A real difference in level suggests that the source (computer?) is not sending bitperfect audio and attenuating it a little. But if you mean it just sounds like a lower level, but measures the same, there are effects (like jitter) that can cause subjective perception of level changes. If all is designed well asynchronous USB should always beat TOSLINK optical. However, it is not always true all things are designed well.
The measures you mention are all useful in attenuation RF, but the USB ground is generally unfiltered, so is connected to the ground of a very noisy source: a computer chassis. So the only way to be sure can be galvanic isolation.
With some very expensive equipment, it is hard to prove, but it would be interesting to try subjectively.
Yes our ears are sensitive to jitter, more so than a lot of test equipment (including Audio Precision). However some kinds of jitter just cause THD, and so cannot be differentiated from THD, others cause noise. So the ear is crucial in the design and tuning of audio equipment, but cannot replace measurement in finding and curing problems.
I was actually looking into galvanic isolation to see if it will benefit audibly as I don't believe my usb output is noisy(only way to tell is to measure) enought to affect the DAC circuit. On the ODAC thread, a fellow has measured his laptop's usb port close to the monitor, and in this case, there were EMI from the monitor it was picking up since it was near by. It can be checked if there is an audible difference at the same time it can be checked electrically by taping into the usb port. Below is measurements of noisy supply, and with a linear supply the noise has reduced. Second is test with an isolator. Once again, this all depends on if this noise is affecting the analog output to be audible.
I'm at least half sure that a battery may be overkill but hey :P have fun! I would love to do battery power electronics as well. Here is ground pollution from my VGA monitor, not sure about how newer monitors would work. Rig this time is Monitor >Fujitsu Lifebook-> ODAC->ADC-> Adum4160 isolator-> Lifebook. Taking out the monitor gives me a clean spectrum.
Maybe the rigs you have been using are cleaner then most but I find that its not the case with some of us which may explain some's negative experiences with the ODAC.Just making a Y cable to inject extra power may not get you the results you really want as you are really concerned with common mode noise which though the ODAC does an okay job, it won't withstand ground loops. Well, whether you believe me ain't my problem so YMMV.
The really bad pollution you see in the previous graph comes from a small switch mode PSU for the Acer. Coupled with the fact that the usb port happens to be exceptionally close to the power input and bam, distortion. In desktops this may come from power built PSUs or you have some crazy graphics(or monitors) or some crappy fans or whatever polluting your usb.