I think the conclusions that all "tier one" chip DACs sound almost the same are pretty close. I do hear a tiny difference it the very top end "air" but it seems to be at the expense of vocal smoothness. Filter 3 smoother, 4 sharper . Overall, I prefer my old JDS Atom+ for just enjoying music. It could be the differences are above the 17K or so range where my hearing is falling off. Or the "massive differences" from the You-Tube shills are all made up. Is it a shame that good hearing is wasted on the young who can't afford top tier equipment?
The front knob is USELESS. One could glue a knob on it. But a desktop device you have to use the remote for is just plain stupid. On the plus side, Remote range is far better than Topping, it has XLR outputs, and does not really do anything wrong, so in a vacuum, most would be quite happy with it. I think they did spend the money on the power and output, but it does not seem to be the determining factor.
So the SMSL is going back. Keeping the DX3pro+ in the main stereo for now and the JDS Atom+ stack on the desk. I was considering a Denafrips/Ladder/Musician entry level R2R, but the horror stories about defective units or returning because they did not like them is preventing me. Not risking that kind of bucks for no support. Jumping up to mid tier is even riskier! There may be great positive stories but I did not find any. As we know, do it right and you tell a couple people, do it wrong and you tell the world. That leaves the Bifrost or Mojo as the only two sub 1K DACs that may sound different with minimal risk. I think I'll wait a while.
I was thinking about the above article I referenced. Do consider the test signal that stepped from 0 to max in one bit is an artificial situation that would never happen unless a corrupted file and it exceeds the rise time of a sine wave within the 20K spectrum. It implies a single sample wide pulse. So his comments on looking at the ripple seems reasonable. Not sure I agree with his filter selection, but of course that is a "hearing" choice. Looking at the response of the Schiit and JDS, it looks like both are close to "linear fast" but as these are older chips, the filters are custom and not the hardcoded ESS numbered ones. I also noticed the documentation in Topping and SMSL do not match the numbers as published in the ESS data sheet. They scramble the numbers a little. You can figure it out looking at their published documentation.
This is a complicated subject and I would like to track down more just for understanding. Bits are bits, but what you do with them matters. Someone asked Mike Moffit which filter he used in the Schiit DACs. His answer was "The right one". I suspect the filter is a large pert of why John Seaber's DACs are so good using older cheap parts. It's the execution, not the parts.