Frequency measurements with different inputs

[krass]

interesting, below 800hz they seem aligned, above 800 there's a frequency shift between troughs. Note how the offset looks about the same for each trough, and because it's a log scale that means the same ratio (and not the same delta). Its about 25% for each trough (925 to 1200) (2800 to 3300) (etc).

If it was aliasing then I believe the offsets would be constant (function of (f1+f2) or (f1-f2)), so on a log scale the gaps would appear to get smaller as frequency increased.

Note also that the troughs appear to realign at about 10k. However I think this is misleading, and that above 10k the same troughs are still offset from each other by 25% but appear superimposed on the adjacent trough......

What's also interesting is that the offsets only affect troughs. The peaks seem to remain aligned throughout.
Why is everything aligned below 10k ?

Troughs = room node cancellations, but these mostly appear at low frequencies. What's going on above 10k I cannot guess. Could it be Nyquist related ? what source are you using ? Is it redbook 44.1k ? A 20k source can only resolve above half he source frequency, ie 10k.......

I can't help much more, but my guess is that there is something wrong with the test setup, but can't really guess what........., so I just leave my observations here in case somebody more knowledgeable about the protocols can see a pattern....