So after 12 hours and change of listening/blind* comparison, I’ve a few thoughts on the updated RME ADI-2 DAC fs:
Normal Use
In normal “for pleasure” listening, with a lossless source, feeding a variety of headphones (dynamic, planar, low and high impedance), and with both RME units configured and fed identically, there’s no audible difference between the two.
Critical Comparison
Using test tracks, and critical/audition/review style listening, and some deliberate problem-causing scenarios, its possible to audibly discern some extremely small differences - and one not so small one (but that won’t come up for most users).
Headphone Output
The original version could exhibit some minor dynamic compression (minimized somewhat in high-power mode). This was most noticeable with dynamic headphones that already had superlative micro-dynamic performance (e.g. the Focal line). It was never a factor with the IEM output. And it was not something I heard with any planar headphone shy of the LCD-4 and AB-1266 Phi CC (which needs more power anyway).
In the updated unit, this is less apparent. I won’t say it’s gone completely, but it wasn’t audible until I was using headphones like the Stellia, Utopia, Vérité and MySphere, and even then it took focus to hear it and the effect was reduced.
Chances are this means you won’t need an additional amplifier with anything less than flagship cans with class-leading dynamics for any reason other than getting balanced drive or more power.
Digital Filters
The updated unit exposes a new digital filter in the 4493 IC that is not present on the 4490, specifically the “SD LD” filter or “Short Delay, Low Dispersion”.
This yields lower latency, shorter/lower-amplitude pre-ringing and reduced phase issues than the more common minimum-phase filters, but still has phase shift vs a true linear-phase filter. So far I like it better than the default “SD Sharp” filter, and would say it’s the second “best” filter on the unit.
My preference remains the linear-phase “Sharp” filter, however.
It yields the most natural, accurate, reproduction of material (using my own recordings as a baseline) as far as I can discern.
Headroom
RME appears to have increased the available headroom for full-range digital signals. This is most apparent when using the unit at 0 dB with material that exacerbates issues with inter-sample overs. The result is such material now sounds just as smooth and refined as non-problematic sources did before.
*I will detail the special box I use to facilitate hardware-assisted blind tests for sources, DACs and amps in another post sometime. But it employs level matching to within 0.1 dB, hardware entropy source/TRNG-driven undetectable switching and capture and logging of elections. It has a ton of other features as well, but I’ll cover those at the same time.