[nicolasb]

I expect DACs sample from the stable part of the pulse (near the center), on a fixed clock. It's hard to imagine they'd take timing from the leading or trailing edge, which is the noisest part of the pulse...!

Some DACs work like this. Most do not. In the majority of cases the timing of the signal is determined by the timing of the input pulses and not by a clock within the DAC itself. This is precisely why, with many DACs, there's an audible difference in sound quality when you use different transports.

Any electrical engineers gurus want to take a stab at this? If edge effects are a measurable factor in digital transmission (and they don't seem to cause ethernet any problem, so I'm skeptical), audiophiles should be screaming for purely optical interconnects

Ethernet is not a valid analogy because it's asynchronous. With ethernet, all that matters is that the actual data makes it through. It doesn't matter how long it takes; the packets can even arrive in the wrong order; all that matters is the data that is eventually reassembled.

But with a PCM stream the timing of the signal is critical: it doesn't just matter that you get all the right ones and zeroes, they have to arrive at exactly the right moment time, and a time inaccuracy measured in picoseconds can introduce audible distortion.

Optical connections don't actually help with this, because there has to be a conversion from electrical pulses to optical pulses in order to send the signal along the cable, and then a conversion from optical back to electrical again at the far end. That double conversion process can, itself, introduce jitter. So, generally speaking, coax sounds better than optical. (And in some cases there's actually a limit on the bandwidth too: my Arcam DV27 player, for example, can output a 96kHz signal over coax but only 48kHz over optical. Clearly this isn't a limit of the cable, it's a limit of the conversion circuitry.)

Another thing audiophiles tend to agree about, btw, is that there isn't an audible difference between different optical interconnects.

[ScottMayo]

Ethernet is not a valid analogy because it's asynchronous. With ethernet, all that matters is that the actual data makes it through. It doesn't matter how long it takes; the packets can even arrive in the wrong order; all that matters is the data that is eventually reassembled.

You're thinking of TCP, a software protocol. TCP runs over ethernet (or a bunch of other physical layer choices), but the physical layer has its own disciplines, and in them, timing does matter. Ethernet in particular has a bunch of different wire protocols in use, depending on the speed; if I remember, 10mb Ethernet uses Manchester encoding, and 100mb uses even weirder stuff.

Anyway, I admit to having a preference for optical cables. I get mine at Radio Shack, because as you say, optical cable is optical cable.

[ctviggen]

But PLLs have many problems.  As a patent attorney, I've written quite a few patents about PLLs.  PLLs are way more complex than people think.  Each design has benefits and detriments.  Some designs allow jitter to pass through and others don't.  Some designs want to quickly track changes in frequency/period and others don't.  Also, many transmission schemes use error correcting codes.  It appears that SPDIF does not.  When you're talking sound, though, even with error correcting codes, the data has to be timed correctly.  If you're talking data sent between computers, no one really cares if the data is a few picoseconds or even milliseconds off of correct timing, as long as the bits are correct.

[BeeBop]

Well back to my original topic. Picked up 3 pairs of 1m DH Labs Air Matrix cables on Audiogon - very good deal at $100 a pair - the guy who sold them to me had 5 pairs he had bought for a HT application that never materialized. Do they ever sound sweet! No doubt there are better to be had out there but I am very happy with these. Really help my BP25/3BSST do their very dynamic, transparent thing. Cables are between the amp/pre-amp, tuner/pre-amp and DAC/pre-amp. Also using 6 foot runs of DH Labs Q10 speaker cables (also $100 on a'gon). Very nice sound all around IMHO. I was once very skeptical of cables but now realize that bad cables act like a choke on sound quality. Good cables let you find out how good your equipment really is. Thanks to all who added their comments. Special thanks to WEEZ who pointed me in the right direction re the speaker cables.