Oooh OK!

I should say at this point that I'm quite happy to be wrong. Best way of learning. Perhaps there are better ways when others are watching, but it'll do.

The wave/particle duality of electrons is pushing it (in my limited experience of physics). You use the one that you're probing at the time. Maxwell's equations work the same for electrons as they do for deBroglie electron waves. You start firing them at a double slit, you'll see the wave. Count them coming from the back of a CCD from light quanta, and you'll see particles. I think I saw them described as travelling as waves, but arriving and leaving as particles, which seems pretty accurate.

Jitter is a problem, and I'm constantly amazed at how low some manufacturers claim to get it (I suspect serious spec manipulation). But in any well designed circuit you should design things to reject power supply noise. I do accept that perhaps the largest source of jitter for a VCXO is PSU noise to the oscillator, but that is far more likely to be caused by PCB layout and poor decoupling than a deficiency in the power supply.

I think someone else invoked the distributed model of a conductor (little aerials). This is why you have decoupling capacitors on the PCB. They are a low inductance current source to filter local power supply problems. The power supply itself is a long way away (electrically), and is largely irrelevant at high frequencies.

Sometimes in hi-fi, we have a tendency to invoke magic. The state of the art for production serial links is 10GHz. 96kHz audio is 2.8MHz. The same laws that keep those links working (and those Boeings and Airbusses in the sky) apply to the hifi stuff too.

Skin effect is real, but it actually helps you from a noise perspective. It raises the effective resistance for high frequencies because they only travel on the skin of the conductor. It's only a problem when you're trying to send really high frequencies across solid core wires.

Jitter, DAC resolution and sample rate all apply to mobile telephone base stations. But here they tend to use 16 bit, about 50MHz sampling and oven controlled oscillators to reduce jitter. Any noise means poor efficiency, which costs a lot of money (see 3G licences!). They don't use oxygen free copper or green permanent markers. (or DACs with a real resolution way below their quoted number of bits for that matter). Just sound engineering practice, and it works.

I've also seen some pretty good data on human hearing. It's a lot better understood by psychoacousticians than people give them credit for. Did you know the shape or your ear isn't there to collect sound? It's there to alter the sound coming from different directions in a complex, but repeatable way. This is the head-related-transfer function. And early research in virtual reality led to those surround headphones you can get now. That's pretty subtle stuff.

Once again, it's nice to have these discussions. No matter what anyone's side, at least they care about their music enough to care!

Regards,

Mark.

861v8, DSP5200, DSP5200HC, DSP5000mk2, UHD722, MF10, Prime, Prime PSU