Originally Posted by drez
I thought ground loop was caused by a potential difference between ground points? EG. connecting a computer at one wall outlet to an amplifier at another wall outlet? Have you had this sort of ground loop happen to you before?
Since the ground paths in the PC have non-zero impedance, and significant currents are flowing from the CPU, GPU, and other devices, there is indeed a voltage difference between different "ground" points of the same machine. If you do not believe, try doing a loopback test of a sound card with its own A/D, and then that of another card in the same PC. If the DAC and ADC are not on the same card, the noise levels are usually much worse. You can see that effect on the following graphs. Notice the large amounts of digital "crud" picked up in the D1->STX and STX->D1 tests:
Xonar D1 D/A to Xonar D1 vs. STX A/D:
Xonar STX D/A to Xonar D1 vs. STX A/D:
With sufficiently low load impedances and/or sensitive testing, effects on the distortion can also be shown. Ignoring/not understanding this problem is a common mistake in RMAA/sound card based measurements.
It is likely that in the USB DAC to sound card test linked above the amount of noise voltage that ends up in the audio signal depends on the impedance of the USB cable relative to that of the audio loopback cable, explaining the small variation in the results. Therefore, the test results are not relevant to typical real usage when there is no loopback cable.