion orders could be in- eral nature of the information provided
creased, if desired, by using a higher by the notches; a full description of the
ratio of low- to high-frequency voltage subject would be a paper in itself.
than the four to one which is custom- I t has been customary to make most
ary. A ratio of eight or ten to one would intermodulation tests with a 4: 1 ratio of
be worth studying. Fig. 7. Push-pull output stage with single-ended low- and high-frequency voltages. This
I t may readily be concluded that the driver stage showing effect of driver overload.
intermodulation method deserves more
intensive us$ than has been customary. notcha depth t etc. This' may lead to a 20
One cause of the neglect has been the total notch depth of over 100 per cent if
high cost of intermodulation measuring each notch depth is expressed in per cent.
18
instruments; another has been the lack The relation between total notch
of realization of the value of oscillo- depth and intermodulation percentage is
scope images as a guide to corrective 16
shown in Fig. 4. This wab experimen-
measures. tally determined, using an intermodula-

A N e w M e t h o d o f Measuring
tion meter similar to that shown by Hil-
liard,l and typical experimental points "
z l4

/
lntermodulation are shown on the figure. The following re

As may be observed, the following

types of amplifiers were used in various
combinations : single-ended and push-
6
a ,o
method deviates from prior practice pull, triodes and pentodes, voltage am-
only a t the finish: plifier and power amplifier tubes, with Y
1. Mix two tones of different fre- and without negative feedback. All con- E

quency, without intermodulation, in any form to the same curve within the limits 'i!
standard circuit. Bridge networks and of experimental error. 6
hybrid coils are the most obvious means. Since this relation is linear over the
2. Pass the two-frequency tone through most important part of the range-the 4
the system under test. lower part-it is possible to use a n oscil-
3. Send the system output through a loscope screen with suitable calibration, 2
high-pass filter, which removes the lower reading the intermodulation effect of
each notch directly on the scale. I n the
.6 H. E. Roys : In@rmodulation distortion less used upper part of the range, the 4
analysis as applied to disc recording and same screen scale may be used in con- AMPLIFIER OUTPUT I N VOLTS
reproducing equipment ; Proc. I.R.E., vol. junction with a corrective graph.
35, no. 10, pp. 1149-1152, Oct. 1947.
6 H. E. Roys ; Determining the Tracking Analysis o f Oscilloscope Patterns Fig. 8. Intermodulation (IM) and harmodic dis-
Capabilities of a Pickup; AUDIOENGINEEB- tortion characteristics of a push-pull amplifier
ING, ~ 0 1 .34, no. 5, pp. 11 & 38-40, May The oscilloscope screen may show a showing that the ratio of the two parameters
1950. variety of patterns. I n a single-ended changes.

20 AUDIO ENGINEERING JULY, 1951

can lead to results of doubtful signifi- modulation distortion and the harmonic much lower ebuipment cost agd clearer
cance if applied universally-a matter distortion produced at the same operat- illustration 01 equipment faults than has
which is evident from a brief considera- ing point. Thus Frayne and Scoville7
tion of the physical relationships in- heretofore been possible.
in 1939 gave formulas which, for the
volved. For example, if we use a test usual four to one ratio of low- and high- I t has been found that the traditional
tone of 40 and 20,000 cps, in 4: 1 ratio, frequency intermodulation inputs, gave "four times" relation between intermocl-
with most amplifiers there will be real intermodulation as 3.2 times the har- ulation and harmonic distortion is of
doubt as to whether low-frequency dis- monic distortion for second order dis- very limited significance. The only way
tortion or high-frequency distorton pre-
dominates in the measurement. If results tortion. They concluded that the overall to determine intermodulation distortion
are to have any significancs separate relation was 3.8 times. with certainty is to measure it.
measurements of low- and high-fre- Hilliardl in 1941 originated the state-
quency conditions must be made. ment that the ratio was approximately
Applying this thought, low-frequency four times. Unfortunately, the word "ap-
distortion should be checked,by using a proximately" was lost when the phrase
low frequency of 40 or 60 cps, and a transferred to the engineering# world's
high-frequency condition whlch would memory, so most engineers have incor-
have relatively no distortion tendency. rectly assumed a rigid four times ratio.
This might be a high frequency of 2000 Warren and Hewlett in 1948 analyzed
cps, and a voltage ratio of 4: 1. Thus the relationship at greater length. For
the fiigh frequency would be primarily the distortion law they first assumed, the
an indicator. ratios became 3.2 times for single-ended
A high-frequency measurement should stages, and 3.8 times for push-pull oper-
be made with 7000 or 12000 cps and,a ation, but if another assumed distortion
low-frequency condition which would law were followed, they found that the
have minimum distortion tendency. This ratio might drop to unity.
could be a low frequency of 100 or 200 In 1950 Roddam8 assumed still an-
cps, and a voltage ratio of 1: 1. other distortion law and found a 2.8
A number of papers have compared ratio for second order distortion. I n the
the intermodulation method with the same year CallendarOobserved that tests
two-tone or CCIF method, claiming on various tubes under various Class
that the latter has a greater sensitivity A and AB conditions gave ratios which
to high-frequency distortion. W e are in- did not agree well with computed values.
clined $0 attribute their result, instead, I n a series of very interesting and sig-
to the fact that they tried, unwisely, to nificant observations on various am-
measure high-frequency distortion with plifiers Pappaslo has observed ratios all
a low-frequency tone of four times the the way from over 6 down to less than
amplitude of the high frequency of which unity.
they were trying to determine the effect. That the ratio can change under
various operating conditions is shown
Background by Fig. 8, which illustates the per-
The examination of oscilloscope formance of an ordinary push-pull am-
images in conjunction with intermodu- plifier with single-ended driver stage.
,lation tests was first suggested by Hil- I t is evident from Fig. 8 and the other
liard1 in 1941, but he supplied no data data that no generalized relation exists
on the types of images to be expected, between intermodulation and harmonic
and their meaning, so very little Ameri- distortion. Under limited conditions a
can use was made of the idea. Some given ratio may apply, but its validity
European use was begun, however. Mc- is very narrow in scope. The "four
Proud in 1947 described the use of times" ratio is a dangerous error, and
screen images for distortion analysis, should be discarded from engineering
but stated that the relation between thinking.
notch size and per cent intermodulation
was only qualitative. As applied to any Conclusions
single notch, in McProud's fashion, this A new method of intermodulation
is correct. but when applied to the total measurement has been shown, involving
notch depth as we have (a method not
contemplated by McProud) his state- J. G. Frayne and R. R. Scoville :Analy-
ment is in error. sis of measurement of distortion in variable
In our experiments we observed a density recording; I. Soc. Mot. Pict. Eng,
new effect: The oscilloscope pqttern vol. 32, pp. 648-673, June 1939.
occasionally shows bulges instead of Thomas Roddam ; Intermodulation dis-
notches, a sign of regeneration, it may tortion ; Wireless World, vol. 46, no. 4, pp.
occur in multi-stage amplifiers if isola- 122125, April 1950.
9 M. V. Callendar : The influence of hiah-
tion and filter condensers are insufficient, order products in -non-linear distortion;
or if an improperly designed feedback Electronic Engineering, vol. 22, no. 272, p.
loop has been utilized. I t seems to occur 443. Oct. 1950.
rnost often in the 5- to 35-cps range, 1; P
: -?a<& : Electronic Development
and explains why some units test sat-
isfactorilv under conventional condi- cation. 19Sb.
-
Laboratory. New York.. ~rivatecommuni-
tions, yet sound bad with
program material.
\
Relations between IM and Harmonic
Distortion
I t has been customary to assume tliat
a fixed ratio exists between the inter-
[Continued on Page 301

AUDIO ENGINEERING 9 JULY, 1951