Intermodulation Distortion

Author: N.H. Crowhurst

Low percentages of harmonic distortion may be inaudible, as such, but the same curvature in the amplifier characteristics causes another kind of distortion, called intermodulatlon distortion (IM for short). The effects of this kind of distortion can be audible when the harmonic distortion is not. There are two basic kinds of intermodulation distortion.

The first kind occurs because the amplification changes during a wave as well as introducing harmonics of this wave. This change in amplification will modulate or change the amplification of higher frequencies present in the same composite audio wave and this modulation of the higher frequencies is what becomes audible.

Intermodulation distortion

Suppose that a 60-cycle wave has 5% of second harmonic. This will mean one-half of the wave will get amplified by 5% more, while the other is amplified by 5% less. If the amplifier is also called upon to handle a 2000-cycle wave of much smaller magnitude than the 60-cycle wave, this will also get amplified by 5% more on one peak of the 60-cycle wave than it does on the other peak of the 60-cycle wave. Thus the 2000-cycle wave will be fluctuating in amplitude at the rate of 60 cycles.

Measuring intermodulation

This effect on the 2000-cycle tone is quite audible as a dithery modulation of the tone. It is often noticed in organ music accompanying deep bass tones. If the curvature in the amplifier is of a kind that produces higher numbered harmonics than second or third, it will also produce increasing amounts of intermodulation because the smaller amount of higher frequencies added to the basic fundamental tone produce more noticeable changes in the waveform.

This kind of intermodulation distortion is measured by using two tones, usually a combined audio voltage at two frequencies, such as 60 and 2000 cycles, with the voltage at 60 cycles 4 times that at 2000 cycles. The combined waveform is fed into the amplifier and a special distortion measuring set applied to the output waveform. First the waveform is fed through a filter that removes the 60-cycle component completely. This leaves the 2000-cycle component which fluctuates in amplitude if intermodulation is present. This 2000-cycle waveform is then rectified, which gives a d-c output with the fluctuation riding on it. The d-c component can now be readily removed by passing the wave through a blocking capacitor, and the fluctuation is measured as an audio voltage. By careful calibration of the whole setup the amount of fluctuation at the output can be measured as a percentage of the total output waveform.

IM test setup

Difference Beat

The second kind of intermodulation distortion is caused by two relatively high frequencies producing a combined tone at a lower frequency. If two frequencies are very nearly the same, the combined waveform will gradually move in and out of phase at a rate dependent on the difference between the two frequencies. At one point, the two frequencies will add, producing a double amplitude, while at a point a little later, the two frequencies will subtract, giving an amplitude which is the difference between the individual amplitudes.

If this combination is applied to an amplifier without distortion, the waveform will be faithfully reproduced as would any other waveform. If the amplifier introduces any asymmetrical distortion, however, the upper part, at the peak in the combined waveform due to addition of the two components, will be amplified more than the lower part of the same peaks. This is equivalent to adding a component of the low frequency corresponding to the difference between the individual test frequencies.

Assymetrical distortion

Suppose one of the frequencies is 4000 cycles and the other one 4200 cycles; the difference frequency is 200 cycles, which is quite clearly audible and of a frequency so different from the original frequencies that quite a small percentage of distortion becomes audible.

Test setup to measure IM distortion caused by two high-frequency components

The method of testing for this kind of distortion is to use two oscillators that generate audio frequencies differing by a fixed amount. For example, if we decide to use the 200-cycle difference frequncy, we would arrange that one oscillator give 4000 cycles and the other one 4200 cycles. Or, to test at a higher frequency, when one oscillator gives 8000 cycles the other must give 8200 cycles. The output from the amplifier is passed through a filter that rejects the high frequencies and picks out any component at 200 cycles.

The problem with this method of measurement is that it only discovers whether there is any distortion due to the curvature that would cause second harmonic distortion. Other kinds of curvature will also produce distortion, but will not result in a simple 200-cycle difference tone. Rather, they will cause all sorts of other unwanted tones. For this reason and others too complicated to give a complete explanation, the results of the two methods of intermodulation test and harmonic measurement are not consistent, but they depend on the amount of different kinds of curvature in the amplification characteristic of the amplifier.