Study: Is I²S interface better for DACs than S/PDIF or USB?

[Killingbeans]

It doesn't matter whether the signal is digital or analog.

It very much matter whether the signal is digital or analog.

An analog signal will have the interference added directly to the noise floor. A digital signal will be largely indifferent to the interference, as long as it doesn't become violent enough to cause bit flips. And if that happens, it will result in horrible artifacts, not just a "mild" raise of the noise floor.

The only reason to transport a digital signal on a differential pair is to ensure reliable transmission on longer cables, at higher bitrates or in noisy environments. It has zero impact on fidelity(?).

 

[Killingbeans]

But if I achieve the same (or better) transfer with one of the other data transfer options AES, SPDIF (Coax / Optical) or USB, then it doesn't matter.

How good does it have to be to be good enough? Amir has tested a large number of DACs using their USB inputs. Many of them show results with dynamic ranges so high that no sane person should be able to claim that they can hear any adverse effects. Nothing points to USB as being a bottleneck of any sorts.

 

[audio2design]

It is a shame that they haven't got together and created a standard. This is not the right way to treat a consumer!

As someone very technical and has worked on standards I find your position and arguments mainly silly. They assume a weak and uninformed consumer and dishonesty on the part of the MFRs that I don't see evident at least not from a usage of this I/F. Whether beneficial is a different argument. While there certainly much dishonesty in audio, I don't see it here. Who cares if it is an offical standard? Beta was. HDDVD was. MQA is. Was the consumer protected? No.

 

[audio2design]

How good does it have to be to be good enough? Amir has tested a large number of DACs using their USB inputs. Many of them show results with dynamic ranges so high that no sane person should be able to claim that they can hear any adverse effects. Nothing points to USB as being a bottleneck of any sorts.

Tests were not system level and not extensive across systems. I have heard and even reports here of obvious analog noise issues. I see it all the time in non isolated analog lab interfaces even with 12 bits analog. The issues happen system level not in isolation.

 

[Roland68]

How good does it have to be to be good enough? Amir has tested a large number of DACs using their USB inputs. Many of them show results with dynamic ranges so high that no sane person should be able to claim that they can hear any adverse effects. Nothing points to USB as being a bottleneck of any sorts.

Did I say otherwise?
I only tried to show possibilities and approaches to the question asked, it does not help to interpret something purely.

But Amir also tested a lot of bad to awful devices.
And I already had devices on the table that sounded better via my USB -> I2S -> I2S / LVDS solution than via the integrated USB interface, but this only indicates a bad implementation, an error or a defect.
But this also applies to all other interfaces.

 

[Veri]

Did I say otherwise?
I only tried to show possibilities and approaches to the question asked, it does not help to interpret something purely.

But Amir also tested a lot of bad to awful devices.
And I already had devices on the table that sounded better via my USB -> I2S -> I2S / LVDS solution than via the integrated USB interface, but this only indicates a bad implementation, an error or a defect.
But this also applies to all other interfaces.

There's also a difference between they sounded better and you thought they sounded better....

 

[Roland68]

There's also a difference between they sounded better and you thought they sounded better....

I absolutely agree with you.
Now comes the BUT, the differences I heard were audible in the resolution. That is, the details of recordings that I know very well were either not audible or only audible with an effort.
Most of these devices also had problems if the cable was not "good". From my IT days I have USB 2.0 A B cables that I once classified from "1a" to "unusable". Many of these DACs already produced clear dropouts with cables in my "OK" or "still works" category. My Gustard A18 MQA runs even with a cable from my category "too unsafe" without a difference that I can hear.

 

[Killingbeans]

I have heard and even reports here of obvious analog noise issues. I see it all the time in non isolated analog lab interfaces even with 12 bits analog. The issues happen system level not in isolation.

Absolutely. But all of those issuses are caused by ground loops? In that sense, it has nothing to do with the fidelity of a specific transfer protocol?

Did I say otherwise?
I only tried to show possibilities and approaches to the question asked, it does not help to interpret something purely.

Fair enough. I just don't understand why we keep discussing the benefits of a solution that was obviously designed to remedy a non-existent problem.

 

[audio2design]

Absolutely. But all of those issuses are caused by ground loops? In that sense, it has nothing to do with the fidelity of a specific transfer protocol?

I don't think anyone, with ignorant exception, is claiming it's protocol related.

Ground loops, yes but not purely in the line frequency noise sense. Ground connections can carry high frequencies too.

 

[Killingbeans]

Ground loops, yes but not purely in the line frequency noise sense. Ground connections can carry high frequencies too.

Certainly. But putting an Adum4160/3160 somewhere in the USB chain still fixes the problem. No need for fancy schmancy i2s hardware.

 

[audio2design]

Certainly. But putting an Adum4160/3160 somewhere in the USB chain still fixes the problem. No need for fancy schmancy i2s hardware.

Well you may have to go for something a bit more sophisticated for USB high speed. I don't see I2S inherently providing noise immunity. I personally would never buy it, but I don't bash based on nonsense arguments either.

 

[Killingbeans]

Well you may have to go for something a bit more sophisticated for USB high speed.

"Fully USB 2.0 Compliant" isn't good enough?

I personally would never buy it, but I don't bash based on nonsense arguments either.

Saying that it's a technology being used in a way it was never designed for in order to solve problems that can not possibly be proven to exist is a nonsense argument?

Does it hurt anything? ..No. But in what way is it not a complete waste of cables, connectors and interfaces?

BTW, I'm not really pissed or anything. I just don't understand why anybody feel the need to defend its existence.

 

[mansr]

"Fully USB 2.0 Compliant" isn't good enough?

The ADUM devices are full-speed only.

 

[Killingbeans]

Ahh, I see. Joke's on me Guess the USB 2.0 compliance only refers to the power requirements?

 

[Roland68]

"Fully USB 2.0 Compliant" isn't good enough?

From the ADUM4160 data sheet "Low and full speed data rate: 1.5 Mbps and 12 Mbps"
The specification you are assuming, and which is usually used in DACs, is USB 2.0 high-speed with 480 Mbps.

 

[audio2design]

"Fully USB 2.0 Compliant" isn't good enough?



Saying that it's a technology being used in a way it was never designed for in order to solve problems that can not possibly be proven to exist is a nonsense argument?

Does it hurt anything? ..No. But in what way is it not a complete waste of cables, connectors and interfaces?

BTW, I'm not really pissed or anything. I just don't understand why anybody feel the need to defend its existence.

It was not "brought into the world" to exceed the capabilities of USB. It was "developed" to replace electrical SPDIF, and in that respect it is far far superior. SPDIF is often carried on RCA connectors that are no where near a controlled impedance and hence generate reflections, worse, right around a typical cable length. It is real, and can generate significant jitter. Yes you can get rid of that jitter, but we know not everything is equal.

USB can have noise issues, and does at times at the system level. I personally have experienced it in audio and lab work and many others have. As noted, easily solved, but not everyone wants a USB I/F.

USB 2.0 Compliant does NOT mean it supports USB 2.0 High Speed.

USB has backward compatibility, but being a protocol does open itself up to interop issues, especially say with a simple Ethernet or WiFi end-point into a DAC. While I2S over LVDS may not be a standard, the odds of interop issues is actually quite low.

I don't see the point of bashing something that is causing no arm, not even to those buying it. USB has its issues. It is not absolutely perfect. Neither is I2S obviously.

 

[Roland68]

In general about USB and in particular about USB audio:
Unfortunately, USB is one of those narrow-track solutions / standards that has prevailed over better technologies such as FireWire due to its low price and the resulting rapidly growing wide range of products.
USB was problematic from the start and did not have a good reputation in the professional IT sector, and that was not without reason either. The problems could mostly only be solved by using different hardware.
The biggest problems occurred during the introduction of USB 3.0. At that time, many forums kept compatibility lists and magazines, including the `ct, tested umpteen devices to give users information about which devices were working with each other and which were not. That gave even the much more expensive Thunderbolt interface a boost in the Windows world.
But the manufacturers have learned that in terms of hardware they have to do more than the standard requires, and USB 3.0 and its successors are now working well.
However, this comes from my professional experience with many thousands of USB connections, including wireless USB, USB over IP, USB over VPN, network USB hubs, hundreds of USB dongels for expensive software, etc.
As an end user, I never have to have a problem, or constantly, the epitome of chance.

But anyone who is technically familiar with USB and its limitations knows that it was never made, let alone developed, for the transmission of high-quality audio.

So I find it all the more amazing how well USB works for high quality audio! But it's easy for me to talk, I haven't had to install a driver for USB in the last 14 years.

But what idiot came up with the idea of laying a power supply line (and one that is so dirty) within the shielding of a data line for which there is a defined characteristic impedance. What rubbish.
But that's all just my personal opinion.

 

[Killingbeans]

It was not "brought into the world" to exceed the capabilities of USB.

Maybe I've just been looking at the wrong places, but every bit of marketing I've seen revolving around i2s seems to say the opposite?

It was "developed" to replace electrical SPDIF, and in that respect it is far far superior.

It was developed as a serial bus for inter-IC communication on PCBs. The further "development" into cable transportation was based on the assumption that straight-from-the-IC automatically means higher fidelity.

SPDIF is often carried on RCA connectors that are no where near a controlled impedance and hence generate reflections, worse, right around a typical cable length. It is real, and can generate significant jitter. Yes you can get rid of that jitter, but we know not everything is equal.

Point taken. That's one of the reasons why I love toslink

I don't see the point of bashing something that is causing no arm, not even to those buying it.

If it was just another standard with more convenience than the old ones, I wouldn't have any beef with it either. It's the fact that it's being marketed as something that's superior in all kinds of nonsense ways that makes me mildly annoyed.

 

[audio2design]

It was developed as a serial bus for inter-IC communication on PCBs. The further "development" into cable transportation was based on the assumption that straight-from-the-IC automatically means higher fidelity.

OBVIOUSLY we are talking about I2S over LVDS which makes you comment irrelevant and pedantic just to be pedantic. It was developed based on electrical SPDIF and TOSLINK being open to poor performance, and by transporting clocking, potentially removing issues of clock extraction that occurs with electrical spdif and TOSLINK. Granted audiophile product designers make some questionable designs, but you would have to screw up pretty bad with I2S over LVDS to make it worse than electrical SPDIF or TOSLINK, and nothing is preventing further clock clean up methods.

 

[Killingbeans]

OBVIOUSLY we are talking about I2S over LVDS which makes you comment irrelevant and pedantic just to be pedantic.

Pedantic, maybe. But I don't see why it's irrelevant. I never questioned its ability to reliably transfer data. What boils my piss is the romantic notions of "purity" that surrounds it.

I absolutely understand the theoretical benefits you put forth, but I have serious doubts about their practical implications.

You see it as harmless and I see it as pointless, and I don't think there's much else to do than leave it at that.