So, I went back and looked at the data...

Though I do not have them all handy, I have them...but for the STEREO and UPSAMPLE modes, I have the values. Here they are:

UPSAMPLE Mode:

Delay: 9.3 ms
Comb Spacing: 107 Hz

STEREO Mode:

Delay 8.0 ms
Comb Spacing: 125 Hz

Again, I am not simply saying that there is a 9.3, an 8.0 or whetever latency - I would be OK with that. What I am saying is that a delay and add scenario is taking place, which is turning the analog signals (at least in the main speakers) into a comb-filtered version of what is coming into the analog inputs; again, zone 2 is not affected by this issue.

How do I know the delay times? I looked at the spectra, and from them, I derived the delay time that is consistent with the peaks in the spectra. That is, in the UPSAMPLE mode, the spectra exhibits regular peaks (about 10-12 dB delta...not insignificant) every 107 Hz and in the STEREO mode, these peaks occur every 125 Hz.

For the math minded you will note that the reciprocal of the delay time indicated you get the spacing of the peaks. That is, take the reciproc al of 0.008 (which is 8 ms) and you get a frequency of 125 Hz - and this is indeed (pretty much exactly) the spacing of the peaks in the spectrum for the stereo mode. For the upsample mode, the peaks are a bit more tightly sapce due to the longer delay time of 9.3 ms (that is 0.0093); take its reciprocal and you get 107 Hz, and this is indeed where the peaks are spaced.

So please, would someone out there play a white noise file into the analog inputs of their 990 and see if you notice the comb-filtering? Again, you will notice this as you change modes (STEREO, BYPASS, UPSAMPLE etc).

Thanks,

Mark