The CO2 is just fine - there is no logic in these devices...bit perfect isn't applicable here.
This goes way back, but if memory serves (and only if one even cares about bit-perfect digital format conversion):
There were multiple anecdotal experiences with the CO2 + various combinations of other gear producing bit-
imperfect digital format conversions. To the best of my knowledge, no one ever confirmed the root cause: source device, cable, CO2, destination device. Additionally, I'm unaware of anyone testing the CO2 under
any conditions and producing a bit-perfect result. Doesn't mean it can't be done, I'm just not aware of anyone doing so.
So while there may not be any logic in these devices, that doesn't mean bit-perfectness isn't applicable within the broader context of the combination of gear used together. (After all, digital format converters aren't used in isolation...there's always other gear involved.)
For what it's worth, of this, I'm certain: I tested the Hosa ODL-312 (AES/EBU > optical S/PDIF) and ODL-276 (coax S/PDIF to optical S/PDIF) under at least a couple of different gear combination scenarios, and they proved bit-perfect each time. Whether those tests would hold up with different, or more current gear, I have no idea.
I'm not saying do or do not use the CO2 (or any other digital format converter, for that matter). But if bit-perfectness matters, it would be prudent to test the combination of gear to confirm whether it does, or does not, produce bit-perfect results.