USB to SPDIF Convertors ideal?

[thumb5]

I've been wondering about that, too.  Off the top of my head, I can't see how the USB input could be improved upon for accuracy of data delivery, since it doesn't require the Devialet to recover the audio sample clock as does AES/EBU or S/PDIF.  Also there is much more robust error detection in USB.  But - speculating wildly - it may be that the USB input is more prone to inject noise into the Devialet ground plane?

I thought that there was no reclocking of SPDIF signal at the target, and that this was the reason for the 'weakness' of using SPDIF as an interconnect format (leading to the problem of poor jitter performance due to the SPDIF cable, either it's quality or excessive length, etc).

Yes, that's my understanding too.

ie; it works a bit like the way synchronous USB DAC's did before the arrival of asynchronous USB architecture and the input buffering/reclocking.

I see what you're saying.  I believe with S/PDIF, the samples are delivered at the audio sample rate which the DAC has to recover and lock to using a PLL (which is of course subject to stability and jitter problems).  But with USB, the time between individual samples depends entirely on the USB clock which is in principle independent of the audio sample rate (so in turn I can't see a mechanism for jitter at that level).  Also with USB there has to be some buffering because the samples never arrive in a nice, evenly-spaced way that matches the DAC's sample rate.  

As I understand it, with synchronous USB, the DAC then has to rate-adapt/re-clock the samples to match its own internal clock, which could perhaps present some problems because it has no control over the overall rate (measured over a long-ish period) at which the source sends the samples.  But with asynchronous USB the DAC provides feedback to the source which then has to deliver the samples at the right overall rate to match the DAC's internal sample rate clock over a long period.

if my assumption is correct, then the timing/jitter performance is derived from:
a) the USB-SPDIF converters own clock if SPDIF input used to Dev
b) the DEV USB reclocking if using the USB input on the Dev

In case (a) there's also the S/PDIF clock recovery circuit in the Devialet to consider, isn't there?  In case (b) with asynchronous USB there shouldn't need to be any re-clocking inside the Devialet.

in this case the perceived performance of the SPDIF solution could be expected to be better than the USB solution if the SPDIF had a great design. 
Or not ?

From a timing/jitter point of view I'd expect an asynchronous USB input to be superior in principle, but I may have misunderstood what's going on, and there are probably other factors than timing (as I alluded to) that could affect the perceived sound quality.

As always, discussing the technical aspects is fun/interesting but reality can be surprising when one actually listens!