[Arny Krueger]

wrote in message


GIven the way THD is measured, an amplifier will
have a THD spec which goes as the reciprocal of power,
simply because THD is a broadband measurement and
can't distinguish noise from harmonics. This is why the
correct designation is THD+N.

Thus, with a constant noise floor, for every halving of
power, the THD+N figure will rise by about 40%.

Secondly, THD as a measure of audibility has been pretty
much shown to be useless, regardless of the claims
seen here and elsewhere. These claims simply ignore
significant factors such as masking.

If you're claiming on reliably and repeatedly hearing
duifferences at the levels of distortion you're claiming,
look somewhere else, it ain't THD.

Agreed. THD|N measurements often raise unecessary concerns by presenting
data that might in the worst case result from low-level distortion like
crossover distortion.

With FFT-based measurement software, ignoring almost all of the noise floor,
and just measuring the actual harmonics is both feasible and often
implemented.

AFAIK this is done by first detecting the frequency of the fundamental. Only
power that is in the FFT buckets that correspond to integer multiples of the
fundamental are included in the final summary.

Similar techniques can be implemented for IM measurements.

When techniques like these are utilized, it is often found that SS
amplifiers have negligable crossover distortion.