So, the mystery to me is simply that us should rely on a hobbyists attempt to derive the said thresholds from somewhere else. With crude methods and lesser success, I assume, unverified, though.
I rarely deal with psychoacoustic, so please correct me if I'm talking crap.
Sources: Zwicker, Fastl - Psychoacoustics,
researchgate.net, hifi-selbstbau.de
(with additions by me)
First we consider how well a masker-tone with 1kHz theoretical can mask a test tone at 2kHz, which corresponds to second harmonic distortion (HD2) - depending on the masker-tone sound pressure. This is really the worst case scenario. This would mean, for example, that only a single test tone is played at 90Hz and you would see when HD2 becomes theoretically audible at 180Hz.
If our fictitious chassis produces a 1kHz sound with 60dB (see no. 1 in the diagram), then HD2 is masked at 2kHz up to a sound level of 8dB (test tone level). Which corresponds to a perceptibility threshold of -52dB/0.25% HD2 at 60dB.
At a sound pressure of 70dB@1kHz of our chassis, HD2 is masked up to -42dB/0.8%.
At a sound pressure of 90dB@1kHz from our chassis, HD2 will be masked up to -35dB/1.8%.
Zwicker continues:
If very narrow-band noise is used as masker-tone noise, masking works better and at
[email protected] HD2 would be masked up to about 8%.
When critical band wide noise is used, it is found that the masking of HD2 (or higher order distortion) works better towards lower frequencies.
In blue is the masking limit of HD2 (2.2%) at 250Hz and 60dB critical band wide noise. HD2 (0.3%) masking at 1kHz is shown in green.
The "hobbyists" of hifi-selbstbau.de, to whom I
referred in Post#490, have taken a program published in DIN 45631, which can calculate the phonecurves developed by Zwicker and Feldtkeller, and have calculated the masking curves at different frequencies and sound level.
If I have understood it correctly, the program uses one-third octave bands. So a one-third octave band was used as a masker and the masking for HD2 at different frequencies and sound pressure levels was calculated - see the diagram in
Post#490 (or below).
Similar to the above with the masker-tone at 1kHz (the sound produced by our fictitious chassis) and the masking at 2kHz (the HD2 of our chassis).
Because the masker consists of a one-third octave band instead of a single tone, the masking for HD2 is somewhat higher:
1kHz@90dB masking of HD2 with single tone is about 1.8%, with third octave band excitation about 2.7%.
Thus the masking of harmonic distortion is certainly higher than with a single tone as a masker. But as already said in post#490, it's about the tendency and not whether the masking of HD2 is 30% or 20% at 90Hz with 96dB sound pressure (where Amir measured the harmonic distortions).
Would also assume that the values above 90dB are pure estimations of Zwicker, because certainly no test participants could be found who had their ears destroyed with 120dB
Hope, however, it has become clear that the masking limits were not simply pulled out of a hat by "hobbyists".
The limits shown are certainly not valid for a single tone masker, so everyone has to decide for himself what is more realistic when stimulated with music (more single tone or more third band stimulation).