The USB cable myth

[Stevii]

Hi

I could not get my head around the fact why a USB cable should be of any influence.This is a shorter, translated version of a dutch article I found. See for the source: http://www.alpha-audio.nl/2011/06/de-usb-kabelmythe/

 

..

 

USB transfer modes

 

isochronous (asynchronous), interrupt, bulk and control.

 

Let's start with the most common: bulk. This method is used for data transfer. The bandwidth is not guaranteed, but that's not important. The bus is trying to get it to the highest throughput. With this transfer mode the host controller and the client (the USB drive / stick) perform CRC checks. When the CRC does not match, the data packet is transmitted again. It can not go wrong.

 

The interrupt mode is for machines that have a low 'latency' require: quick response. Think of a mouse. The control mode is used to initialize a device.

 

Finally, the isochronous transfer mode. This is used for audio and video streaming, as USB dacs do. The main advantage of this mode is the guaranteed bandwidth on the bus. This rate is agreed between the controller in the PC and the USB receiver in the DAC. If the speed is not feasible for some reason - too many other devices on the bus, or a bad cable - it will not work.

 

Control

 

In bulk mode (data transfer) crc packets are transmitted. This ensures that errors in transmission are not possible. Or the PC gives an error. In isochronous mode is a crc-check is performed, but the controller does not send a packet again if something goes wrong. The package is therefore 'dropped'. Is that bad? no, it's not audible, as it is a constant flow of data and we talk about a millisecond. But is there a chance a packet gets lost in the first place? More on that later.

 

How is the flow of data is now governed by usb? Very different - much easier and more controlled - than spdif. Every millisecond (1000 Hz) The host sends a USB SOF (start of frame) token. Using this package, with a unique 11bit ID, the receiving side can Synchronize the data. Both sides know exactly what was sent and where the package belongs in the chain.

 

Testing

But, skeptics will say, in isochronous mode is no-crc correction. No, that's right. In theory, data loss is possible. To find out how much we did a test.

 

The following software tool shows the data over a USB connection: http://www.usblyzer.com/download.htm. And it shows if packets arrive, crc-checks are correct, etc.

 

U.S. 300 euro 5 euro

We purchased a AudioQuest Coffee USB cable, sized 1, 5 meters (ca. 300 euros). Yes: we too were convinced of the usefulness of a high-end USB cable. The AudioQuest entered the fight with a standard cable of 5 euros. This of course meets the USB 2.0 standard.

 

We played a complete song by Massive Attack, twice. Once over the AudioQuest Coffee and once over the budget cable.

 

In both cases, error free data is transferred. A song of more than 5 minutes, more than 300,000 milliseconds ... error free in both cases.

 

Logical

And that is very logical when you consider that there is no other information going from a to b then data and synchronization data. There are packets of data going through the cable, supplemented with an ID token which indicates where the packet belongs. The receiver chip in the DAC creates spdif.There it can go wrong: not with the USB transfer.

 

Conclusion

Do not be fooled by expensive USB cables. It's a waste of money. We have believed, but are now convinced of the nonsense. with a thorough digital protocol it is simply impossible to ruin the data transfer. The cable is broken or simply does not meet the standard.

[PeterSt]

The package is therefore 'dropped'. Is that bad? no, it's not audible, as it is a constant flow of data and we talk about a millisecond.

 

Of course this is audible, especially when the drop out would last one millisecond (a crazy long ltime btw);

When so much data is lost, chances are the most fair that two subsequent samples which made it (with the gap in between them) jump so much for voltage that a loud click can be heard.

Of course it entirely depends on the music (samples) currently playing whether that voltage jump is large or (inaudibly) small.

 

Peter

[Stevii]

Hi Peter,

 

You miss the point of this article: nothing gets dropped in the first place.

 

stefan

[Richard Dale]

Hi Peter,

 

You miss the point of this article: nothing gets dropped in the first place.

 

stefan

 

But nobody has been claiming that USB cables sound different because of packets being dropped, and so the whole article is just attacking a straw man.

[Stevii]

But nobody has been claiming that USB cables sound different because of packets being dropped, and so the whole article is just attacking a straw man.

 

Ok, so when digital data is transported 100%, The only way to alter the sound is to deliberately altering the packets.

This is not preferred or is it?

 

stefan

[souptin]

If I have understood the article extract posted above, the dropping and re- transmitting of packets is not the main point being addressed.

 

To me the research is trying to verify that both cables tested are able to transmit bit perfect data in isochronus mode (aka asynchronous USB, as is popular with recent DACs). And that their answer is yes - the cheap cable transmits the exact same data as the expensive one. So far so good...

 

Now the "consensus" seems to be that bit perfect data transfer to the dac stage is an essential starting point for high end audio, but that it is only a starting point, nothing more. So if you are hearing differences between digital cables and playback software, for example, you need to look further than just confirming that all the bits have arrived in good order.

 

I say "consensus" with quotes for added irony, it's a controversial topic.

[Stevii]

Now the "consensus" seems to be that bit perfect data transfer to the dac stage is an essential starting point for high end audio, but that it is only a starting point, nothing more. So if you are hearing differences between digital cables and playback software, for example, you need to look further than just confirming that all the bits have arrived in good order.

 

Only the usb cable is addressed, so bit perfect transmission is also where it ends. The only way to alter the sound in a bit perfect transmission is to alter the packets deliberately.

[Richard Dale]

Only the usb cable is addressed, so bit perfect transmission is also where it ends. The only way to alter the sound in a bit perfect transmission is to alter the packets deliberately.

 

No it isn't because in audio applications the timing of when the bits arrive is important, which is not the case with USB bulk mode transmission. Please search for the many discussions about USB cables on this forum and read them if you're are interested. Of course not everyone agrees whether or not USB cables sound different when used in audio systems, but those people who think they sound different don't think it is because of the bits.

[Stevii]

No it isn't because in audio applications the timing of when the bits arrive is important, which is not the case with USB bulk mode transmission.

 

The timing is very important I totally agree, but that is done by the Host/client. The USB cable has nothing to do with the timing of the bits, only with the bits themselves

[Julf]

The timing is very important I totally agree, but that is done by the Host/client. The USB cable has nothing to do with the timing of the bits, only with the bits themselves

 

I think that there are two main theories as to why (if at all) a USB cable itself could affect sound - one is that the electrical characteristics (frequency-dependent behaviour, reflections and possible nonlinearity) affect jitter (easily verified with the eye pattern test), and the other that the cable conducts earth loops, HF energy and EMT between the computer and the DAC, thus increasing jitter.

[Stevii]

I think that there are two main theories as to why (if at all) a USB cable itself could affect sound - one is that the electrical characteristics (frequency-dependent behaviour, reflections and possible nonlinearity) affect jitter (easily verified with the eye pattern test), and the other that the cable conducts earth loops, HF energy and EMT between the computer and the DAC, thus increasing jitter.

 

The USB protocol (unlike SPDIF) does not have a single problem with jitter. In Asynchronous mode, the audio device does the clocking. It tells the PC to regulate its speed. As the clock is running on the audio device this solution can be implemented in such a way that the the DAC receives a signal with a very low jitter.

[Julf]

The USB protocol (unlike SPDIF) does not have a single problem with jitter. In Asynchronous mode, the audio device does the clocking. It tells the PC to regulate its speed. As the clock is running on the audio device this solution can be implemented in such a way that the the DAC receives a signal with a very low jitter.

 

I agree - but a lot of USB implementations out there are not (yet) asynchronous. Not that it makes any difference if the DAC does propeer buffering.

 

I think the key is that the solution *can* be implemented so that you have very low jitter - but not all systems are that well designed.

[jerry_rig]

I'm not sure this is helpful, but at least some of us who have experimented with different USB cables can hear a difference. I too was a sceptic until I tried it myself. Strangely, in my system, I preferred the cheaper (not cheapest) cable.

 

Since the authors of the article in the post above can not measure any difference, they claim none exists. This is an old issue in the audio world. It often simply means they are measuring the wrong thing. Why beat a dead horse?

[wgscott]

Strangely, in my system, I preferred the cheaper (not cheapest) cable.

 

That at least adds credibility to your observation, since presumably it is counter to expectation-bias.

 

Since the authors of the article in the post above can not measure any difference, they claim none exists.

 

Here's the thing: If you can measure a difference, it at least becomes plausible that there really is a difference that might (or might not) be audible.

 

If you can't measure a difference, despite trying to measure all the right things, then it places an additional "burden of proof" on the claim that there are audible differences.

[jkortela]

With Cardas Clear Serial Buss USB, the sound is almost perfect. I compared the Cardas Clear Serial Buss (USB) -> WaveLink HS 24/192 USB to SPDIF converter -> Cardas SPDIF connected to digital in of PS Audio Perfectwave DAC Mark II compared with "bridge" and sound is almost identical.

 

With Belgin Gold USB or Audioquest Cinnamon USB (almost identical), the sound has really clear magnetic echo in it and you loose a lot from a music.

[Paul R]

Deleted

[Jud]

The timing is very important I totally agree, but that is done by the Host/client. The USB cable has nothing to do with the timing of the bits, only with the bits themselves

 

Just wondering if you know anything regarding how jitter might be affected by frequency effects in the cable potentially altering the zero-crossing points of the waves representing the bits.

[Jud]

the cable conducts earth loops, HF energy and EMT between the computer and the DAC, thus increasing jitter.

The USB protocol (unlike SPDIF) does not have a single problem with jitter. In Asynchronous mode, the audio device does the clocking. It tells the PC to regulate its speed. As the clock is running on the audio device this solution can be implemented in such a way that the the DAC receives a signal with a very low jitter.

 

Your response does not account for the effects Julf described. First, if electrical noise is brought into the DAC clock, the DAC clock itself becomes less accurate. Since now your timing source, the DAC clock, is less accurate, that increases jitter. Second, electrical noise can be conducted by the USB cable through the ground plane into the analog circuitry of the audio system, where it appears as plain old analog electrical noise.

 

Electrical noise can be eliminated by an optical cable, but now you are not dealing with an async USB interface.

[esldude]

Maybe we should also start talking about the nocebo effect as well as the placebo effect.

 

This is when you are given something that you think makes matters worse, and end up perceiving a worsening of sound, when perhaps nothing is changing the sound. Maybe it plays into thinking software load on PC's adversely effects sound quality, or that cheaper USB cables are causing harm to sound, or that PS noise regularly causes problems in the resulting sound output by a DAC connected via USB. Lots of things that could conceivably make some difference, some that even causes measurable differences in some areas, but that haven't been shown to actually change the sound coming out of the system.

[sandyk]

"or that PS noise regularly causes problems in the resulting sound output by a DAC connected via USB. "

esldude

PSU noise can sometimes be heard via coax SPDIF to a DAC as well.For example, if you are listening to music while your email client is open, and the email client does it's programmed checking for new emails, you may sometimes hear little clicks when this is happening. Part of the problem here may be due to many ATX PSUs having one very high current capability +12V supply that is shared by everything. ( +12V 18A to 25A, and higher.)

Usually there are no separate regulated +12V supplies, although occasionally there may be 2 +12V supplies.. Looking at a schematic of a 300W ATX PSU, as an example,(Google Images) you may find this VERY high current supply goes through a choke , bypassed by a single 1,500uF capacitor at the output..

Regards

Alex

[esldude]

Yeah, I couldn't agree or disagree with you Alex. I did say USB connected computers, as I haven't spent all that much time with digital out from sound cards in the past year. Of course using analog out (even though I have done only a miniscule amount of that) one hears lots of stuff get through like moving a mouse or cellphone interference, so I know such stuff is there.

 

I have used a desktop I have built with good PSU , two netbooks, an old Dell laptop, a Shuttle (miniATX) and very old Dell desktop (PIII) with a USB connection. Never have I heard anything get through the USB connections that were simply spurious noise. I have heard drop outs on some of those when using a Firewire connection. I have heard just literally only a handful of dropouts on USB. But noise of other things getting through doesn't seem to be a common issue. Whether it is perfect or not using an asynch USB to SPDIF connection seems to go a long way toward isolating junk from the PC getting into the rest of your audio system.

[Musicophile]

Maybe we should also start talking about the nocebo effect as well as the placebo effect.

 

This is when you are given something that you think makes matters worse, and end up perceiving a worsening of sound, when perhaps nothing is changing the sound. Maybe it plays into thinking software load on PC's adversely effects sound quality, or that cheaper USB cables are causing harm to sound, or that PS noise regularly causes problems in the resulting sound output by a DAC connected via USB. Lots of things that could conceivably make some difference, some that even causes measurable differences in some areas, but that haven't been shown to actually change the sound coming out of the system.

 

I think the nocebo effect has got it right. I auditioned the other day the Audioquest Carbon against a generic printer cable. I somehow believed I heard something, but the difference was to weak to justify the >$300 price tag that came with it (i need 5M, which makes me additionally nervous). So I'll probably settle for the Audioquest Cinnamon in the next days, as

 

1) I like the looks (psychologically important),

2) the price is about 10% the price of the DAC, which sounds like a healthy ratio

3) I simply couldn't stand the idea of a generic printer cable in my nice Hifi system

4) My dealer who I very much appreciate and who's been very supportive in a lot of listening sessions and lending equipment to take home makes some money

[PeterSt]

"or that PS noise regularly causes problems in the resulting sound output by a DAC connected via USB. "

esldude

PSU noise can sometimes be heard via coax SPDIF to a DAC as well.For example, if you are listening to music while your email client is open, and the email client does it's programmed checking for new emails, you may sometimes hear little clicks when this is happening. Part of the problem here may be due to many ATX PSUs having one very high current capability +12V supply that is shared by everything. ( +12V 18A to 25A, and higher.)

Usually there are no separate regulated +12V supplies, although occasionally there may be 2 +12V supplies.. Looking at a schematic of a 300W ATX PSU, as an example,(Google Images) you may find this VERY high current supply goes through a choke , bypassed by a single 1,500uF capacitor at the output..

Regards

Alex

 

Alex, I am sorry, but this is not correct. Plausible in your eyes, I'm sure, but this is just missing samples. That they may be caused indirectly by a poor PSU (setup) is another matter, but what you hear is missing samples.

This, despite what was claimed earlier on in the thread - that no missing samples will ever occur. They will, easily (and with all the connection means mentioned). So, as long as missing samples can occur (the most easy to prove with the proper tools) that comes first for a reason of "clicks". Only when it can be proved that no samples are missing, then your idea is in order.

 

Peter

[sandyk]

Alex, I am sorry, but this is not correct. Plausible in your eyes, I'm sure, but this is just missing samples

 

Peter

I have found that when a JLH PSU addon is fitted in line with an internal LG BR writer, that most likely due to it's very low impedance and high simulated capacitance, that noise/ missing samples is much more obvious if the email client is open and does it's regular check for new mail. That is why I mentioned the very poor output filtering in typical ATX PSUs.

There is clearly some PSU interaction involved here. I have been thinking of fitting a 100uH choke with a DC resistance of 50milli ohms in series with the JLH's input in order to try and reduce this effect.Either that or switch off the DC input to the JLH PSU when listening to music.

I have on one occasion tried recording 24/96 via loopback, and some program has caused missing samples as seen in the .wav files using Sound Forge 9.

Alex

[Jud]

Maybe we should also start talking about the nocebo effect as well as the placebo effect.

 

This is when you are given something that you think makes matters worse, and end up perceiving a worsening of sound, when perhaps nothing is changing the sound. Maybe it plays into thinking software load on PC's adversely effects sound quality, or that cheaper USB cables are causing harm to sound, or that PS noise regularly causes problems in the resulting sound output by a DAC connected via USB. Lots of things that could conceivably make some difference, some that even causes measurable differences in some areas, but that haven't been shown to actually change the sound coming out of the system.

 

Before we get to Ptolemaic epicycle levels of complexity with placebos, nocebos (semi-cebos? para-cebos?) and the like, we may want to think about the possibility of simply believing what people are hearing.

 

No doubt people's senses are affected by appearances. I don't question the applicability of studies done with wine labeling, for example, to judgments of audio equipment sound quality. But when the conclusions of subjective evaluation don't fit neatly into those sorts of boxes - when subjective impressions of sound quality don't smoothly correlate with cost either positively or negatively, for example - then instead of designing ever more complex boxes based on "placebo effect," we just may want to have a look at real differences as a possible source of the reported distinctions in sound quality.