I just wanted to come in here and say that the 16/48 S/PDIF input is indeed bit-accurate.
This is based on an set 2 of the show that I taped last night. V3 > D8 via AES1 output, V3 > Microtrack via the coax output.
I haven't yet transferred set 1, and I'll test that one as well. but set 2 was an hour long. Microtrack file copied with a USB 2 card reader.
the DAT was transferred via D8 > EgoSys Waveterminal 2496 > Cool Edit 2000.
I then trimmed both files to start and end at exactly the same place. Then I used mkwACT to generate md5 files for both.
89c23072bd5dd4c2d86d6abb13cf2847 *set2-DAT.wav
89c23072bd5dd4c2d86d6abb13cf2847 *set2-MT.wav
So amid all the bugs and quirks of the initial firmware, this gives me a good deal of hope. I know I'm not the first one to do bit accuracy tests, but I don't think John's tests were over the same length of time. So basically, in my mind, this proves that not only does it not resample the digital input, but it also doesn't have any periodic or random blips over a long period of time.
after I transfer the 2 hour long first set, I'll do the same thing and post here with results.
Edit: typos