This is the extracted and amplified distortion signal using a music signal and a speaker load, a Mordaunt-Short MS20, the lower trace being the music, played a little above my normal listening level. The oscilloscope is in the alternating trace mode, so there is a time delay between the traces, but this and other photos showed similar low outputs. It can be seen that the result looks like noise, and if amplified and recorded, then played back alone and listened to, it appears to be both noise and uncancelled music. With a signal to noise ratio of maybe 6dB it is difficult to judge distortion, but the remaining music component sounds fairly undistorted. In normal operation the noise of the amplifier is inaudible even with an ear next to the speaker, so the distortion, at a lower level, should be well below audibility at the normal listening distance. Add back the original music playing at peak level up to 90dB and there seems no possibility of the distortion then being audible.
Some speakers have lower sensitivity and more difficult impedance characteristics than the Mordaunt-Short MS20 used here, so this is certainly not a worst case result, simply an indication of the performance at moderate levels with a typical load.
The many claims to hear significant differences between even good quality amplifiers are difficult to take seriously when the real errors are known to be at such a low level that they are inaudible even without the masking effect of the music. The same sort of inaudible 'distortion' signals were being demonstrated for the Quad 303 amplifiers more than 30 years ago, so I have not done anything unusual other than achieving such results with a relatively simple design.HOME.