On contrary, my view was more or less supporting you, I tried to separate noise and distortion. Maybe your more technically oriented colleague will reply. I am not theorizing and I have stayed firmly in technical terms.
To continue:
1) have you made any study that would support the assumption that ultrasonic noise does not degrade tweeter linearity at acoustical side,
2) will you post full swing CCIF IMD 19+20kHz to verify good linearity at high frequencies.
I have done the 19-20kHz test, but only at 5W average power. We can see thet the in band distortion is at maximum -115dB at 1kHz. This is at (or close to) the limits of my analyzer. No high frequency noise.
Will do a full power test later.
I can do a THD measurtement also, but it will be the same as THD+N if the bandwidth is less than 40kHz. At 80kHz the noise shaping might come into consideration. I can check it for the fun of it.
Regarding the tweeter issue we have not tested if it heats up. But when the amplitude of the remaints is only 0,3V at 500kHz that equals only 0,01W not taking into account the inductance of the tweeter itself.
The amplitude is also symmetrical around zero so any offset is not a problem either.
Found a measurement of THD+N with bandwitdth up to 40kHz. The test tone is 15kHz.
370W continous power both channels at 15kHz with total of -108dB THD
including N.
In other words, we have 41,4 volts RMS of clean 15kHz signal, and the
total sum of noise and THD up to 40kHz is 108dB below the signal.
Not too bad?
NOTE! The "jumps" in the measurement is my mesasuring gear changing range. Better measuring gear would measure this down to at least -112dB