Why do you think it is snake oil? It’s just an alternative connection. And they don’t have to try and sell it. It is widely accepted amongst Chinese manufacturers so you simply buy a DAC and it is there. People have a choice to use or not use it.
I have nothing real to contribute here other than increase my post count but since you’re here…
USB has been “bad” for a while compared to other interfaces. XMOS provides most of the USB chips in audio products. XMOS provides a “reference” circuit for their chips. Most reviews I’ve read over the many years note that the USB sounds inferior to other inputs and when I lookup the USB implementation its usually this XMOS design. Now, I’m not saying XMOS is bad. It’s just a common problem with a lot of audio manufacturers of “outsourcing” a part of the design they don’t understand and therefore its not integrated or implemented at a standard the other stuff is and you end-up with an inferior input.
I2S is a weird thing. I agree, it should never have been a thing to connect boxes. But, those who implement it, unlike USB, do it competently because its not something that can be outsourced. It must be designed in-house and that requires an understanding of how the implementation impacts the whole. On these devices it is often reported the I2S sounds better than the USB.
Technically is there a reason one should sound better than the other? No. But the bits are bits argument falls apart because there are a few things in both these circuits that are not bits (like clocks and grounds). Those other things can, and often do, find their way into hearing something different.
So, in conclusion to get my post count up, all this better or worse debate is going to come down to your specific kit. Try both if you have the ability. Be happy with what sounds better. Unless you’re a manufacturer, then I’d advise you pay attention, get good at USB, and do it right. Maybe even join the USB-IF and roll your own.
4 Likes
Everyone seems to forget, that using I2S is slaving the DAC to the steamer, just like with the other legacy connections, which actually can’t be a good thing, technically…
It’s best practice these days, to keep the master clock as close to the DA stages as possible.
2 Likes
I have tried I2S versus USB for many configurations and I2S consistently sounds clearer with more dynamics and a blacker background. In a specific case the PI 2 Design Mercury 2 I2S sounds better to me than a Lumin U1 USB into a Denafrips Venus II.
Now run the Lumin thru a DDC and then into the Venus II with I2S and the sound becomes equally as good if not better.
The fact is I2S separates the clock signals from the data stream and does not have the possible interference of a power lead like USB. Where USB also multiplexes the data with the clock signals and requires a more complicated circuit on both ends for transmission.
I also understand that I2S cable length limits are similar to that of USB.
The point about clock location is valid so perhaps the best upgrade for most is a DDC with OXCO clocks and I2S output like the GAIA rather than upgrading the streamer to one with I2S as the choices are limited.
2 Likes
When a given DAC model does not have an I2S input, the question if I2S is better is irrelevant.
When the DAC has both an I2S, and a USB input, they can be compared, and I2S sounds better to the ears of most users.
Why?
When the USB port of the DAC is used, the sound is transferred to the DAC by USB packets.
It is clocked by the DAC during the asymmetric transfer.
Once inside the DAC, the USB packets are translated to I2S. Then, the I2S sound is transferred to the chip of the DAC to process it.
When the I2S port of the DAC is used, the I2S sound is clocked by the word clocks of the DDC or of the streamer. (Usually, these devices have better word clocks than the ones of the DACs).
The I2S sound is transferred then directly to the chip of the DAC to be processed.
So, in this case, the word clocks of the DDC clock the sound, and the double translation of I2S to USB packets, and back from USB packets to I2S is avoided.
That’s what I know about the difference between the two ports.
This difference is not due to the quality of the USB port of the DAC. My previous DAC, a Musician, had an excellent USB port that was designed by AMR. The present DAC, a Gustard R26 Discret, has also a good XMOS USB interface.
I tested the difference between the two inputs of the R26 by playing the same track on two iPhones. One iPhone was connected directly to the USB port of the DAC. The second to the USB port of my DDC. (The DDC during this trial was an Audio-GD DI-20HE that was connected to the I2S port of the DAC.)
While playing the same tune simultaneously on both iPhones, I had only to switch between the two sources with the input selector of the DAC to hear the difference.
It was a perfect A/B comparison.
With the I2S port of the DAC, the sound stage is larger, the sound is punchier, and there’s a better clarity.
So, the improvement is not snake-oil. I could observe it with more than one DAC, a DDC or a streamer. Most users who compare both ports reach the same conclusion.
3 Likes
I didn’t say that it is. I said that the same argument you used to answer the question whether there is evidence for a noise over USB, i.e., this one …
No. But that isn’t because it is not true. It is because no one is financing this sort of research for public consumption or peer review. If it is being done, it is being done informally by the usual suspects who sell their ideas as products or to manufacturers.
… can also be used in the very same way to justify any snake oil. It’s simply a very weak argument. I can always use this for anything. “No, I don’t have evidence for [insert whatever], but that’s just because nobody provides it / pays for it, but it’s real nevertheless”.
1 Like
Yeah, reasonable enough.
It’s the standardization angle that gets me, though. A widely used standard like USB gets lots of eyes on it, lots of debate, lots of inspection, lots of experience with implementation, lots of versions and improvements. And lots of mistakes, which are also important. Henry Petroski wrote a great book called “To Engineer is Human: The Role of Failure in Successful Design”, which points out that we often learn more from engineering mistakes than from successes. Not Petroski’s only great book, but highly recommended.
Furthermore, audio devices are not the only circuits fed by USB with delicate analogue stages. Lots of instrumentation is also built that way. Problems with USB would be discovered quickly in those kinds of application.
I tend to think this whole noise shibboleth is more of a lingering myth than a reality. Consumer environments are highly noise-regulated these days. Probably my garbage disposal and garage door opener are the only electrically noisy things in my entire house. Worrying about noise over USB is pointless, especially when we are comparing it to noise over ad-hoc LVDS circuits.
I don’t know what you might mean by that. Do you have recordings of the two, so that we could examine the differences?
Why would this be? Every modern DAC re-clocks its inputs; why would the designers use bad clocks? And if they do, why would anyone want to buy that DAC?
Did you perhaps record those sessions, so that we could examine the differences?
1 Like
I presume that it’s a question of cost. A DAC is designed to produce a reasonably good sound at a given retail price.
I didn’t record the sessions, because I don’t need to prove anything to anyone.
People who can test both ports reach the same conclusion.
You can check it on other forums.
No I am not going to produce sound clips. I provided comments on my own experience. You can consider it or ignore it.
As far a Lumin is concerned - on a side note - there does seem to be some digital processing that rolls off the highs and makes the overall notes sound rounder (high frequency noise filter?). I did not realize this until I started comparing with other streamers. Some may like this “house sound” very much but I personally found the clearer less processed sound of other streamers more to my liking. Neither good nor bad - just different.
So is a streamer, or a DDC. Why would they have better clocks?
1 Like
Of course. The evidence that could end this discussion with proof of your claims is never available, it seems.
Can you point me to those discussions, please?
1 Like
The DDCs are devices that are specifically designed for a single task: clocking the sound. They are not cheap.
And it’s not something that is new. Professional DACs, like the Merging, do not have an I2S port. But they do have a port for an input of the word clock of a an external DDC. In this case, the sound is transferred to the DAC through its USB port, but the DAC works with the word clock of the external DDC.
Bill - I do not have the recording equipment that would be able to show the differences but they are clearly there. It does seem that you have never done this test yourself otherwise you would not be so determined to argue about it.
Just by common sense a circuit that is more complicated like USB that multiplexes the signal is less likely to sound as clean or clear as a circuit that uses a more straight forward signal path (even if there is some buffering etc required in both).
Of course the build quality of each individual component is still a factor. A DAC with the best USB implementation may not sound much different than a DAC with a typical I2S input.
But a DAC with the best USB circuit and the best I2S circuit will sound better thru the I2S. There is no question of that. To argue that point is to show a lack of understanding of the two systems. In my opinion.
Sure.
Go for instance to Head Fi forum. Users discuss there how they optimize specific models of DACs.
There are threads about most of the DAC models of many brands, including Gustard, Audio-GD, Denafrips, Topping, Holo Audio… You’ll see that many people use these DACs with a DDC or a streamer that has an I2S output. They do it despite the fact that buying these devices requires an additional expense that is not negligible.
Sure there is. That’s why we’re talking about it. I’m not arguing for the sake of argument, I’m trying to learn. But there’s no evidence on the “I2S is better” side of the discussion. And I’m not hearing any logical reasons for it to be better, either. While on the USB side of the discussion, there are standards efforts, lots of implementations of USB signalling, and most particularly people measuring noise over USB and not finding any. I’m willing to believe that one is as good as another, but I’d find it likely that the I2S connection would usually be more problematic, because of that.
Just as a side note: a lot of “common sense” ideas about how things work are often confounded by actual engineering, which involves measurement.
But look: the PS Audio lashup of I2S over LVDS is every bit as complicated as using USB Audio. In both cases, the digital signals have to be converted to serial differential signalling, and then back again at the DAC input. What makes this simpler with the PS Audio setup?
1 Like
But those are two completely different kinds of clock - right? The audio sample clock is not carried on the USB at all, so it is not “multiplexed with the data” as you put it.
(ETA: I assume you’re talking about asynchronous USB audio.)
1 Like
So the evidence is that audiophiles are buying stuff they supposedly (on the “USB works fine” side of the discussion) don’t need? That’s a pretty weak argument. 
How many are also putting linear power supplies on their DDCs? 
I did poke around head-fi.org a bit, looking for more info, but I have to say, most of the discussions of I2C on that site are 10-12 years old, and the correspondents seem kind of clueless. Is there anyone on that site you’d recommend reading?
1 Like
Just to give some background. I am an engineer and am used to making assessments based on evidence. I went thru all the various configurations using several DACs and streamers that I had and adding in a DDC as well.
In the end it was very clear that I2S sounds better than USB.
For all those doubting that or saying it is not so then please share with us the testing you did to come to that conclusion. It seems many posters arguing for USB have never even heard a system connected by I2S.
1 Like
There are no theoretical discussions about I2S there.
You are wrong to presume that people are stupid to buy unnecessary gear, or to presume that they have hearing issues and can not differentiate properly between the sound quality of two inputs of their own DAC.
The discussions on this site are about how to optimize the sound of a given gear, including many DAC models. People exchange practical feedback about the DAC model that they are using. There are many technically competent guys there.
You can check threads that are about DDCs, the Gustard U-18 or the Audio-GD DI-20HE for instance…
Anyway, I explained already what is the technical difference between the two inputs, and why there is a difference in sound quality.
If you have an occasion to test, and to compare between them, it will be the best thing to do.
There are exceptions…The Holo Audio May KTE sounds just as good (maybe better) using the USB input as the I2S input and you don’t have to deal with the PLL syncing with the DDC clock. I compared using both USB and I2S using the Matrix Audio X-SPDIF 2 and the Holo Audio Red.
With my previous DAC, the PS Audio DirectStream, the I2S input was far better than the USB input.