Yes they sound different.
I have heard the chord range of dacs from mobile Poly-Mojo to the DAVE sequentially in the same set up - all different
chord vs DCS - very different
cables - different
power supplies - different
cables off the ground vs on the ground in my system as assessed by non audiophile friends on many occasions - different
end of the day - if you enjoy it listen to it
I couldn’t possibly enjoy music knowing that the Moon’s phase could affect SQ, but that’s just me I guess.
@Marian @mjw could you guys more elaborate on FPGA? Do you suggest that all DACs sounds the same due to this with SoC?
1 how different: more depth, air, detail at every step up. I prefer dcs to cord - more substance then chords sound ‘skinny’ and fast, the dcs sounds real
It is a bit like putting the same engine in three different cars - the suspension, height, drive computer, transmission, tires all make a difference. The analogy here is the numerous components that surround the DAC chip, the power supply, the system that receives the digital input, etc
To Marian, not sure what you meant by the Moon phase - if it is an offhand comment about a comparison you have not done yourself - will leave that with you.
FPGA - again that is only one component of the whole
dcs haș a white paper on this in their website
It seems FPGA is an audiophile buzzword now, but there’s not much to it really. I’ll try to explain how it works, to the best of my knowledge.
In general, DACs need to do some DSP - basically oversampling and bit reduction - before the actual D/A conversion. Third-party DACs incorporate this processing in their chips so it doesn’t have to be done externally. Some people think these chips don’t do a very good job here, considering they are resource-constrained (e.g. in the internal bit depth or the number of taps for oversampling or the modulator order for bit reduction).
You can address this by oversampling the signal before feeing it to the DAC, which bypasses the internal oversampling. You can do that using Roon or HQPlayer, for example. But that is only a partial solution, since the DAC still does the final up-sampling and bit reduction, subject to the same limitations. If you want to bypass those too, you can start your own boutique and build your own DAC in-house.
But you don’t want to produce your own DSP chips, or outsource them, so you have to find something that you can program yourself. This is where FPGAs come in handy, as they bridge the divide between a full fledged computer and a dedicated (ASIC) chip. Now, you can do the DSP the way you want to (e.g. with 64-bit samples, a ton of taps, a 7th-order modulator and whatever secret sauces you can think of) and produce a 128x or 256x oversampled signal with one bit or 4-6 bits. If you choose to go with 1 bit (a.k.a. DSD), you could use a 3rd-party chip with direct DSD conversion at this point and save some trouble. But if you go for multi-bit, you have to continue to reinvent the wheel and do your own D/A, since you don’t have a 3rd-party chip that accepts these inputs. So you end up with a discrete DAC. It does what a 3rd-party chip does, so if you think about it, it could be thought of as the ultimate filter choice.
Interestingly, AKM did come up with the idea of a chip that takes a 6-bit 256x-oversampled signal, the AK4498 - most probably to cover this very scenario - but I don’t think they ever produced it.
I think the questions are flawed, no one listens only to a DAC chip, or only to a filter. You cannot swap these out of a DAC, like changing a fuse. Each is part of a piece of equipment used to create a “sound”. It would be the implementation of the DA conversion using a particular chip, or the implementation of the filtering in a piece of hardware that could have an effect on the sound.
Why theorising about this?
There is the RME ADI-2 DAC FS.
This devices exists in three iterations with three different DAC chips. AKM 4493, AKM 4497 and ESS9028Q2M.
I am pretty sure they kept the changes in implementation of the analog side to an absolut necessary minimum.
Buy all three and find out for your self and have fun while doing it. 
As soon as you add an IC DAC it’s not discrete… it’s integrated. Typically, an integrated DAC only needs additional passive components since the output stage is on the substrate.
And what about ladder DACs, which do not have “chips” ???
I very much follow struts.
Even the 0 and 1 are transferred as analog voltage or current, and in the audio field the reestablishing of the analog music signal is very difficult.
But, my experience is that it is very difficult to differentiate between different DACs if the downstream equippment is not able to show very small differences.
Can anyone actually point me towards a proper ABX test that would indicate our ears are able to distinguish between properly engineered DACs? Not subjective and certainly not sighted impressions – valid blind tests please! I’d be particularly interested in findings re a comparison between a, say, 1000 € DAC and a boutique DAC (10000 € +). This is not about measurements, but listening impressions following the ABX protocol. (I know that these tests have been conducted for amps, and the conclusion is that there is no difference – as long as amp and speakers are well matched in terms of power --, with the exception of tube amps because of distorsion, but that’s measurable).
What does “properly engineered DACs” mean? How can we (Roon users) identify how a DAC is engineered and if it is done properly?
It would be great to see two otherwise identical DAC’s built but each with a different DAC chip for comparison.
Unfortunately, you can’t really do that. Each different DAC chip will require different scaffolding circuitry.
It has been said few times already. SINAD, linearity etc.
Then you can be sure that there is no audible difference by distortions etc.
As ROON user you can check manufacturer data or reviews.
Of course many will complain then that test signals are not music, but even for them it should be solid base for other testing/ comparisons.
But I still believe that if PCB is modified in only minimum required to be able to use other chip and both DACs will measure well, there must not be any difference.
Perfect, so a plug in that includes their required scaffolding or if needed just build two DAC’s that are set up as DAC chip is needed and then same back ends. To be honest I would be surprised that not more than one manufacturer had done this already for R&D.
My question was whether we’re able to hear differences between two properly engineered DACs (meaning: based on measurements that are state of the art, or close to, that is variations are beyond the audible range, regarding frequency response, noise level, THD, dynamic range…). Just to make sure that there’s no clear technical problem in the implementation of digital and analog circuitry.
I guess such an ABX test is difficult to find. From what I gather via a cursory trawl through the internet is that those who have done such a test cannot perceive any differences, or at least, no differences that would actually matter, and certainly no differences that would allow one to judge one DAC as better or worse than the other.
I bring this up because if we can’t reliably differentiate between SOTA DACs, then the whole question of which type of DAC you have, something that was addressed in this thread, becomes kind of moot (except for those who are interested in the technical side of things, which, however alluring that may be, has nothing to do with actually listening to music).
This would also mean that one needn’t spend that much on the best DAC possible – more money simply means more functionality (doesn’t cost that much) and prestige casing (can cost a lot; I know for instance that half of the cost of a dCS DAC is casing, which has no demonstrable relevance re sound).
The better question is maybe whether or not filtering and resampling make a meaningful difference. If it is important then well made NOS DACS offer the advantage of dialing in exactly the implementation your may prefer.