Air 3.0.4 Officially Released

[Confused]

Well executed digital coax in home applications, over a meter or three, won't be worse than AES/EBU (XLR).  It can be even better. Reason for existence of differential signaling is resistance to noise over long distance. There should be no difference between the 2 interfaces until very long cables are involved.
Analog signalling is different animal, differential transmission makes sense even over short distances.

That's theory. In practice, quite often, coax outputs are implemented horribly. That is why so often we hear, that AES/EBU is better. If you hear AES better than coax, then it means coax is poorly implemented in your source.

I suspect most people would agree that any difference between AES/EBU & coax S/PDIF is at most very small.  However, my experience with a Devialet is that the USB input does sound different.  I suspect this may be due to galvanic isolation issues or 'leakage loops' with the USB implementation.  Although I note that there is a letter and response in this month's (April) issue of 'Hifi News and record review', here HFN's Paul Miller speculates the difference is due to the USB input using XMOS firmware and asynchronous driver, whereas the both the S/PIDF and AES/EBU use Wolfson WM8805 digital interface transceiver.  Irrespective of the technical stuff, my experience is that I get more accurate bass via AIR, S/PDIF & AES/EBU in comparison with anything via USB.  The differences are fairly minor, but they are discernible.

This also appears to concur with Hifi_swlon's bass observations now he's switched from the microRendu back to AIR