February 1960 Electronics World
Table of Contents
Wax nostalgic about and learn from the history of early electronics. See articles
Electronics World, published May 1959
- December 1971. All copyrights hereby acknowledged.
Audio distortion is most often
expressed as total harmonic distortion (THD) as opposed to intermodulation distortion
(IMD). As the name suggests,
is a measure of harmonic power content relative to the fundamental frequency (a single
tone) from which harmonics are created.
IMD on the
other hand, is generated from the nonlinear mixing of two or more tones, with the products
being non-harmonically related to the original tones. The author's discussion of audio
frequency IMD applies equally to radio frequency IMD. Intentionally generated harmonic
components can enhance sound quality due to being
whereas IM products create
tones not directly related to the originals. Audio amplifier frequency vs. power specifications
typically include the sum of total harmonic distortion and noise (THD+N), signal-to-noise
ratio (SNR), crosstalk between channels, response flatness, but at least for consumer
grade products rarely include intermodulation distortion. That is probably because very
few people buying stereo systems - even very expensive models - have any idea what THD
refers to and would be totally clueless about IMD.
Intermodulation Distortion: What it is and how it's measured
Fig. 1 - How intermodulation distortion results from nonlinearity.
By Milton S. Snitzer / Technical Editor, Electronics World
An important specification in evaluating hi-fi gear that should be thoroughly understood
by the audiophile.
The earliest and still most common method of checking distortion in hi-fi gear is
by the use of a single input frequency in a harmonic distortion check. (See "Harmonic
Distortion - What It Is and How It's Measured" in our November, 1959 issue.) Although
this method is easy to use and readily duplicated, it may not tell very much about how
an amplifier would handle a somewhat more complex input. For example, if we were to apply
a mixture of two input signals, one at a low frequency and the other at a high frequency,
and we were to check how the amplifier handles these simultaneously and what interactions
there are between them, then we would have a better indication of how the even more complex
speech and music signals would be handled. This is exactly what we do when we measure
intermodulation distortion (IM).
There is currently some disagreement on the value of an IM measurement. There are
many in the field who feel that a total harmonic distortion (THD) test made over the
entire frequency range adequately describes the amplifier's distortion performance. Following
this reasoning, recent standards have been set up for THD measurements but not for IM
measurement. Despite this, a good many manufacturers quote IM distortion figures so it
is important to understand these figures and how they are obtained.
How IM Is Produced
When two different frequencies are applied to a perfectly linear device (one whose
output varies directly in accordance with the input), then the output of the device contains
only these two frequencies. On the other hand, when there is any non-linearity, one of
the signals is changed or modulated by the other. (See Fig. 1.) When this modulation
takes place, additional sideband frequencies are generated just as they are in the modulated
stage of a radio transmitter. These additional sideband frequencies are not harmonically
related to either of the original frequencies. For example, if a large 60-cps signal
is combined with a small 6000-cps signal in a non-linear device, the additional frequencies
of 5940 (6000-60) cps and 6060 (6000+60) cps are formed. What is more, harmonics of the
original two frequencies combine with each other to produce still more "sum and difference"
signals. None of these additional signals was present originally, so that we have now
Fig. 2 - Functional block diagram of IM analyzer. Waveforms at various
points in the instrument are shown in Fig. 1
The amount of this distortion is the r.m.s. sum of all these added sideband signals
expressed as a percentage of the modulated signal. In the example just given then, if
the square root of the sum of the squares of all the sideband frequencies is 0.1 volt
and the amplitude of the 6000-cps signal alone is 5 volts, then the IM distortion is
There are two audible effects of IM distortion: one of these is the result of the
amplitude modulation that occurs and the other is the result of the unharmonically related
added frequencies, Assume you are listening to a Bach Chorale with the sopranos being
accompanied by a loud, bass passage on the organ. If your equipment has high IM distortion,
the voices may have an unpleasant, "fluttery," and unclear sound because these high frequencies
are being amplitude modulated by the low-frequency tones of the organ.
In addition to this effect, any time that you listen to any combination of tones,
or even a single tone that is anything but a simple sine wave, the added frequencies
due to IM distortion introduce a roughness or harshness that seems to "muddy up" the
original tones. Any added frequencies, whether harmonically related or not, constitute
distortion. But when these frequencies are not harmonically related, then the audible
effects are quite unpleasant.
Limits of IM
In measuring IM distortion it is common to apply
to the equipment under test a large low-frequency signal along with a small high-frequency
signal. The usual ratio between the two amplitudes is 4 to 1 (12 db). Low frequencies
in common use are 40 cps, 60 cps, 100 cps, and 150 cps; high frequencies in common use
are 3000 cps, 4000 cps, 6000 cps, and 15,000 cps.1 Depending on which frequencies
are used, and their related amplitudes, different IM distortion figures will be obtained.
Therefore, it is important to state just what frequencies and amplitudes are used in
a given test when specs are quoted. It is also important to state the power level at
which the measurement is made.
The amount of IM distortion that should be present in top-quality hi-fi equipment
has been the subject of much discussion. According to one authority" the following arbitrary
grouping is given (above, right):
Recently the Heath Company has set up its own standards
for IM distortion limits on power amplifiers. Its groupings, using test frequencies of
60 and 6000 cps at a 4:1 ratio, are as follows (right):
In our own lab-tested reports as well as in the specifications quoted by a large number
of companies in the field, an IM distortion figure of 1 per-cent is used. If we find
that an amplifier will not produce its full rated output power at 1 per-cent IM, then
we down-rate the unit. Admittedly this is an extremely strict specification to meet but
we feel that it is fair provided all units are subjected to the same criterion.
IM Compared with THD
In general, when an amplifier has low intermodulation distortion, it also has low
total harmonic distortion (THD). When the IM is high, so is the THD. However, it must
be remembered that these two methods of measuring distortion are quite different, so
it is logical to expect that the percentage figures for IM and THD will not be the same.
Fig. 3 - Relation between IM and total harmonic distortion in hypothetical
The relation between IM and THD in a hypothetical amplifier is shown in Fig. 3. These
curves are fairly typical. Note that at low power outputs (up to about 11 watts in this
case) the IM distortion is less than the THD, as measured at the extremes of the frequency
range to be covered (say 20 or 20,000 cps). At 10 watts output, the IM of this particular
amplifier is 1% while the THD is 2%. It is quite accidental that these two figures, which
we use as our own standards, occur at the same power level. But observe what happens
at the higher power levels. First, the IM goes upward a little more suddenly than does
THD. For this reason, many feel that the IM measurement is a much more sensitive indication
of amplifier overload. Second, the two distortion curves cross each other so that the
value of the IM becomes higher than the value of the THD.
If the harmonic distortion measurements are made at the mid-frequencies (400 or 1000
cps) , our amplifier is seen to produce much less THD. Under these conditions, it is
not uncommon for the value of THD to be less than the value of IM at all power levels.
In the case of our hypothetical amplifier, an output power of 13 watts is produced at
a THD of 2 per-cent. At this power level, the IM is close to 8 per-cent, or almost 4
times as great.
It is quite possible to have a totally different relation between IM and THD than
is shown in Fig. 3. The design of the amplifier being checked, and particularly how it
overloads, may change this picture markedly.
Although IM distortion can be measured with separate audio generators, filter circuits,
oscilloscope or meter indicators, common practice is to combine many or even all the
functions needed into a single instrument. This instrument may be called an IM analyzer.
A block diagram of all the functions performed by such an instrument is shown in Fig.
Two stable audio generators are used, one of these produces the low frequency and
the other produces the high frequency required for the test. In some instruments these
frequencies are adjustable, in others they are fixed. The two signals are then fed into
attenuators for isolation and for adjustment of the amplitude ratio (usually 4 to 1).
Some units permit the user to switch internal v.t.v.m. circuits in at these points to
measure the relative amplitudes of the two audio signals. After suitable mixing, the
combined signal is applied to the input of the amplifier to be tested.
Fig. 4 - Test setup employed for checking IM distortion of a high-fidelity
The amplifier is terminated in an output load resistor. Some IM units even have such
loads built in for limited power use. The v.t.v.m. is then switched across this load
so that the output power may be monitored. It is common to operate the amplifier "wide
open," and simply keep increasing the mixed input signal until rated power output (or
whatever power output level is desired) is obtained. See Fig. 4.
In this regard, it is important to use the same peak output voltage as would be obtained
under single-frequency conditions, since it is usually at the voltage peak that distortion
is generated. If the amplifier is delivering 4 volts of low frequency and 1 volt of high
frequency at the same time, its total output voltage would be 5 volts. The distortion
would be equivalent to that produced by a 5-volt signal. The power that should be produced
by such a signal is E2 (52, or 25) divided by the load resistance.
On the other hand, the true power (as indicated by a voltage measurement taken across
the load) is only 42 plus 12 (16 + 1, or 17) divided by the load
resistance. This means that the indicated power output must be multiplied by 25/17 (or
1.47) when testing for IM under these conditions to obtain an equivalent sine-wave output
power. Hence, if the voltage measures 8 volts across an 8-ohm load, the true power is
8 watts, but the equivalent sine-wave output power is 1.47 x 8, or 11.8 watts.
The amplifier load signal is then applied through an input attenuator to a high-pass
filter. This circuit removes the large-amplitude, low-frequency signal and passes on
the high frequency. If the amplifier under test has IM distortion, the amplitude of this
high-frequency signal will vary, perhaps as shown at the upper waveform No.5 in Fig.
1. If the amplifier has no IM distortion, the amplitude will be constant, as shown in
the lower waveform.
The high-frequency signal is then amplified and detected. (See waveform No.6.) After
detection, the signal is fed to a low-pass filter which removes the high-frequency ripple.
Now the amplitude variation shows up clearly as a low-frequency signal (upper waveform
No.7). Had there been no IM distortion, the output here would be pure d.c. The remaining
low frequency, representing the distortion, is then applied to the a.c. v.t.v.m. circuits
for measurement. Since this meter had previously been adjusted for full-scale reading
with the undistorted high-frequency applied, it will now read directly the percentage
value of the intermodulation distortion.
IM distortion measurements, taken in accordance with any standard method such as the
one described, allow an important specification to be quoted for evaluating high-fidelity
1 This method is the one specified by the Society of Motion Picture & Television
2 "Radiotron Designer's Handbook" Fourth Edition - edited by Langford-Smith.
Posted July 9, 2018