Gordon Rankin Says I'm Wrong About USB Cable Sound!

[Blake]

[/color]I've been down the "neurotic audiophile" route, and have come out the other side. I do not have any intention of having that monkey on my back again. All that money spent chasing an elusive goal that can never be attained.

 

Well said George.

[kumakuma]

I'll go you one better. Understanding the mechanism better, now, I'd be willing to bet money that the high-resolution files that get downloaded from HDTracks and equivalent sites sound different from the same files actually stored on the supplier's servers. That thought ought to get the neurotic juices flowing!

 

Why would this be the case? You aren't playing the files while downloading. You are downloading and then playing back the files locally. Both versions should be identical although I know there are some who hear differences between bit identical files.

[sandyk]

Why would this be the case? You aren't playing the files while downloading. You are downloading and then playing back the files locally. Both versions should be identical although I know there are some who hear differences between bit identical files.

 

There are also some very experienced Recording Engineers that hear differences between discs manufactured at different disc replication plants, and prefer those where the master is created at a lower speed.

[wgscott]

I wonder why it is that I can use various cheap USB-cables that came with various USB-equipment to connect my Windows 7 laptop to a USB(2)-harddisk and send terabytes of data over a period of 3+ hours, and not have a single error occurring?

 

Well, if OS X suffers one transmission error per month, that is 0.00000038580247 errors/second. If Windows really is twelve percent worse, that means it suffers an error rate of 0.000043209877 errors/second (or a .155% chance of suffering an error during an hour of listening).

[wgscott]

There are also some very experienced Recording Engineers that hear differences between discs manufactured at different disc replication plants, and prefer those where the master is created at a lower speed.

 

Are those discs made of vinyl?

[kumakuma]

There are also some very experienced Recording Engineers that hear differences between discs manufactured at different disc replication plants, and prefer those where the master is created at a lower speed.

 

This is interesting, quite possible, and totally irrelevant to the question I asked.

[sandyk]

This is interesting, quite possible, and totally irrelevant to the question I asked.

 

Do you REALLY expect a definitive answer to a question like that, when even the "experts" aren't in complete agreement about more mundane topics such as the title of the thread?

[kumakuma]

Do you REALLY expect a definitive answer to a question like that, when even the "experts" aren't in complete agreement about more mundane topics such as the title of the thread?

 

Yes, I did. I was interested in why George thought possible differences in Ethernet cables would cause downloaded data to sound differently. I am willing to keep an open mind about different cables causing streaming data to sound different but the idea that cable quality could cause downloaded data to change is hard to fathom.

[gmgraves]

Why would this be the case? You aren't playing the files while downloading. You are downloading and then playing back the files locally. Both versions should be identical although I know there are some who hear differences between bit identical files.

 

I was being somewhat facetious about the HDTracks downloads. But, if a USB cable can't transmit a streaming music file a meter or so without changing it, then What would thousands of miles of Internet connection do to say, a streaming performance of the Boston Symphony or the New York Philharmonic?

[Jud]

Yes, I did. I was interested in why George thought possible differences in Ethernet cables would cause downloaded data to sound differently. I am willing to keep an open mind about different cables causing streaming data to sound different but the idea that cable quality could cause downloaded data to change is hard to fathom.

 

So we probably have to look elsewhere than "caus[ing]...data to change." Think of John Swenson's explanation for how USB cables affect the sound - no bits are flipped, but cable qualities cause the DAC's USB receiver chip to work harder with some cables than others, resulting in noise that slightly changes the ground-to-signal comparison timing and amplitude in the DAC chip, causing jitter and thus distortion. Or you could just get some cables transmitting more electrical noise than others; or it could be Gordon Rankin's explanation of impedance mismatches.... Notice none of these involves bit flipping (changing data), and the effects are often felt not in the cable itself but in the DAC chip way down the line.

 

Now why it should surprise us that this stuff isn't intuitively obvious to us non-audio engineers, I don't know. For my own part, I don't expect to be able to understand the way various audio circuits work properly, let alone the sundry ways distortion can sneak past more or less knowledgeable engineers' best efforts.

[thesurfingalien]

There are also some very experienced Recording Engineers that hear differences between discs manufactured at different disc replication plants, and prefer those where the master is created at a lower speed.

 

Hi Alex,

 

I believe I have mentioned this before (more than a year ago I guess). I believe these differences could be related to the tracking-mechanism in a CD-player. If more position-corrections need to be made by the tracking-mechanism to accurately read the bits of a less accurate pressing, is it not possible that this causes distortion in the analogue part of the CD-player?

 

Kind regards,

Peter

[kumakuma]

I was being somewhat facetious about the HDTracks downloads. But, if a USB cable can't transmit a streaming music file a meter or so without changing it, then What would thousands of miles of Internet connection do to say, a streaming performance of the Boston Symphony or the New York Philharmonic?

 

There are protocols such as RTP (Real-time Transport Protocol - Wikipedia, the free encyclopedia) that deal with streaming data to ensure that the quality is as high as possible although there may be glitches just as there may be glitches when data is pulled from a CD when it is played.

 

There are other protocols such as TCP that are more reliable and ensure that received packets match sent packets. If they don't, then they are resent until they do.

 

Note that Internet protocols for sending data *assume* that data will be corrupted in transit and packets will arrive in the wrong order.

[kumakuma]

Hi Alex,

 

I believe I have mentioned this before (more than a year ago I guess). I believe these differences could be related to the tracking-mechanism in a CD-player. If more position-corrections need to be made by the tracking-mechanism to accurately read the bits of a less accurate pressing, is it not possible that this causes distortion in the analogue part of the CD-player?

 

Kind regards,

Peter

 

This is probably also why the ripping speed slows down when you try to rip certain CDs.

[sandyk]

Hi Alex,

 

I believe I have mentioned this before (more than a year ago I guess). I believe these differences could be related to the tracking-mechanism in a CD-player. If more position-corrections need to be made by the tracking-mechanism to accurately read the bits of a less accurate pressing, is it not possible that this causes distortion in the analogue part of the CD-player?

 

Kind regards,

Peter

 

Hi Peter

Yes, interaction between the current draw of the laser mechanism and the rest of the PSU may account for some of this.

However, it still doesn't explain why very high quality discs such as those burned to MAM Gold CD-Rs are able to be ripped by EAC at around double the speed of a generic CD-R AND still sound a little better when played at normal speed, even sounding slightly better after being ripped, and showing the same checksums, just like with BluSpec comparison CD sets. The same observations have been made when the player's analogue output is bypassed, and Coax SPDIF to a high quality DAC is used.

Many people have also reported that their burned copies sound better than the original CD when high quality CD-R blanks are used. Perhaps in that case, the original master may have been used for longer than it should have been ?

Kind Regards

Alex

[sandyk]

This is probably also why the ripping speed slows down when you try to rip certain CDs.

I think that the actual polymer used, and the quality of the reflective layer plays a part here too, as claimed by several Japanese companies.

[Stereolab42]

So just use one of the interfaces that use bulk mode transfers instead of isochronous ones...

 

I never liked USB Audio Class, it is over complex while at the same time having stupid design for the transfer model (1 ms packet rate with variable packet size based on async feedback).

 

Just for fun, compare source code of Linux UAC driver vs hiFace driver.

 

P.S. Or even better, use PCIexpress or Thunderbolt interfaces and gain direct busmaster DMA without wasting CPU time building special audio transfer packets.

 

I suppose a DAC manufacturer could do something like write a custom USB driver that presents itself as a streaming audio driver to Windows, but on the backend use the USB bulk protocol to communicate with the DAC. Thus getting error correction for "free". Or heck, even develop their own protocol, since they control both endpoints. But why stop there? Let's have the USB "driver" instead communicate with the DAC on the backend using TCP/IP over Ethernet.

[beanbag]

Whether Mr Rankin is right or wrong, this illustrates a fundamental flaw which frequently arises in the objectivists' logic. They assume that they have a logical model which completely explains a phenomenon and reject any observations which do not quantitatively agree with their model, then someone comes along with a different, but equally valid model and their argument goes out the window.

 

This is a misrepresentation on the "objectivist" position, whoever these people are.

They do not go around "rejecting any observations which do not quantitatively agree with their model", they go around rejecting observations that are faulty and unverifiable. Maybe that is where the confusion lies.

 

This is an example of a faulty observation: "I listened to these two cables, which I swapped out myself, and heard a difference. My friend was there too, and I asked him, 'you notice how the veil was lifted when I swapped these cables?', and he was like 'yeah, I guess'."

 

Valid evidence / observations would be something like a measurement result that could be verified by others, and / or an ABX test to check for audibility. Hey, it's what the AES does. BTW, I dunno why some people have such a phobia to this. It's not like there is some guy out to "get you", looking over your shoulder and giving you "performance anxiety" (LOL). And then after you fail the test in shame, he smirks at you with a mixed look of condescension and pity.

 

Yes, there are different rigorous ways to do the test, but fundamentally, it is something that you can do yourself, in the comfort of your own home, with nobody watching, and you can switch back and forth as many times as you want and take as long as you want. Goal is to remove placebo effect and other user biases. Main drawback is that it can be a PITA to perform if hardware is involved.

 

Objectivists (whoever these people are) are only interested in reality, and not made up impressions.

 

Oh, and also Laissez-faire capatalism.

 

Oh, and also blowing up ugly buildings.

[sandyk]

They do not go around "rejecting any observations which do not quantitatively agree with their model", they go around rejecting observations that are faulty and unverifiable.

 

Just because they are presently unverifiable, does NOT mean they are faulty !

 

Objectivists (whoever these people are) are only interested in reality, and not made up impressions.

 

And who decides the impressions are made up ? Newbies like yourself , and EEs who rely solely on what they were taught at Uni many years earlier, and have since specialised in other areas, and not updated their knowledge in other areas since then ?

[Snowmonkey]

This is a misrepresentation on the "objectivist" position, whoever these people are.

They do not go around "rejecting any observations which do not quantitatively agree with their model", they go around rejecting observations that are faulty and unverifiable. Maybe that is where the confusion lies.

Are you familiar with the logical fallacy of begging the question? This is an informal fallacy where the conclusion that one is attempting to prove is included in the initial premises of an argument, often in an indirect way that conceals this fact. (Wikipedia).

For example: "Objectivists go around rejecting observations which are faulty or unverifiable".

How do they know the observations are faulty? - because they are inconsistent with their model of reality.

How do they know their model of reality is true? Because all the non-faulty observations confirm it.

 

As to rejecting observations which is unverifiable, this is the fallacy of argumentum e silentio – where the conclusion is based on the absence of evidence, rather than the existence of evidence. For example (pre 2012): The Higgs Boson does not exist, because its existence is unverifiable.

 

Objectivists (whoever these people are) are only interested in reality, and not made up impressions.

Definition of reality: those things which conform to their model of reality.

Definition of made up impressions: everything else.

[acg]

Mea Culpa, I guess. I just had a nice discussion with Gordon Rankin of Wavelength. Gordon is the man who designed the Ayre QB-9 and the AudioQuest DragonFly DAC, as well as the WaveStream® IsoSynchronous USB protocol. He said a number of things that really took me aback. The first thing he said was that yes, USB cables do sound different. They shouldn't but they do. He says the reason is because most manufacturers are unable or unwilling to make their cables with the correct 90 Ohm impedance and be able to pass up to a 2.6 GHz perfect square wave. He says that unless those two parameters are met, the USB feedback mechanism doesn't work because reflections get the signal confused on the computer end. Without it, there's no way for the computer to know that it's sending the data at the proper rate and bits get dropped. Since USB receivers have no error correction (this was news to me) a dropped bit gets interpolated at the DAC, and distortion increases. He says that this situation is not helped by the fact that Windows, including Win 8, generate 12% more hard errors in a digital audio bitstream than does Apple's OSX! He also says to play your computer audio directly from memory for lowest error rate on either platform, and if possible, keep your music on a Firewire, eSATA or Thunderbolt drive rather than USB.

So, If you want your computer sourced digital audio system to sound it's best, find a cable manufacturer that will guarantee his cable's impedance at 90 Ohms* and that his cable can pass, ideally, a 2.6 GHz, or at least a 0.5 GHz perfect square wave (seems like a pretty high frequency to me), and that your music server runs OSX rather than Windows.

 

*Gordon also told me that virtually NONE of the cable manufacturers out there have any way of measuring the impedance of their cables and many believe that a lower than 90 Ohms is preferable because it allows them to sell longer runs. He says that because any deviation from 90 Ohms (higher or lower) is damaging to the data stream, lowering the impedance of the cable doesn't work. Also, an apps engineer at Belden just told me that most cable companies buy their actual wire from suppliers who will customize things like sheath color and dielectric material, and it's possible that many suppliers supply essentially the same cable to multiple audio cable makers.

 

Hi George,

 

Interesting topic, I've just found it and now jump into the discussion four pages late.

 

My understanding of the USB Audio protocol is that there is no error correction but some form of error detection at the dac side. Also, I understand that if a dac is immune to USB cables, that it is not necessarily immune to electrical noise from the computer, which I experienced with the Phasure NOS1 which has been immune to cable for a few years now, but not immune to PC noise. The new Phasure NOS1a is now immune to both USB cable and computer noise, and is the only dac that I am aware of that can make this claim.

 

PeterSt, the developer of the NOS1a, has made observations regarding USB Audio drivers in windows and their level of "bit-perfectness" which he has not had time to tease out in full, but there seems to be something going on in the o/s that is not what it should be. Just how this ties into what Gordon Rankin has said to you I have no idea, but PeterSt got USB cable immunity in the NOS1 sorted with proper grounding (among other things) and the solution was not reliant on the computer end of things as Gordon Rankin suggests. However, immunity to USB cable and computer noise do seem to be two slightly different beasts, each with their own solution.

 

Cheers,

 

Anthony

[Allan F]

Are you familiar with the logical fallacy of begging the question? This is an informal fallacy where the conclusion that one is attempting to prove is included in the initial premises of an argument, often in an indirect way that conceals this fact. (Wikipedia).

 

As to rejecting observations which is unverifiable, this is the fallacy of argumentum e silentio – where the conclusion is based on the absence of evidence, rather than the existence of evidence. For example (pre 2012): The Higgs Boson does not exist, because its existence is unverifiable.

 

Of course, you are right. But surely I am not alone in finding more than slightly tiresome the appearance of another evangelistic objectivist newbie on a mission to convert and save us from our ignorance. Thank God for the Ignore List!

[One and a half]

There are protocols such as RTP (Real-time Transport Protocol - Wikipedia, the free encyclopedia) that deal with streaming data to ensure that the quality is as high as possible although there may be glitches just as there may be glitches when data is pulled from a CD when it is played.

 

There are other protocols such as TCP that are more reliable and ensure that received packets match sent packets. If they don't, then they are resent until they do.

 

Note that Internet protocols for sending data *assume* that data will be corrupted in transit and packets will arrive in the wrong order.

 

Shhh, don't talk about Real time Protocols, get your head chewed off, see here.

[Paul R]

I was being somewhat facetious about the HDTracks downloads. But, if a USB cable can't transmit a streaming music file a meter or so without changing it, then What would thousands of miles of Internet connection do to say, a streaming performance of the Boston Symphony or the New York Philharmonic?

 

Not at all the same thing George. Totally different transfer protocol. There is what we call streaming but is really a TCP/IP data transfer, and then there is USB...

 

Paul

[manisandher]

Now why it should surprise us that this stuff isn't intuitively obvious to us non-audio engineers, I don't know. For my own part, I don't expect to be able to understand the way various audio circuits work properly, let alone the sundry ways distortion can sneak past more or less knowledgeable engineers' best efforts.

 

I recently asked Pat at AR-T what causes sound quality differences in USB. He said he didn't know for sure. So I guess no one does.

 

Mani.

[manisandher]

... immunity to USB cable and computer noise do seem to be two slightly different beasts, each with their own solution.

 

Thanks Anthony.

 

Mani.