I ran 9 hours of 24/96 testing on my laptop last night (under Linux) with USB and a UA-5. There were no USB transport errors reported and that is encouraging. The source was analog, so I am unable to do any extensive correctness testing in this case.
Ultimately, I want to do this with the Minime, not the UA-5. And because the Minime lacks digi-in, it does not seem possible to do digital comparison based correctness testing.
I have considered writing a software tool to scan the samples and look for excessive sample to sample amplitude changes that would suggest noise. While that wouldn't catch everything I think it would cover some important cases. A tool that could do a 'fuzzy' compare of the original digital source and the final digi->analog->digi result would be ideal.
Any suggestions on how to validate the Minime at 24/96 with USB for Many hours (ideally automated)?