Did a few measurements today and noticed that the distortions have gone completely:
RAA: 48 Khz, 24 Bit, 3/4 out to 3/4 in, -5dB
Had a ton of noise above 1 kHz and some spikes at 50 Hz and 150 Hz before. After reconfiguring the whole cabling of the system the spectrum is much cleaner. And the strange thing: even without the ALLDAQ in place! The graph above is basically the same with and without the isolator, which tells us that the ALLDAQ works transparent without introducing additional noise.
As for the noise measured before, I guess it was just a faulty cable somewhere in the setup.
So what's the point with the ALLDAQ?
The isolator comes into play when I connect other components to the Babyface Pro such as my active speakers. Without the isolator in place I hear very loud buzzing noises when connecting through unbalanced TRS. Going with balanced XLR is mostly clean. But there still appear to be some random spikes going through the chain that always trigger both speakers even when no signal is played back. Those issues have completely vanished with the isolator between the interface and the computer. Ground loop solved.
Looking at my in ear microphones I cannot hear or measure any difference between both configurations. I guess it is just a flaw of the capsules being very prone to EMI.
As for the playback, I consider to send the isolator back and go for an external DAC which I can supply with an optical signal. Like I said, those galvanic isolators come with a price of 200 to 300 Euros. That is already the scale of a Topping D50s, a Soncoz LA-QXD1 or a SMSL SU-8. Those would not only bypass the the ground connection but also add a few more INs and OUTs.
I'm still waiting for the Intona to arrive. But I don't think it will change anything with the observations described above.
Regards
Ellef