Aleg is completely right, and if one searches the forum this has been discussed and explained many times before.
A USB, Ethernet and SPDIF are electrical cable(s). They carry analogue signals. These signals represent digital values at a period of time.
As these cables are like any electrical cable they are affected by analogue signal properties such as being bandwidth limited, acting as a transmission line, electro mechanical emmisions etc.
These properties interact with the connected equipment.
This in turn produces an analoge response in the connected equipment. This is quite seperate to any digital value that was represented by the analogue voltage. This for the most part stays intact; which is of course a benefit of digital encoding.
However if the connected equipment have functions that are affected by these analogue interactions such as digital clocks, and level detection and level transition detection, then these analogue interactions often show up as some sort of modulation of the digital data timing.
Now in some systems this is of no consequence. In digital to analogue conversion systems it can become audible.
Now USB by its nature can be noisy. It uses a twisted pair as its primary data carrier, which is deliberately unbalanced by signalling protocol. In addition a USB lead carries a DC supply line along side the twisted pair which can have RF currents induced on it, even though they are usually reasonably well shielded. Finally i seem to remember the USB termination spec is reasonably loose to keep costs down, so there can be a variation in reflectivity.
With all this 'science' going on it is hardly surprising that different USB cables in the real world 'sound' different when connected to audio equipment.
I think with USB there is a strong argument for Naim producing a reference cable because of the large amount of variation possible.
And none of this involves sales conceptual illustrations of being 'bit perfect' or 'ones are ones and zeros are zeros', which really is not that relevant in this area.
Simon
