There are two main areas of distortion measurement. The most common is total harmonic distortion plus noise – that’s what most audio distortion analyzers will characterize. It’s a good measurement of performance, but the area where it falls apart is measuring distortions at high frequencies, especially in band-limited devices (and so many components are band-limited). All digital systems are limited to half of the sampling frequency, so they’ll automatically be limited with the anti-aliasing filters to 20 kHz or so. That means it makes no sense to measure THD above around a third of that frequency, so the THD readings you do above 7 kHz don’t mean anything. They’ll give you a number, and it’ll probably look really good because the filter is rolling off the harmonics, so it’s really just measuring noise – not distortion.
That doesn’t mean there’s no distortion at high frequencies. Your ear will certainly tell you that there’s indeed distortion. How do you characterize that? In a band-limited medium, such as analog tape in addition to new digital systems, intermodulation distortion measurements are a way to characterize higher frequencies. One type of IMD measurement is the so-called “Twin Tone,” where you take two high-frequency signals (15 and 16 kHz, or 18 and 19 kHz) and you look for the difference frequency component at 1 k. That will give you a true, accurate, and usable characterization of high-frequency distortion – right up to the band edge limit. If your system cuts off at 22 kHz, you could measure 21 and 22 kHz and get a true characterization of distortion at high frequencies.
This method was discovered in the ’40s while measuring optical film soundtracks on which all film sound was done. It was an optical track on the edge of the film, before it started being striped with a magnetic coating for magnetic soundtracks. The problem there was that the upper frequency limit of an optical soundtrack was 7 kHz, so all of those early films from the ’40s, ’50s, and even the ’60s stopped at 7 kHz. People in the film industry and SMPTE recognized that total harmonic distortion measurements above 1 or 2 kHz were meaningless, so they came up with the SMPTE Intermoduldation Distortion Method which used a 7 kHz and 60 Hz signal and measured the inter-modulation products developed from that. It ended up being a realistic, accurate, and useful characterization of the distortion of an optical film system.
My advice is that if you’re looking at a band-limited device, as most things are now, be careful measuring THD above a certain frequency, and use other techniques to get a better characterization of what’s really happening.
Wayne Jones has almost 40 years of experience in the pro audio and audio test and measurement fields. He’s served on various standards and has been a consultant to companies like Intel, Microsoft, and SigmaTel in recent years.