@
Olli@, I do not understand your logic (only one channel -> must be software). It's the other way around: the audio hardware - however small it is nowadays - has physical dimensions. The parts that make up tne one channel might be nearer to some source of inference than the other. It can even shield the other channel from inferences. Besides that, it's well known that one shall not run the PSU @ it's limits for a longer period.
Well, then please explain to me exactly how the PSU can cause a regular buzz on just one channel that appears both on an analog 3.5mm jack
and on the digital signal of an USB port (and it obviously affects USB audio only, but not USB keyboard, mouse or other protocols).
For example: Does the PSU have anything to do with USB? Does it “know” how to dissect the various layers of the USB audio protocol, and replace just the sample bytes of the right channel with bytes that represent a buzzing sound, and recalculate the CRC checksums of the USB packets? I think this is extremely unlikely. Unless your PSU comes from the NSA, of course.
Note that there are
no analog parts involved when using USB audio, the path is completely digital, so the PSU would have to inject digital „buzzing bytes“ at some point into this path, where ever that may be.
Also note that the data paths for the 3.5mm jack and for USB audio are completely different. The former comes from the audio hardware chip on the mainboard, goes through a DAC (which is often – but not always – inside the audio chip), and then through a small amp (which is usually separate). Conversely, USB audio does
not come from the audio hardware on the mainboard. It comes from the USB controller (which can be inside the processor, inside the chipset {sometimes called north bridge or south bridge}, or it can be a separate chip on the mainboard), goes through one or more USB hubs, and finally reaches an USB port.
So what do analog audio and USB audio have in common? Almost nothing. It’s even unlikely that they’re on the same PCI bus (although it is possible). The last thing that they have in common is the
snd(4) framework inside the kernel – After that, they go their separate ways.
Therefore I’m positive that it must be a software problem.