Best USB cable to use between computer and dac?

[Julf]

We still have to keep in mind that those links are about speaker cables, where we are talking about definitely analog things going on, and where the current carrying capability, resistance and inductance (all measurable quantities) definitely make a difference (and we can show measurements to that effect). But in this thread we are talking about USB cables, where (as long as we remember the constraints - we are talking about USB 2.0 cables, not any of the older specs, and asynchronous USB interfaces), nobody has really showed a good physics-based theory of what could be causing audible differences.

 

That is why some of us (with an engineering/scientific background, or with a rational approach to things) are struggling with the concept of audible differences between USB cables. Employing Occam's Razor, it is unfortunately rather tempting to attribute the effects to placebo effects, marketing and snake oil. At the same time, I can see why the more existentialist/belief-based/touchy-feely ones amongst us point to their own subjective observations as proof that we bitheads are plain wrong.

 

I think we are dealing with an issue that transcends USB cables, audio or technology, and I don't think any of our debates here will change that. See "Zen and the Art of Motorcycle Maintenance"

 

Julf

 

 

[Mark Powell]

I heard a definite difference between an 18 inch USB cable and one 6.7 times as long (10 ft), and reported it here. As Pass says about speaker cables, length differences will tend to be far more significant (if at all) than anything else. But dropping it in to a cable discussion tends to be rather a 'show stopper' and those who believe in alien abductions and the like are not fond of it.

 

[Jud]

Thats all well and good but will be seen as errors. So it will either pass or it won't.

 

Wowee, how many times....

 

(1) The fact that a cable is passing digital information - more accurately, an analog signal representing digital information - does not immunize it from passing electrical noise or RFI through signal or ground (or in the case of USB, power) connections to the digital and analog circuitry of the DAC.

 

(2) Perfection or total failure are not the only two modes for transmission of digital information. It is impossible to reconstruct analog from digital correctly without accurate timing information. If the digital input is not asynchronous, the cable affects the timing information, inaccuracies in which can cause distortions in the reconstructed analog signal short of total failure.

 

(Even if the digital input is asynchronous, I wonder whether electrical noise or RFI might affect timing of the data out of the DAC's buffer. I've read some things that make me think it's possible, but as a layperson I don't know nearly enough to be sure.)

 

I've been reading this "it will either pass or it won't" stuff since DACs and digital cables originated over 20 years ago. (Not from audio engineers.) I'm thinking it's about time that people graduated from a cartoon vision of 1s and 0s happily marching down a cable untroubled by any actual physics such as electrical or timing effects. Simplification is an excellent tool in making models to investigate or understand various phenomena, but in the process of simplification we've got to be careful not to throw out data that will be material to obtaining results that agree with reality. Ignoring electrical and timing effects in digital cables doesn't comport with reality.

 

If people don't hear differences in their own systems, or hear differences, but very inexpensive cables do at least as well as more costly ones, that is wonderful. Save your money and spend it on components, wine, women or men, song.... And for those of us who hear differences that sometimes (not always) favor more costly cables, hey - let us enjoy our great sounding music and have our discussions about which cables we like, OK?

 

[Julf]

Absolutely - if a cable has influence on the sound, the influence will be proportional to the length - a cable 10 times as long will have 10 times the effect than a shorter one.

 

[Mark Powell]

But doesn't it all do that? The PC tracks, the resistors, the wires from the chip pins to the silicon chip itself, etc?

 

[Jud]

But in this thread we are talking about USB cables, where (as long as we remember the constraints - we are talking about USB 2.0 cables, not any of the older specs, and asynchronous USB interfaces), nobody has really showed a good physics-based theory of what could be causing audible differences.

 

Hi, Julf. It is always a pleasure discussing audio with you.

 

As noted in my most recent comment, I think we must be careful not to forget that cables passing digital information are not immune to all the same physics that applies to analog cables. That means electrical noise and RFI can be passed through to the digital and analog circuitry of the DAC, and it is possible that some USB cables may do better in that department than others. Also, I am sure you know that inaccuracies in timing information short of total failure (i.e., jitter) are quite real and can cause distortions in the reconstructed analog signal. If we are *not* talking about an asynchronous DAC input, then the cable is part of the system that can cause these inaccuracies.

 

If we *are* talking about an asynchronous USB interface at the input of the DAC, then there should be no direct contribution of the USB cable to the jitter at the output of the DAC's buffer. What I do wonder about, as noted in my previous comment, is whether electrical noise, RFI, etc., might indirectly affect jitter at the DAC's buffer. (See Damien Plisson's paper at the AMR site for what seems to me as a layperson a plausible discussion of how this might occur.)

 

[Mark Powell]

I agree. I might even try one of the lower cost 'specialist' ones that you have suggested. But I still think that the difference between an 18 inch one and a ten foot one might be far greater. Why do so many refuse to discuss it? Others have made this point, on other threads, and it is never (Julf excepted) responded to.

 

[Julf]

No, digital communications aren't immune to noise and distortion. But how often do you get "almost but not quite" the right characters on your computer screen, or in your text files on your computer, or on a web page you download? The beauty of a properly designed digital transmission protocol is that it allows you to reconstruct the exact data that was sent, despite the analog characteristics of the transmission media. So I hope we can agree that the actual data values won't be changed by using a different USB cable? That leaves timing as the only possible variable - but no modern system relies on the synchronization of individual samples of sound over the length of the whole chain. The computer sending data over USB sends packets, and the receiver receives packets, and then passes out individual sound samples based on it's own internal clock.

 

Where things can go wrong is at the boundaries of packets - what happens if the receiving end runs out of data before the sending end has sent more? That is why the later asynchronous USB spec is better than the older, data-oriented synchronous spec, as the asynchronous mode allows the receiver to determine the timing.

 

So, if you still have buffer boundary issues even with asynchronous USB, it is a sign of badly designed software or hardware that is not fast enough to keep up, not of noise or interference on the USB bus.

 

Julf

 

 

[Jud]

So, if you still have buffer boundary issues even with asynchronous USB, it is a sign of badly designed software or hardware

 

I don't think it can just be data running out of the buffer before more data is delivered. That would cause audible dropouts. Short of such audible dropouts, there is jitter, which I don't think is primarily caused by data insufficiency at the DAC's buffer. (I'm ready to be corrected by those who know more.) We've got some very smart engineers here, including the person who has as much as anyone to do with the current popularity of the asynchronous USB interface (Gordon Rankin), and none of them is claiming either (1) zero jitter, or (2) that they have resolved exactly what is causing sonic differences between cables.

 

I've mentioned a couple of times here Gordon Rankin's comment at how frustrating it was to him that even with his careful research and good measuring equipment it was near-impossible to completely quantify all interactions between various phenomena affecting sound quality, or be certain he'd captured all the phenomena that *could* affect sound quality. Of course that doesn't mean people aren't making some damn fine equipment based on sound science and engineering.

 

[Jud]

I agree. I might even try one of the lower cost 'specialist' ones that you have suggested. But I still think that the difference between an 18 inch one and a ten foot one might be far greater. Why do so many refuse to discuss it?

 

I don't know at what point any other distinctions in sonic quality between two given cables might be swamped by a length difference. The question hasn't really interested me simply because for reasons of both cost and sound quality, I buy the shortest cables that will connect two pieces of equipment, in this case my laptop and the DAC. Since I place my laptop right next to the DAC, I'm able to buy the shortest, least expensive version of any given cable. (These days for the cables I've tried that appears to be about .6m, or 2 feet.) Thus the question of how different a 10 foot cable might sound has never arisen.

 

[Julf]

Jud,

 

The buffer runout wouldn't necessarily be that clearly audible - after all, it is pretty much a lesser case of what happens every time a CD player can't read a block and has to "fake" it. It is a form of jitter, albeit a rather abrupt one.

 

All I am saying is that short of running out of buffer data, the (asynchronous) USB interface does not affect output jitter - but you still get output jitter based on the precision of timing in the DAC. And there I think we can find much more significant sources of jitter...

 

I would love to hear Gordon Rankin's views - and, once again, I am not saying I am totally ruling out the possibility of audible effects, but I would like to hear some scientifically valid arguments for it.

 

Julf

 

 

[Jud]

So I hope we can agree that the actual data values won't be changed by using a different USB cable? That leaves timing as the only possible variable

 

The answer to that question is worth examining closely, because I think it is easy to leave out something important.

 

- Agreed that the USB cable won't take this data stream:

 

00001000

 

- and change it to, for example:

 

00000000

 

- absent gross errors, which we can agree to leave out as irrelevant to this discussion.

 

But where the data *can* and *will* change with timing errors not so great as to cause dropouts, is in the reconstructed analog waveform. Two bitstreams of identical values that only vary in terms of timing (that is, one bitstream has more jitter than the other, though insufficient to cause dropouts) will when reconstructed give different analog waveforms. The one reconstructed from the bitstream with more jitter will have at least a higher noise floor (less effective dynamic range) and greater intermodulation distortion. There may be other effects I'm unaware of.

 

[Julf]

Likewise, I find your opinions very level-headed, informed and enjoyable. And if we didn't enjoy the discussion, I am sure we can find better things to do with our time

 

I have been involved in embedded communication systems in industrial settings, where control information is transmitted through plants and factories with electrical machinery, arc welders etc. - a somewhat rougher environment, in terms of electrical noise and RFI than the average listening room (one would hope!). Yes, the signal coming out of the other end doesn't look pretty when looked at with an oscilloscope or spectrum analyzer, but the data is still reconstructed perfectly - or flagged as transmission error.

 

There is of course a scenario where the cable affects the receiving DAC not by affecting the data, but by helping conduct RFI and electromagnetic noise into the DAC where it affects the internal operations in the DAC, but in that case the effects would be very much DAC dependent. That is something I miss in the debate. Mostly I hear "cable X is better than cable Y" or "cable X has an airier, but less crisp sound, with a slight hint of raspberry". I don't hear "cable X seems to make a big difference with DAC Y, but not with DAC Z".

 

Would you have an URL for the Plisson paper?

 

Julf

 

 

[Julf]

Totally in agreement there. But isn't that kind of what I stated - that timing is the only possible real variable. But there the issue is that with the asynchronous USB interface, the data gets re-clocked (so jitter gets reset to 0) on reception, and all jitter that you can see in the output is caused by DAC jitter introduced after the USB link.

 

[Julf]

This one definitely takes sides, but is still worth reading. Matrix-HiFi

 

[Jud]

That is why some of us (with an engineering/scientific background, or with a rational approach to things) are struggling with the concept of audible differences between USB cables. Employing Occam's Razor, it is unfortunately rather tempting to attribute the effects to placebo effects, marketing and snake oil. At the same time, I can see why the more existentialist/belief-based/touchy-feely ones amongst us point to their own subjective observations as proof that we bitheads are plain wrong.

 

That's not a very scientific observation. ;-)

 

I certainly want to get emotional enjoyment from my system, but I am not interested in any explanations for its sonic qualities that don't make scientific and engineering sense. I'd like to think my contributions here are arguments that we should in fact be more careful not to leave out important data when constructing our thought models; and that we should therefore not accept the oversimplified cartoon version of digital audio as some sort of ideal world of happy little 1s and 0s immune from physics.

 

Though I've tried to quite careful and thorough in 30 years or so of selecting components and cables for my various audio systems, it's not impossible that I could have been fooling myself. In fact there is at least one occasion on which I almost certainly have. (I heard improvements when I put noise reduction equipment into my system 20 years ago, and again when I took it out 2 years ago. :-)

 

However, there are reasons to think my subjective observations are not totally random or driven by marketing, and therefore have some value: the consistency of my subjective preferences for certain manufacturers' designs over others regardless of price; the consistency in many instances of those preferences with my wife's in the best blind testing I can manage for her; and the fact that I have never yet read any of the very fine audio engineers who participate in this forum (who don't work for cable companies) saying there are scientific and engineering reasons why different USB cables *cannot* sound different.

 

I think we are dealing with an issue that transcends USB cables, audio or technology

 

Oh, absolutely, one sees it all over the Web. People don't need years of education and experience, they *know* by "simple logic" that much of modern genetics (evolution), microbiology (vaccines), etc., is just plain wrong. One must have enough facts before it is safe to wield Occam's Razor.

 

[CG]

First, length does matter, both in speaker cables as well as data cables. It is not just a question of resistance, inductance, and capacitance, either, although that is a good basic approximation for the primary effects.

 

Try this... Somehow patch together a bunch of USB cables so that the overall length is 20 meters. The absolute length really doesn't matter; just make it long. Now, try running your DAC through that. How's it working? Try your printer. How's that one working?

 

In both cases, neither the source nor the load are perfect or ideal. The waveforms do matter, not just the levels that define zeros and ones. The source generates an (imperfect) time varying waveform that creates a time varying electromagnetic field in and around the cable. The (imperfect) receiver tries to make the best of what it gets. Little electron guys are not carrying signs that say either 1 or 0 down the cable.

 

But, rather than me trying to explain this - it would take too long, and nobody would believe me anyway - why not go read a publication by someone who is regarded as an expert in the field for this sort of thing?

 

http://www.amazon.com/Electromagnetic-Compatibility-Engineering-Henry-Ott/dp/0470189304

 

The link is for Amazon, but the book is also available elsewhere and is found in most technical libraries.

 

And yes, the printed circuit board materials and layout do make a difference. As do a lot of other factors in the associated circuitry. All the analog and digital components are far from ideal. These aren't just ideal black boxes on a block diagram as so many people, both in industry and elsewhere, like to treat them. The physics really does matter.

 

 

 

[russe41]

Thats the way I look at things. The sound is dependent on room,Speaker location(I've changed the sound with NOTABLE differece by changing speaker location),Equipment and Quality of the recording. I used to biamp with 2 cary v12's using nordost red dawn for the lower cabinet of my speakers and nordost blue heaven for the m/t cabinet and now use canare 4s11 and can't hear any difference. There for sale if anyone's interested.

BTW amp stands and fuses that change sound unless the stand is for a turntable are funny also.

 

[Jud]

all jitter that you can see in the output is caused by DAC jitter introduced after the USB link

 

That's certainly the design intent, and it is evidently from published measurements extremely effective.

 

But I simply want to be cautious about agreeing right now that there is absolutely no mechanism by which the USB link might conceivably affect jitter after the async USB interface. As I noted, I still wonder whether electrical and RFI effects transmitted by the cable might impact jitter at the output of the DAC's buffer.

 

Here is the citation to Damien's paper. You may well have seen it before:

 

http://www.amr-audio.co.uk/large_image/MAC%20OSX%20audio%20players%20&%20Integer%20Mode.pdf

 

[CG]

"I have been involved in embedded [and a bunch of other good thoughts...]"

 

In your embedded communications system, you are not converting the data into an analog format are? I think you may have already answered your own question in the next paragraph very well. Ever calculate what level of noise and other junk might be audible in an audio system with respect to the nominal 2 Vrms full scale output of a DAC box?

 

You do have a very good point about what is missing in people's reports of what they hear. Knowing the system context would be very helpful. The cable is hardly the only unknown in the equation.

 

Personally, I fail to understand why this is a debate at all. It's really just a bunch of wildly different personal observations that people want to argue over. Kind of like an old Monty Python sketch.

 

[Julf]

Yes, I had came across that paper - and it is a good read. It does point out the difference between sync and acync USB, and hints at the possibility of the USB cable itself acting as an antenna. So yes, I'm there with you on the point that there might be some mechanism whereby the USB cable affects output jitter - but just like with magic crystals, the Loch Ness Monster and Bigfoot, I would like to see a bit more firm evidence before becoming a believer

 

That's not to say that people who feel calmer in a room decorated according to feng shui giudelines, or feel refreshed by touching a magic crystal to their forehead are wrong per se, but I am not sure I would want them to design my computer or my audio system - and I would be very careful about buying extremely expensive products from them.

 

[Julf]

I can just hear John Cleese going "yesitisyesitisyesitis"!

 

[Julf]

I don't think anyone here is arguing physics doesn't matter. I guess all I am arguing is that those physical effects can, and should, be measured. And we need to get better at measuring them. Once you can measure it, it gets easier to improve the product (as well as avoiding spending money on snake oil).

 

[Julf]

I agree - let's try to keep an open mind, while approaching things rationally and based on technical knowledge. That is what I would expect on a forum for *computer* audio. We all know our audiophile hobby has gotten a bad reputation among non-audiophiles because of some of the snake-oily excesses in the business.

 

 

[CG]

There is snake oily excesses in every business I have ever seen. Every single one. Audio is just easy for some to pick on, because it's seemingly so simple and obvious to anybody who is at least partially sentient.

 

It's just like cooling and air flow. Everybody knows everything on the subject because they think it's easy and they have some innate experience with it.

 

Beyond that, the truly cruel part is that based on how scientists understand our auditory systems, everybody is somewhat different in how they perceive and interpret sound. Music especially so, since there is again another level of processing that takes place for that. So, what is important or evident to one person is likely different for the next person.

 

People have no problem comparing wine, coffee, chocolate, food, and a whole bunch of other products mostly on a subjective basis. In every one of the cases just mentioned, it is entirely possible to analyze the chemical and physical content of each. Yet, I don't see that many web sites that proclaim that all coffee tastes the same, and that a full set of measurements is needed of the composition to determine whether it is worth drinking the stuff. People just drink it and say they either like it or don't like it and what they do or don't like.

 

Why audio is subject to a different standard is hard to explain. Perhaps the usual sorts of people who are attracted to the hobby, or at least a significant subset, just happen to be wired that way. I think that photography may be similar in this regard.

 

Besides, everybody likes to pick on people who they think may be geeks. Even other geeks like to join in the melee.

 

Back to the topic. According to the Amazon web site, nobody has jumped to order the hard cover edition of Ott's book...