A:
@Josh83 could also distinguish between other DACs. If ONLY the D90 would be distinguishable it could make sense to investigate the polarity as a possible reason but that wasn't the case.
B: output buffer circuits are usually all 'signal followers' 1x.
C: even if the buffer would be inverting (-1x) which is highly unlikely then polarity would still be correct if the - signal were input.
The ABX box is very likely not to blame here despite not being blameless.
The distortion the box does add would be equal for both DACs
if... at least both A and B input were attenuated and the signal thus passes though the attenuator.
A question could be whether or not the volume setting change settling-time during switching is always faster than the 'mute' during switching (the always present 'tick') but reckon the settling speed is not an issue as relays aren't fast switching anyway (mechanical component).
If the 'softest' DAC is not attenuated AND the louder one was attenuated then it is possible (not proven until tested) that the 'loudest one' would have had a bit more distortion added.
Seeing the null results in the follow-up and looking at the linear scale between 1Hz and 20kHz we see around -100dB difference and that could mean 0.001% distortion which is inaudible.
BUT this null is between A and B and actual distortion (because it is the same and thus won't show in a null as it is common) could well be much higher.
Alas it seems we need to see if Erin (
@hardisj) at some point does a distortion measurement and verifies what the attenuator does.
It
would be interesting to know how much distortion the attenuator adds and what the spectrum of that distortion is.
This would give some info about Erin's box (which is the same type).
It is not likely in Josh's advantage to post recordings nor measure distortion.
The relays, ground loops, etc are most likely not the issues.
As annoying as it is it seems Josh was able to ace the test by looking at minute audible differences between the different DACs in specific parts of a song.
It was not about an obvious 'sound signature difference' that the better hearing people are always on about.
We know not all DACs measure exactly the same (if only because of the filters) but let's face it.
Josh used 96kHz files which means any reconstruction filter issues would have been around 45-50kHz so would not be an issue.
Likewise for the up-sampled 44.1 file as that would have a around 20kHz difference that would be equal in both DACs and would treat that the same at it would be well withing the audible band.
Until someone repeats it or the test by Josh was redone and checked by a witness as well as witnessed all that can be done is either dismiss it because not all is known or see this as one of the rare 'positives' that took many, many attempts and training over months to detect.
Personally I would like to have seen a THD measurement, attenuated and not attenuated (which could potentially differ a bit) and a recording with the 2 most differing DACs while switching between DACs.
Not going to happen I'm afraid (not in Josh's interest nor intention) so best to accept the results for what they are.
No one will argue that some amps can not sound different or that speakers/headphones/microphones/instruments/recording(masters) and vinyl cartridges as well as tape can not sound different enough to be audible.
Just like that it is equally understandable that reconstruction filters (around 22.05kHz) can be different enough for young folks to be audibly different and that there are DACs that are purpose made to 'sound' different.
This makes this whole discussion test interesting but looks like we won't get to the bottom of the how and why... alas.
