Some interesting discussion here around the right THD+N spec. We conducted an empirical test a number of years ago where we had the test subject sit in the anechoic chamber and listen to test tones (yikes). The operator in the lab would slowly introduce a second tone. We had the subject raise their hand whenever they could detect any change to the test tone. A camera in the anechoic chamber allowed the operator to observe at what distortion level the hand was raised. We did this with quite a few people and the results were very consistent. We introduced tones that were both harmonically related to the fundamental and ones that were not. For the harmonically related tones the distortion level for detection was incredibly high for the 2nd harmonic and dropped off rapidly so by the time you were at the 5th harmonic there was no difference between a harmonically related tone and a non-harmonically related tone. There was also a fairly linear downward detection level with frequency. We are much more sensitive at higher frequencies. The absolute lowest detection level we got was 0.1% for a non-harmonically related tone at 10kHz, and this is with test tones in an anechoic chamber so music masking in a real room would be a much higher level. So here is the problem; the spec does not give you any information about which harmonic. If it is the 2nd or 3rd then a THD+N well above 1% would actually be ok but outside of this caveat the most extreme argument you can make is that it needs to be below 0.1% (how much below is irrelevant). All this said there is another real-world problem that can make even considering the THD+N spec completely irrelevant, and that is if you hit the dynamic limits of your amplifier. Amp clipping is measured in double digits of distortion that will go through the roof with the addition of only a few watts above the clipping threshold. So, the moral of the story is to be sure you have lots of power to spare before worrying about the THD+N spec.

Ian Colquhoun
President & Chief Engineer