I2S: Is it a good external interconnect? Give your opinion and findings!

[Mr.C]

I don't know the technobabel for this, but doesn't firewire transfer data and not a signal? I thought that the data was transfered to the DAC and then converted to a signal by the dac. This way the clock would tell the computer when to send the data packets but maybe the clock is also data rather than a signal? Is that on the right track?

 

[cfmsp]

 

 

"For the record, isochronous IS used in audio (by Weiss at least)."

 

It seems I've been laboring under the incorrect assumption that isochronous and asynchronous are mutually exclusive.

 

"But my understanding is that asynchronous/isochronous USB/firewire DACs don't simply 'communicate' the rate/flow of data, but actually provide the clock to the computer. I.e. the DAC's clock is in someway imbedded in the protocol."

 

I'm not aware of any clock signal being imbedded within Firewire.

 

"The only alternative I can see is if a FIFO buffer is used in the DAC, in which case the rate/flow of data from the computer wouldn't be an issue (as long as it was high enough)."

 

I think the communications is to insure that the buffer does not 'underflow' (as you suggest) or 'overflow'. As I understand it, there is no provision for a resend of data, so we wouldn't want to lose any data due to buffer overflow either.

 

clay

 

 

[cfmsp]

I_S

as always, thanks for your clarifications.

 

I"m always happy to learn something, esp when what I thought I knew was imprecise, or just plain wrong.

 

 

"There is definitely no I2S async spec."

 

Yeah, that didn't sound exactly right when I wrote it. But, I know I read recently that one of the protocols not noted for Async behaviour does have an 'optional' spec for async, but now I'm confused about which one it was.

 

clay

 

[cfmsp]

 

Is it correct to assume then that isochronous and asynchronous are not mutually exclusive, just as is the case with USB?

 

I had assumed it was one or the other.

 

clay

 

[cfmsp]

"I don't know the technobabel for this, but doesn't firewire transfer data and not a signal?"

 

I'm not exactly sure of your distinction between data and signal. If by 'signal', you mean data with embedded clock in it, as is the case with AES/EBU and it's variants, then no, Firewire does not transfer a 'signal'.

 

One of the benefits of Async Firewire, Async USB, and network players is to hold off on combining clock and data (at the source) and thereby limit the potential effects of RF / EMI / etc. on the clock while data is being transferred. This I believe is one of the reasons S/PDIF cables are so critical, and is also one of the reasons why people don't understand why USB cables can make a difference (assuming that they do, I don't have firsthand experience, not having a USB DAC) - that is, they think of non-Async USB as being just data.

 

Another benefit of async firewire/USB is that you don't suffer the challenge of keeping two clocks 'in sync' at the pico second level.

 

Perhaps you are asking if the content of the transmission is binary numbers, or an electrical representation of same? I still get confused on that. I know that many, myself included, tend to think of all digital data as just 1s and 0s, but I recall someone (CG? I_S?) recently clarifying that it's not exactly like I had imagined. Obviously, I don't know the techno-babble for this, either.

 

clay

 

 

[manisandher]

Thanks all - I've definitely got a better understanding of how USB and firewire work.

 

But I2S???

 

I_S, you said, "Mani, you are correct to worry about impedance matching, but this will only affect the DAC if it is using a real wordclock for timing..."

 

Does I2S require a real wordclock from the DAC for timing? If so, is there any attempt to match the I/O impedance? And if not, how badly does this affect the performance of the connection? I would have thought it'd be a disaster...

 

Mani.

 

 

 

[cfmsp]

 

 

"Does I2S require a real wordclock from the DAC for timing?"

 

I don't think it requires one, but it might be possible to implement one.

 

Where's Steve Nugent when you need him? If anyone has done this, it would be Steve at Empirical.

 

I'd be surprised if he hadn't done it, actually.

 

clay

 

 

 

 

[Wavelength]

Clay,

 

No OS supports asynchronous Firewire audio. So most vendors who use Firewire use a bulk mode and write drivers. This would make for an async setup. If they use the adaptive (native) expect some heavy jitter reduction as most of these interfaces that were tested by Paul Miller were in the 5000-10000ps of jitter.

 

I2S is not a protocol, there is no async method it is totally synchronous protocol.

 

The use of word clock is not really a good one just for playback. It is really only useful if you are trying to synchronize several devices together.

 

Thanks

GOrdon

 

[manisandher]

Gordan, you said, "The use of word clock is not really a good one just for playback. It is really only useful if you are trying to synchronize several devices together."

 

This is NOT my experience at all. If your DAC has a wordclock output and your interface a wordclock input, I would strongly recommend that you use them... even just for playback.

 

Indeed, this was the setup at the CA symposium... for a reason!

 

Mani.

 

[cfmsp]

 

"I would strongly recommend that you use them... even just for playback."

 

Mani,

it's my understanding that this is sound advice with AES/SBU and S?PDIF, although I can't speak from personal experience.

 

For Async USB and Firewire, Gordon's comments seem proper, based on the nature of how they operate, i.e., there's not a separate clock that 'needs' to be synchronized.

 

YMMV,

clay

 

[manisandher]

Yes, agreed.

 

Mani.

 

[Wavelength]

Mani,

 

Enough with the I2S... it is not an external interface. Ok let's just look at the cable...

 

RJ45

8 - Serial Bit Clock 64fs BCLK

7 - BCLK return (ground)

6 - Data clocked by pin 8 BCLK

5 - Data return (ground)

4 - Left Right Clock LRCLK WCLK (Word Clock)

3 - LRCLK return (ground)

2 - Master Clock (256fs) MCLK

1 - MCLK return (ground)

 

Ok let's just look at this from the Master Clock down... MCLK/4 = BCLK (bit clock) which clocks out the data. MCLK/256 or BCLK/64 = LRCLK/WCLK which determins if the data is for the Right or the Left channel.

 

If you have jitter on the BCLK at all that will cause the DATA to have jitter related data errors. That's bad stuff...

 

Anyways... this is silly... this is not a standard and nobody uses it. Those that do are fringe and it basically has never caught on.

 

Thanks

Gordon

 

[cfmsp]

 

 

"Anyways... this is silly... this is not a standard and nobody uses it."

 

Gordon, I believe the OP was interested in I2S due to the recent release of the PS Audio Perfect Wave DAC, which uses I2s (over HDMI, not RJ45) between Transport and Dac.

 

clay

 

 

[PeterSt]

I don't think what Gordon means is related to a cable topology.

Why he presents it like that, I don't know ...

 

I guess he tries to set the standard for the pinout. Just in case. Hahaha.

 

Well, it is about time that my own temporary I2S solution is going to be replaced by the intended one (Firewire -> few cm I2S), so let's see what a difference it will bring. I guess I'll first have to visualize the current jitter, in order to have a reference.

I will let you all know what came from it (might take a week or more).

 

Peter

 

[timequest]

First, I am nowhere close in proximity to being as informed or knowledgeable about digital audio (and hifi audio tech in general) as is Mr. Rankin. I do know that the Perfect Wave DAC (as well as the combination of the DAC and Transport) sounds simply wonderful. Evidently PS has managed to effectively utilize an I2S external interface – I couldn’t begin to explain the details (I simply don’t know).

 

If Wavelength has perfected the technology (my words), then why did Mr. Rankin pull his product from review in TAS? Evidently, he would not allow the Cosecant to be reviewed in TAS because the test “bed” was not to his liking. By bringing this up, I may come off as smug, but so does Mr. Rankin when he openly disregards (totally disregards) any possible solution to the I2S external interface matter. By doing so, he – whether it is unintentional, comes across as…well, a bit self-righteous. I have no doubt that Mr. Rankin is familiar with PS Audio’s recent efforts; and also of the recent industry praise bestowed on the very products that effectively (my words) implement the I2S external interface (which, as was pointed out, does not utilize RJ45).

 

Obviously, these guys (the true industry leaders – like Mr. Rankin) are audiophiles just like the rest of us and they no doubt are passionate about the issues – and, like the rest of us, opinionated.

But it is really unprofessional to let your passions (and opinions) go unchecked in a public forum (this isn’t a round table discussion with fellow designers and manufacturers – is it??).

 

Instead of knocking someone else’s tactical implementation strategies, offer some reasons (proof) why someone’s product doesn’t sound as good as it could. Then, allow the reviewers to actually review your products.

 

No disrespect intended (really).

 

 

[The Computer Audiophile]

Hi Timequest - Can you tell us what you really think? Only kidding of course.

 

Your questions are valid and I also suspect there are valid answers.

 

[Wavelength]

Timebandit,

 

The reason for demanding a with draw from the TAS review was simple. It was going to be a hack job and I could see that last September when i sent the Cosecant. The interface on all my digital products is USB. It is a spec there is a complete industry built around it. Internally all my digital products use I2S so I am and have been using this since the late 80's.

 

The thing that kills me about the Taffel/Stone debunkle is that they killed another complete sector of possible Audiophiles. Their subscription is in the can and now they can count out getting any of these people on board and with that I would say that maybe the US would be down to one magazine which would be a crime.

 

But both the reviewers really did not take any time to really investigate and review anything. That makes it a total waste of paper in my book.

 

It would be like Auto Week reviewing the latest Alpha, Audi, Porsche but required to do so with 13" wheels.

 

Peter... I was trying to show the connection and the relevance of the associated connections. Probably pretty stupid because most people have no idea what it takes or even how it works to assure I2S on a cable is even functioning.

 

But anyone can see a cable adds capacitence to the load and that will kill any square wave. So when the I2S enters the receiving end then everything just falls apart. If you have any good 350-500MHz scope you can see how bad the I2S cabling makes anything. Heck I do this all the time for prototype stuff, cables everywhere... it's easy to see why I2S out of the box with a cable is just a complete failure.

 

Thanks

Gordon

 

[idiot_savant]

I've read the article, and I would say Gordon was correct not to participate - the article was clearly a hatchet job on all things USB - I swear, I will be impolite to the next person who comes up with the "USB is for printers" line.

 

We now have two sets of measurements from stereophile indicating async USB has excellent jitter characteristics. Where are the I2S ones? I am not trying to start a fight, but jitter is something you can measure, and if you use a proprietary interface because it improves something you can measure, you must be able to measure it, right? I have seen the article in HiFi News, which illustrates high jitter via I2S...

 

Again, I do not doubt that the PS audio stuff sounds lovely.

 

I do doubt it is anything to do with the reasons they say.

 

your friendly neighbourhood idiot

 

[timequest]

Look, I know Mr. Rankin is an industry leader and I certainly do respect his thoughts. In fact, his opinions about I2S, quite frankly, leave me with questions about just how good (or poor) an external I2S interface may actually be. The fact still remains that I, along with many others, have heard the Perfect Wave products connected via I2S and they not only sounded superior to a cheap Wal-Mart CD player, but they actually sounded remarkably good. A blanket statement like Mr. Rankin’s pretty much leaves no room for the Perfect Wave products – as if there is no way they could actually sound excellent (as in high-end quality) because of the issues inherent with the problematic external I2S interface.

 

I am more curious about this whole issue now than ever before. Obviously, Mr. Rankin is not alone in his beliefs. On the other hand, it has been pointed out that there are other industry leaders who disagree. Damn. I just want it to be easy.

 

 

[[email protected]]

hello

 

last year when the stello sig dac came out in vegas they had all the inputs of the dac hooked up playing all sorts of stuff and I2s input hooked up with the stock cable that they make and that comes with the transport sounded the best. now i'm not saying it is the best way to go but im saying that it sounded better then anything else hooked up to those three days. it may not be the best but it sure doesn't suck.

 

[[email protected]]

hello

 

last year when the stello sig dac came out in vegas they had all the inputs of the dac hooked up playing all sorts of stuff and I2s input hooked up with the stock cable that they make and that comes with the transport sounded the best. now i'm not saying it is the best way to go but im saying that it sounded better then anything else hooked up to those three days. it may not be the best but it sure doesn't suck.

 

[cfmsp]

In response to an ongoing discussion about the possible efficacy of the I2S spec as an interface protocol, Gordon said:

 

"I2S is internal protocol good for about 2cm...anything more and it will sound like crap. If you go between two different units don't expect this to work as good as say a $49 wal mart special cd player."

 

Timequest apparently took umbrage at Gordon's remarks, saying:

 

"Then why does the new PS Audio PWT/PWD sound so (extremely) good utilizing their proprietary I2S interface? After extensive auditioning of these components, your statement (for me) truly impacts your credibility..."

 

and later Timequest says:

 

"...my initial response was to Gordon, who made a really poor decision to post such a statement on an open forum. I’m not smart enough to understand whether, or not the reason the Perfect Wave devices sound so good has anything to do with the implementation of I2S – according to PSA it does."

 

and then today, Timequest says:

 

"Evidently PS has managed to effectively utilize an I2S external interface..." followed by...

 

"I may come off as smug, but so does Mr. Rankin when he openly disregards (totally disregards) any possible solution to the I2S external interface matter. By doing so, he – whether it is unintentional, comes across as…well, a bit self-righteous."

 

So, is Gordon self-righteous, or is he just right? I set out to decide for myself, which I believe I did.

 

Initially I thought that Gordon's comments should NOT be taken literally. Certainly, saying that something 'sounds like crap' is a figure of speech, and should NEVER be taken literally, nor should anyone be offended if someone says that their favorite piece of gear 'sounds like crap'. One man's crap is another man's 'sand amp'.

 

But we can take a peek to see how literally accurate the following statement by Gordon was:

"If you go between two different units don't expect this to work as good as say a $49 wal mart special cd player."

 

please note: Gordon says "don't expect this to work as good", not "sound as good", which relates more to the engineering aspect than say, 'purely sound quality'.

 

I'll report what I found, and you the reader can be the judge as to it's accuracy.

 

In a Stereophile review of 3rd generation iPod (circa 2003), John Atkinson said this about the iPod's jitter measurements: "The overall result is superbly low, at 225 picoseconds peak-peak."

 

http://www.stereophile.com/mediaservers/934/index5.html

 

In a HiFi News review (published this month) of the PS Audio Perfect Wave Transport/DAC, the I2S interface was reviewed extensively, and the jitter in Native (aka I2S) mode was reported to be 300 psec.

 

http://www.zinio.com/browse/publications/index.jsp?productId=204644780

 

note: you don't have to pay for the magazine in order to read the review. After clicking on the 'see inside' icon, drag the slider at the bottom of the page to the spread for pages 20 / 21. From here you can zoom in and read the pages. To read the last two pages, including measurements, drag the slider once more to pages 22 / 23 and zoom in again. There are a limited number of 'zooms' allowed, which will prevent you from reading the entire magazine, but one can easily read the PS Audio review.

 

 

Here you can buy a more recent generation ipod at Walmart for $45.

 

http://tinyurl.com/45-ipod-shuffle-Walmart

 

or you can buy a 3rd generation ipod (exactly as was measured by Stereophile in 2003) for $49 on ebay.

 

http://tinyurl.com/49dollar-ipod-3G

 

 

Self-righteous or right?

 

you be the judge.

 

enjoy

clay

 

edit: PS, yes I know that an ipod is NOT a CD player, but it's the cheapest device I could find with some objective measure (in this case, jitter) that could be compared to the PS Audio device to see if, as Gordon says, it could "work as good as say a $49 wal mart special cd player."

 

 

 

 

 

 

 

[timequest]

Well there you have it. The Ipod is better.

 

And Gordon....that's Mr. Timebandit pal.

 

Cheers (really)

 

[cfmsp]

 

Timequest,

 

No one is saying that your DAC doesn't sound good, but that I2S is probably NOT the reason.

 

It doesn't sound like you'll be using the I2S feature any way, as you're not planning to buy the transport.

 

 

Enjoy it (really).

 

clay

 

 

 

[Wavelength]

Timebandit,

 

In the case of the PS audio you are connecting two pieces that were designed together.

 

The big failure for any NON supported interface would happen if you connected two pieces from different companies. This is when I2S really fails.

 

But even more so take the comments made by many respectful reviews when they talk about the difference in sound between a good single box and multiple boxes. Most of them feel the single box sounds a lot better. In this way the I2S is kept really short and the impedance matching of traces can be held to a really high standard.

 

Timebandit, ever wonder how the PS audio would sound if it was a single box?

 

Thanks

Gordon