Finally...
Setup: MacBook Pro into iFi micro iDSD, micro iDSD's line out into Gustard H20, micro iDSD's coax out into Audioquest RCA splitter, from there with identical cables into Yggy (coax) and DX7s (coax). From Yggy's balanced out and DX7s's balanced out with identical cables into Gustard H20's two balanced inputs.
First impressions: Yggy sounds sweeter, has more grip on the bass, cymbals sound nicer, more depth in soundstage. DX7s has something unpleasant about it so far, at least with the default apodizing filter and the brickwall filter. Switched to linear phase, slow roll off and actually liked it afterwards, but had to leave after a few songs (to be continued). At one point, Yggy was somehow less loud than the DX7s and the micro iDSD. Cycled inputs and it was fine again. Weird.
Blind A/B is more difficult here because the DX7s fades in after locking in on a new signal, like after changing the sample rate, thus revealing itself. Just like with the Yggy, the input gets changed by clicking a button (the volume knob in this case). But unlike with Yggy it's not obvious what the next input in sequence is. There's also an annoying delay until the display responds by telling you which input is now active. Yggy's problem is that the button doesn't register reliably. I do like that the DX7s clearly displays the sample rate via its display. The Yggy encodes it via LEDs for which the small labels are offset when looking from the side, making it hard to read correctly.
Feeding the DX7s with my old phone overnight, just in case it makes a difference. The Sound Blaster E5 is great for that, using its host mode it can charge my phone while serving as its DAC, and its optical output can feed the DX7s, albeit at a fixed 48 kHz sample rate despite using USB Audio Player Pro, which has to go through Android's sound stack in this configuration.