"Audiophile Grade" Ethernet Switches - The new generation

[thumb5]

...  If it's all that bad one has to ask why can't the effects be easily measured?

We have measured them. At the clock input pin of several DACs. Just not ready to post the graphs.

Alex, first let me say that I found the white paper interesting and it explains the principles behind the design of the EtherRegen quite clearly.  As someone who works daily with digital hardware the ideas were not entirely new to me and I can see in theory how ground plane noise could affect the output of a DAC.

That said, I remain sceptical about the extent to which that theoretical effect actually translates into reality, if one has a reasonably well-engineered DAC.  To be clear, I wonder whether the effect can actually be at a level that makes any audible difference (relative to other noise and distortion) at the output of the DAC, which after all is the only thing that really matters.

The white paper states that clock phase noise will cause sidebands around a pure tone played through the DAC, which seems plausible depending on the design of the DAC.  Presumably during the design of the ER and before it was launched to the market you've been able to measure the level of those jitter-related sidebands with some real DAC, or possibly several DACs, connected first to a standard commercial grade switch and then to ER to give "before" and "after" results.

That sounds like a pretty straightforward measurement given the right equipment, which I guess you must have if you're designing and building this kind of device.  So, would it not be possible to simply tell us the actual magnitude of this problem in the real world, and how much it's reduced by ER?  You'd need to specify the test protocol, what DAC you were using, etc. but naturally that's information you'd already have to hand I suppose.

Would it also be true to say that if the problem is significant in the real world it should be possible for it to be measured independently to verify your own results and consequently the effectiveness of the ER?

Unless I've misunderstood something, in which case please say, it seems to me that the tests carried out at ASR (leaving aside any subjective comments and subsequent discussion) attempt to measure exactly these jitter sidebands that your white paper predicts.  If you agree, could you comment on why their measurements didn't show the presence of these jitter sidebands?  Otherwise, what do you think was wrong with their measurement approach that caused the effect of the ER to be hidden?

Thanks.