And with an analog cable to each other. Since the digital cables are not connected to exactly the same point of the computer, there will be a small potential difference between the ground of the interfaces, resulting in common mode noise on the analog cable.
The resistances might be small, but they are not zero. Also, the ground currents from some devices in a desktop PC are huge, and the amount of noise that was measured is equivalent to only a small voltage (-92 dBA relative to 2 Vrms is only 50 uV A-weighted noise voltage).
A differential amplifier does not eliminate power supply ripples. Nor does the power supply theory explain why the noise is there only if one sound card is recording the output of another, without using a differential amplifier.
Also, if I connect the line output of both sound cards to two inputs of an external (not grounded) amplifier, and I turn up the volume on the amplifier, I can hear the computer noise. But unplugging the other sound card (not the one that is selected as input) makes the noise go away completely.
The only plausible explanation to all of the above is a ground potential difference between the sound cards.
Edited by stv014 - 2/26/13 at 5:43am