I am also interested in audio HiFi design, and have wondered why some IM tests
for products in that field use three (non-harmonically related) frequencies,
while most only use two. Using three well-spaced frequencies should greatly
increase the chance that IM signals will be detected, since there is a greater
chance that such frequencies wouldn't get filtered out. I suppose that the
reason that some still use only two frequencies, of course, is that with
properly selected frequencies one can hide one's product's poor linearity.
Frequencies of 400, 500, 900, 1100, 2300 and 2600 Hz to a mike input should
provide a particularly good test, although the signal generator would be more
complex than the usual. An audio clipper following these frequencies would
probably show a mess.
TNX, Gene May
Amps mailing list