I have been using a Topping D90 (MQA version) for about a week now, and I still need more time before I can make a final decision about this DAC and its value. It certainly ticks many boxes in terms of functionality/features. But I have to hear it through many different familiar sources before I can really judge it appropriately.
So far I'm quite impressed by the implementation of the AKM flagship chip at this price point (hopefully they get the factory going again soon). Topping seems to have kept costs as low as possible while still delivering a solid piece of electronics (time will ultimately tell just how robust it is). There are trade-offs, but so far it looks as if those trade-offs are reasonable. Tear-downs of the D90 show solid components in the places where they need to be.
My ears are sensitive to very high frequencies, and for me various DACs really only differ in the highs. As a physicist, this makes sense, reconstructing low frequencies from a much higher frequency digital signal should be much easier than massaging the highs around Nyquist. When people talk about the "tight bass" or "upfront mids" and such for a DAC, I can't help but think that it has more to do with other pieces in their chain...or how those pieces work together with the DAC, which is non-linear, but we know that "pairing" is important and so by default we know that we live in a non-linear space with audio components.
Anyways, I've been playing with the various filters on the D90, and at least for me they make a big difference. It isn't subtle at all. But then again, I will recoil in severe ear pain at high pitched squeals that are inaudible to others (this isn't necessarily a good thing, since most people don't take care to regulate the volume those pitches).
So I was listening to a Tidal stream that started with Pat Metheny's latest album (very nice!) but ran off on a stream of consciousness to a bunch of other random tracks, and I had filter 1 selected on the D90 (it is the sharpest and most analytical around 20kHz). There was a tune that had a super high frequency resonance bell ring, and it really got to my ears. This doesn't often happen with other DACs, and I was impressed, as well as disturbed, this was one of the first times I've had that happen through a virtual audio source. Is it the Tidal track? The MQA compression/folding (I'm skeptical of this format, but it could have some weird artifacts)? I don't know. I'll have to keep listening and comparing. But I hope that Topping can come up with some more filters, via a firmware update, giving us more options. Or is it limited by the chip? I don't know.
Anyways, one day I'll write a full review, but for now I'm just enjoying the exploration of this new generation of DACs, in the past couple years it seems like things have really gone up a notch, but at the same time there is still no universal solution to deal with the highs and the roll-off. This seems to be where the game is at, now.
I think it's an interesting and honest observation. It makes me think, and I'm still thinking if and how your hearing is different. I can't speak for you since my hearing over 15k has been lost with age but I am very critical about high frequencies. Not that it is too much (usually too slow or rolled off) but simply not 'right'. There is some artificiality about it. What I have found out over time (double entendre) that it has to do with rise and decay of notes. Timing errors, phase, pre- and post-echos. Not with amplitude.
The problem is that conversion can be done in two ways and to confuse the issue further; a dac is not only the transition between digital and analog but part digital and part analog and the method you chose has consequences for the analog part as well (with current technology).
If you choose multibit, or R2R or ladder you might think you have to 'massage around Nyquist' (you don't actually *) but you can keep the analog part very, very simple (*) so that it doesn't get in the way of the music (ie no information loss).
If you choose single bit sigma delta you are picking up the sum of the fractures, glue them together and then you have to get rid of all the excess glue so to speak (that's the high frequency noise and smearing). DS chips are easier to fabricate due to looser tolerances but because of the high amount of oversampling you get a lot of HF hash and phase smearing. That impacts transients and that is what makes the unnatural sparkle in the high frequencies. You need a lot more filtering and you need a complex I/V and analog amplification stage that does get in the way of the music (information is lost).
If you convert over a ladder of resistors you need to calibrate them well but without oversampling (NOS) there is no unnatural pre-echo or pre-ringing. There will be some post-ringing but that is common in nature (except of course in Star Trek where they have 'inertia dampers'). There has been a lot of improvement in delta sigma dacs, especially in reducing this effect but I don't agree that even more filters will fix that. Actually, less is more and I'm back to where I arrived 10 years ago. (*) Better than new DS dacs (unless you play DSD where I'm not sure what sounds better).
As to physics and Nyquist, that's all very nice. In my experience that is very instrumental in developing a professional tunnelvision. It's always the same kind of people namedropping 'Nyquist this', 'but Nyquist', 'you don't understand Nyquist'. He might have been an engineer and physicist but his theory is more what we call information sciences. It's all virtual, not physical. And that's quite descriptive of the chasm between believers in scientism and people who believe their ears (just fill in the title of your favorite eternal flamewar and you see the two camps emerge). In the words of Chris de Burgh:
"And a chance to work it out,
For we cannot live together, and we cannot live apart;
It's the classical dilemma between the head and the heart;"
* after playing with a nice ES9038Q2M DS Dac that has all the modern features for convenience and a NOS, filterles R2R with only 1 resistor for the output stage I wanted to upgrade. I was looking at 2 choices: SMSL M400 (about the same as the A90) and the Denafrips Ares II. After lots of searching for reviews, opinions and comparison, with my experience in modifying and comparing dozens of dacs, I came to believe that the Ares was the better choice. And I'm quite certain that it is. Only when I compare the Ares to another good DS and my oldest dac, an R2R/multibit with 4 parallel vintage Philips chips that I bought for €45 10 years ago and I modified to the simplest form of output that is not according to spec, the outcome was very surprising. The oldest and cheapest and most limited dac (24-96 max) that I know measures like s#it won hands down. It has better space, clearer defined, highs are clean but detailed, it doesn't clip or saturate with soprano or piano high notes. And the bass is fast and clean too. It's not just the highs that are in order. And 'Nyquist' might say that I will get sidetones or spuriae but I haven't heard him play the hobo yet (used to tune a symphony orchestra).