I think perhaps some of the differences may not necessarily lie in how clean and beautiful 1's and 0's look in the scope, but perhaps how different non-ideal issues couple into the analog circuitry among other things.

Jitter at the receiving end of a DAC may not be a big deal in terms of properly decoding values. But residual jitter having an impact on the DAC output clock might be a different story. Noise from a noisy power source coupling into the traces of the analog section of the board is also perhaps a bit problematic and not necessarily negligible.

As I read about DACs and ADCs I find that even datasheets for DACs recommend special considerations for digital and analog sections of a board's layout. I think I have found this fairly often in some multi-channel DACs for receivers. Consider Fig 43 of the PCM1690 datasheet:

http://www.ti.com/lit/ds/symlink/pcm1690.pdf

Considerable amount of jitter in the signal source might be too much for certain DAC implementations. How much jitter and noise there is in the line proly depends on the connection type, source, cable construction and so forth.

It also depends on how well things are handled at the DAC itself.

To counter some of these problems I think PCB layout, i.e. implementation, matters a lot. In some cases, the traces themselves are used to filter signal stuffs. For the PCM1681, on page 33 there are even recommendations to use the PCB trace parasitic capacitance "to form a low-pass filter that removes high-frequency noise from the digital signal, thus reducing high-frequency emission":

http://www.ti.com/lit/ds/symlink/pcm1681-q1.pdf

Again, I don't necessarily think that 1's and 0's and how pretty they look matters as much as how their ugliness affects other parts of the system.

Dunno TBH, but best guess is that there are some fairly large buffers there.

For me the significant part is that until Wyrd and a super-zoopy usb cable, the cat7 data path was clearly superior.  I figured it was due to the extra buffering/reclocking etc. in the bridge card that gave it the 'edge'.  And the difference wasn't slight either.  Inner detail, bass impact, sound stage etc. etc. all clearly favored the cat7 data path.

But now the oft maligned usb data path is the equal and in some sonic characteristics superior to the more sophisticated cat7 data path, and for the first time in my system.

All because #1, the Wyrd reduced the noise on the power being sent to the DAC.  AND #2, because the data stream has been re-clocked.  Then #3, this data stream is passed along using wire that has been silver 'cladded' and has other 'tricks' in its construction.  AND #4, was actively cooked.

And the results are not just noticeable but delineate-able as to FR and other easily describeable sonic attributes.
And I do agree that 'noise' from whatever source that is deposited smack into the input circuitry can 'influence' the rest of the functional operation.  But the effects are not what I would call 'errors' which is what the digital domain would have us believe is the 'only' option.  Either the correct data is present, or not.

And I do understand that digital circuits are in reality re-tasked and refined analog circuits which in these cases are designed to act as near to an on-off switch as possible.

Except that the sonic differences sound more akin to 'analog coloration' especially with those aspects I mentioned above among others.  Thus my referece to tube rolling.  Which is a way of 'customizing' the sound to your own tastes.  These efforts can yield impressive results to say the least.

All due to improvements in the delivery system of the data bit stream that is subsequently re-clocked, again, and 'properly' presented to the dac for re-construction into an analog signal.
 
IOW the sonic effects of signal degridation from whatever source (analog or digital) can have the same or very similar sonic effects.

I find this fascinating myself.


JJ   

And yes, I still have problems with inconsistent ODAC performance. I jumped in the deep end without being able to swim. Reading the subsequent posts with interest.

I still wonder, though: we know that USB can transmit data accurately, and, in the digital domain, out of my depth or not, I do still insist that music is data. Are we to believe that the Audio implementation somehow got broken?

Actually, the ODAC is my first USB DAC: I started with optical, a decade or so ago, and then went to firewire.

Next time, I will give my self a choice of input options.  Although I am easily able to believe that the in-DAC circuitry may not treat all inputs equally, at least I will get the experience. And choose the one I like best. Placebo or not!

USB isn't electically isolated and in some cases this MAY affect the analog signal or even 'decision making' errors (timing related) in the recieving circuit.

It can be with the Adnaco optical card. Part of the issue with USB is that there's so many different ways to implement it as compared to more traditional methods. USB can be either terrible or fantastic, it just depends on design.

This is obvious when you look at all of the USB > S/Pdif converters on the market which vary hugely both in price and performance. The Hiface and OR5 for example share the same basic technology. One is arguably the worst performing converter on the market, the other is one of the absolute best. Neither ever enters the analog domain though, so clearly there's more to USB implementation than say the effect of a noisy ground on a DAC's analog output stage.