.. or (more likely perhaps) in the output stage. But if you connect a DAC to a massive source of noise at one end (a pc) and don't do much about it, it's not surprising that some of that noise can get out the other end. But spending money on quieting the transport is arguably throwing good money after bad; why not spend money on a better designed DAC that is better isolated from the electrical noise of the pc? For example, Naims new V1 optically isolates the digital and analogue sections. Other DACs make claims about galvanic isolation. Good to see these things being measured rather than just speculated about.
From the point of view of the analogue output stage of a DAC, a digital signal
is noise, on a massive scale. It's an absolute roar. Think 33k modem signal. So this idea that poncing around with jitter rates is somehow going to immunise a DAC which leaks noise is a bit silly.
My TOSLINK cable cost about 2 quid. If I had known I was implementing a state of the art "galvanic isolation solution" for leaky DACs, I would have put it in a fancy box and flogged it for three hundred quid.
A DAC which lets miniscule timing variations from the input data through into the analogue signal is a shit DAC and needs a much better buffer. The thread and measurements linked to said nothing of that sort was happening. The measurements suggested that the proximity of a computer to unbalanced output cables was transmitting noise into them. Efforts to differentiate sources (e.g. ethernet vs wifi vs properly shielded output when the DAC was driven by USB) in any significant way were a spectacular failure. The linked experiments do not back up the idea that it's worth spending money on a transport. They do support the idea that it's worth using XLR outputs from a DAC if it's near a noisy computer.
Note the first test manages to establish practically no difference between XLR output measurements when the PC is idling and when it is running full tilt. According to item's broad listening experience, this fellow's testing equipment must be broken broken. Here's the point: "In theory I would have expected that a busy machine worsens "jitter" through the USB interface with all the audiophile talk about minimizing # threads and CPU load... At least no evidence from my tests to suggest this makes any difference SO LONG AS YOU MINIMIZE ELECTRICAL NOISE from getting into the system."
Note also that even the difference between XLR and RCA output is probably near or below the inaudible threshold: "However, the RCA 'quiet i7' (white) has a number of spikes evident including a -115dB hump at the 60Hz powerline frequency (of course this is inaudible at normal volumes)."
This is his conclusion: "If you're listening with the computer, computer LOAD could make a difference but if your equipment has good noise rejection (eg. use of balanced cables), this does not seem to be an issue... If computer load correlates to USB jitter, I'm not seeing it showing up in these measurements either because the asynchronous DAC is doing a good job rejecting it or the increase in jitter is not enough to overpower the existing jitter in my measuring equipment."
How anyone can turn that into a win for foo transports I have no idea.