First impressions of new MH-DA006, Musetec flagship

@sns 

If USB is so flawed, why hasn't the market driven it to the trash bin? Every DAC manufacturer has the ability to provide an I2S and/or external clock input, yet most don't even though it would provide an enhanced revenue stream in the form of DDC and clock sales.

The top DAC manufacturers (DCS, Weiss, MSB, Wadax, etc.) eschew the use of I2S for USB. Sonore doesn't provide an I2S output on their Rendu devices because they found that implementation matters more than the "format". Jusi Laako, founder of Signalyst/HQ Player is fairly vocal regarding the negatives of I2S due to the clock being so far removed from the DAC.

PS Audio was one of the first manufacturers to promote I2S in their original PerfectWave DAC for which I was a Beta Tester. At that time (~2007) I was all-in on I2S and it definitely sounded better than USB. As technology has progressed, improvements in USB isolation and better clock implementations have reduced any perceived benefit provided by I2S.

If you feel that your system sounds better via I2S compared to USB, great, but that doesn't mean it's empirically better and it's certainly not an absolute.

@sns 

First of all we agree that what matters most is the sound, and not measurements.

Obviously, what triggered my last post was your "unfortunately only Femto clock" remark.  But when you are talking specifically about clocks, you are focusing on jitter (and phase noise) and that is measureable.  And it would appear that the Gaia's OCXO has no partiular advantage over the ubiquitous Accusilicons and Crysteks that you seem to disparage.  And there are yet better Femto clocks available!   In identical set-ups simply substituting better clocks has clearly been shown to improve the sound in ways considered imprtant to audiophiles like space, air, focus, soundstage, etc.

As for I2S vs. USB outputs, I was only considering your remarks about those of the Gaia.

And last of all we agree that what matters most is the sound, and not measurements.

@melm 

Well stated, and spot-on.

@catastrofe

The answer is that for most people USB is good enough.  It also easily allows people to use their laptops directly into a DAC and a conveniently long cord.  Their streamer and file storage are already at hand.  We used to put a filter on the line and call it a day.  And if you can make USB work as well or better than the alternatives, that's great.

For the great majority that do not use USB out to their DAC, they use a device with spdif, which is very closely related to I2S.  The great majority of all the RPi hat devices output spdif.  Some reviewers, Hans Beekhuyzen for one, says that for him I2S and spdif or AES/EBU (that is similar to spdif) are virtually indistinguishable coming from the same streamer.  And the spdif cable can be conveniently long.  The highest performing streamer he and a lot of other people have ever used is the Grimm Audio MU1.  It has no I2S output, but only spdif and AES/EBU.

I don't see any reason to have USB in the mix if it avoidable.  Going from ethernet to I2S is such a clear path.  And so too it would seem, based on what hardware is available, going from ethernet to spdif-AES/EBU as the next best thing.  No need for what we used to call USB decrapification.

I'm not stating an overarching disparagement of usb, in many cases usb may outperform I2S, I keep on reiterating my experience with Singxer SU6 vs usb. I will continue to affirm what I believe to be inherent advantage of dedicated lines for data and clock, and no need to convert from usb to native I2S within dac. Does it seem reasonable to believe extracting data and clock streams from one cable is an advantage, same with detour through usb boards. With I2S one is simply supplying isolated and dedicated clock and data streams directly into native I2S path of dac.

 Now I agree the downside of I2S is the cable, however, I'm reasoning the superiority of the clock in Gaia over 006 clock supersedes or overrides the downside of clock being further away and having to travel through short cable to I2S stream inside dac. Superior sound quality I'm getting via I2S is provides evidence for this.

As for choices made by manufacturers in regard to interfaces, plenty of dacs provide I2S inputs including Rockna, Mola Mola Tambaqui, PS Audio, most of the Chifi, and some others I've seen listed on forum. More at issue here is the failure of streamer manufacturers to provide I2S outputs. With so few streamers offering this output easy to understand why dacs wouldn't provide for it. Now, as to why streamers don't offer it, here is how I see it. Virtually all off the shelf computer motherboards (which is what all streamers are) have usb capability on board. These boards were originally developed for general purpose, only in relatively recent years have they become optimized solely for audio. We also have follow the leader syndrome, usb became the de facto interface because it was no hassle   to integrate, integrating I2S requires more work, another signal path, clock and power supply. I also agree it can sound fine since so many years and so many manufacturers working on optimizing it. I've used many usb optimizing solutions over the years, each subsequent solution bettering the previous. If I hadn't purchased  Laiv Harmony with the specific intention of trying I2S with it I wouldn't have even bothered with I2S to 006.

As to doing away with a DDC altogether, I have option of installing Pink Faun I2S board in my custom build streamer, I'd power via my Uptone JS 2 choke based lps. In going this direction I'd completely eliminate my Sonore OpticalModule, OpticalRendu (no more optical network conversion and total elimination of usb, AQ Diamond usb cable, lps, AC cables, DC cables). Question becomes does less become more?

Finally, based on reviews of the world class streamers and dacs, I have no doubt the proprietary interfaces they use are superior to both usb and I2S. Still, the native signal path within dacs remains I2S so the ultimate solution is to provide a superior clock and power supply for that clock within the dac. As for the streamer I haven't researched these proprietary interfaces enough to know exactly what they do, however, it seems reasonable to believe they provide dedicated runs for clock and data, I can't imagine mixing up data and clock on a single cable is the optimal solution.