A USB Cable Test Designed to Satisfy Skeptics AND Believers—Feedback Welcome

[solderdude]

It was not audible.

So it would be interesting to find out just what error rate becomes audible.

In 1987 I had a Technics CDP (SLPJ33 modified with SLP500 filters and DAC chip) and on its processor it had C1, C2 and CU error flags.

Connected a monostable to all 3 outputs with 0.1s time prolonging + 3 LEDs.

I found that all CD had plenty of C1 errors (which are fully correctable).
Sometime C2 errors with some CD's (which should be correctable but indicate poor discs) and with some damaged discs there were CU errors.
These are not correctable and the CDP reacted with a sample hold.
It did not loose track though and just a few CU errors were completely inaudible.
With a boatload of errors you could hear it but in most cases the CD hung or skipped as the errors became too severe and started loosing track(or focus or both?)

 

[danadam]

I thought the same, but a single sample error seems not to be audible. I edited a song, found a sample that was close to full positive, and changed it to full negative. I thought it would manifest as a click or tick sound.

It was not audible.

I did something similar. Here are bitflips:

Intel CPUs Don’t Support ECC Memory: How Bad For A/V Quality?

This discussion may only serve to reveal my less than complete knowledge of A/V digital signal processing rather than raise questions which audibly or not impact signal quality. However, the concerns raised stem from here. https://www.google.com/amp/s/arstechnica.com/gadgets/2021/01/linus-torvald...

audiophilestyle.com

and here are zero-ed samples:

Time resolution of digital sampling

audiophilestyle.com

("musiclong.flac" is original and "musiclongmissing.flac" is with some zero-ed samples)

The change is not necessarily audible in every instance a change was made, but it definitely is audible in some.

Also, AFAIK, USB 2.0 transfers one packet every 1/8000 s, so in case of 44.1 kHz sampling rate a single packet will contain 5 or 6 stereo samples. When CRC check fails, the whole packet will be dropped, so the "hole" in the audio data will be at least that and not just a single sample. Here's an example what it could sound like if 0.1% of packets were lost this way:

2.04 MB file on MEGA

mega.nz

 

[antcollinet]

I did something similar. Here are bitflips:

Intel CPUs Don’t Support ECC Memory: How Bad For A/V Quality?

This discussion may only serve to reveal my less than complete knowledge of A/V digital signal processing rather than raise questions which audibly or not impact signal quality. However, the concerns raised stem from here. https://www.google.com/amp/s/arstechnica.com/gadgets/2021/01/linus-torvald...

audiophilestyle.com

and here are zero-ed samples:

Time resolution of digital sampling

audiophilestyle.com

("musiclong.flac" is original and "musiclongmissing.flac" is with some zero-ed samples)

The change is not necessarily audible in every instance a change was made, but it definitely is audible in some.

Also, AFAIK, USB 2.0 transfers one packet every 1/8000 s, so in case of 44.1 kHz sampling rate a single packet will contain 5 or 6 stereo samples. When CRC check fails, the whole packet will be dropped, so the "hole" in the audio data will be at least that and not just a single sample. Here's an example what it could sound like if 0.1% of packets were lost this way:

2.04 MB file on MEGA

mega.nz

Thanks for that - that last sample sounded exactly like what I got when I was using a long toslink connection to ??? something, can't remember what - and I turned up to a sample rate the connection couldn't deliver reliably. Except the error rate gradually ramped up until there was more noise than music. Of course - toslinik/spdif uses multi sample frames also.

 

[SanojS]

I thought the same, but a single sample error seems not to be audible. I edited a song, found a sample that was close to full positive, and changed it to full negative. I thought it would manifest as a click or tick sound.

It was not audible.

So it would be interesting to find out just what error rate becomes audible.

That's not a signal error, you changed one bit.

 

[SanojS]

I did something similar. Here are bitflips:

Intel CPUs Don’t Support ECC Memory: How Bad For A/V Quality?

This discussion may only serve to reveal my less than complete knowledge of A/V digital signal processing rather than raise questions which audibly or not impact signal quality. However, the concerns raised stem from here. https://www.google.com/amp/s/arstechnica.com/gadgets/2021/01/linus-torvald...

audiophilestyle.com

and here are zero-ed samples:

Time resolution of digital sampling

audiophilestyle.com

("musiclong.flac" is original and "musiclongmissing.flac" is with some zero-ed samples)

The change is not necessarily audible in every instance a change was made, but it definitely is audible in some.

Also, AFAIK, USB 2.0 transfers one packet every 1/8000 s, so in case of 44.1 kHz sampling rate a single packet will contain 5 or 6 stereo samples. When CRC check fails, the whole packet will be dropped, so the "hole" in the audio data will be at least that and not just a single sample. Here's an example what it could sound like if 0.1% of packets were lost this way:

2.04 MB file on MEGA

mega.nz

Exactly, errors will manifest differently to a single swapped bit or zeroed bit.

 

[antcollinet]

That's not a signal error, you changed one bit.

It simulates a single bit flip. Bit flips are all you can get on digital connections.

Of course -as we have learned, a bit flip results in a frame checksum failing and the whole frame being dropped.

 

[SanojS]

It simulates a single bit flip. Bit flips are all you can get on digital connections.

Of course -as we have learned, a bit flip results in a frame checksum failing and the whole frame being dropped.

No, you can also have jitter, reclocking issues and introduce ground loop noise into the system if the cable is crap.
But a good enough (cheap) USB cable with sufficient shielding won't have those issues - going with a multi-hundred USD cable will not affect sound quality.

A good DAC can fix some of the issues coming from lower quality USB cables (e.g. reclocking, jitter and some noise filter).

 

[somebodyelse]

No, you can also have jitter, reclocking issues and introduce ground loop noise into the system if the cable is crap.

None of those are a function of the cable.

1. Jitter - for the quality of DACs we're talking about, the transfer is asynchronous, with data buffered on the DAC and played using a clock in the DAC. The cable doesn't influence this clock, so has no influence on the jitter.
2. Reclocking - the cable doesn't reclock anything.
3. Ground loops - a standards-compliant USB cable won't break a ground loop as the shield connections and ground connections aren't optional. For that you should use a proper isolator, now relatively inexpensive even for full speed USB2. There are some cables and devices that break the loop in other ways, but they don't comply with the standards and can cause problems with some types of autodetection.

 

[AdrianusG]

USB is a standardized protocol.

So either the cable transmits what it's supposed to, as mentioned above, or it transmits nothing at all.

This is one of those ludicrous debates in subjective forums, or simply a misleading sales tactic to sell at a higher price.

As for me, the USB cable I use with my converter cost €15.

But I'm not stopping anyone from doing comparative tests where there's nothing to compare.

For example, when I download photos from my camera, the colors are more vibrant with a certain USB cable.

I knew it,

 

[antcollinet]

No, you can also have jitter, reclocking issues and introduce ground loop noise into the system if the cable is crap.

1 - Jitter is not a thing for USB. Data is clocked by the DACs cock - not by the source. Similarly therefore, no re-clocking issues. Even if there were, then jitter rejection and clock recovery in modern dacs is good enough to reduce timing distortion well below the level of audibility. (We don't have to worry about jitter anymore for Toslink or Coax connections either)

2 - Ground loop noise is not altered by a USB cable in any shape or form. While the cable shield will conduct the ground currents - that is regardless of the quality of the cable - all shields conduct ground currents - that is what they are for. Ground loop currents can create signal noise along an unbalanced analogue interconnect - but not along a digital connection. If you have ground noise then you need to eliminate it by breaking it (eg. use toslink) or use balanced analogue interconnect. Playing with USB cables will deliver precisely nothing.

 

[SanojS]

None of those are a function of the cable.

1. Jitter - for the quality of DACs we're talking about, the transfer is asynchronous, with data buffered on the DAC and played using a clock in the DAC. The cable doesn't influence this clock, so has no influence on the jitter.
2. Reclocking - the cable doesn't reclock anything.
3. Ground loops - a standards-compliant USB cable won't break a ground loop as the shield connections and ground connections aren't optional. For that you should use a proper isolator, now relatively inexpensive even for full speed USB2. There are some cables and devices that break the loop in other ways, but they don't comply with the standards and can cause problems with some types of autodetection.

So you are saying the cable can't cause jitter or reclocking issues? Are you sure? Because I know it can.

 

[solderdude]

Is there a cable that causes (audible) jitter and reclocking issues in all DACs (so really a cable issue, assuming it isn't broken) or do you mean that some DACs have a poor USB receiver and/or PLL or internal layout issue that can cause certain (audible) issues ?

There is the crux... the reaction of the USB receiver to incoming data that could potentially cause issues... is that a DAC receiver issue or cable issue ?

 

[Ropeburn]

ABX tests are also statistical and if you repeat the cable test enough times with enough participants, somebody is going to "guess" 10 out of 10 correct (etc.).

That's an interesting point I had never considered. It's rather counterintuitive.

The more runs/cycles/guesses you do per participant, the more the result approaches statistical perfection. That's what you want; sometimes 10 guesses just isn't precise enough.

But! The more participants you test with a low number of guesses each, the more likely you'll get funky outliers. The average of all participants will still approach precision, but everyone will look at the outliers and interpret that according to bias. Not good and the opposite of what you want.

 

[solderdude]

Another downside of AB(X) is that when one does say ... 20 attempts in a row where the differences are hard to hear (but are present) is that several minutes in a continuous test the brain gets 'tired' of having to strain and is likely to make decisions they may not make when 'starting fresh'.
For that reason alone it may be best to divide a listening test in say 4x5 or 5x4 attempts with enough time between them.

With listening tests one always deals with humans (which is a problem in itself).

This is an aspect that should be incorporated in blind test protocols perhaps.

 

[Roland68]

So you are saying the cable can't cause jitter or reclocking issues? Are you sure? Because I know it can.

Is there a cable that causes (audible) jitter and reclocking issues in all DACs (so really a cable issue, assuming it isn't broken) or do you mean that some DACs have a poor USB receiver and/or PLL or internal layout issue that can cause certain (audible) issues ?

There is the crux... the reaction of the USB receiver to incoming data that could potentially cause issues... is that a DAC receiver issue or cable issue ?

If the USB 2.0 cable is built according to the usual specification, with the 5V positive and negative lines being used as spacers for the data line—essentially a starquad arrangement, with all four lines within the shielding—then a noisy 5V supply can, of course, contaminate the data line.

However, the problem lies with the source device, not the cable.
Cables where the wire pairs are shielded separately can mitigate this problem. But naturally, it also depends on the receiving DAC and how it handles the 5V line.
A DAC that creates an additional problem can only be described as extremely poor design and defective.

But the cable itself, if not defective, cannot cause jitter or reclocking problems, at least not more than the cable specifications allow, which is extremely small and often immeasurable.

This means that besides the cable, we still have two variables that can have an influence.
Of course, the cable can't change anything if the data signal has already been corrupted by the source.
That's why it always makes sense to connect a USB isolator as close to the source as possible to prevent unnecessary data corruption.

To be perfectly clear, good, standard USB 2.0 cables, like the Lindy Anthra Line or Cromo Line (€3-9 in standard lengths), do a perfect job, even under poor conditions.

If there's a problem with a device (usually the source), I always recommend fixing the problem at its source instead of investing money in treating the symptoms (USB isolators, cables, etc.).

The Supra Cable Excalibur would be an example of what I consider a perfect USB cable, featuring completely shielded wire pairs, but it's also very expensive. Under normal circumstances, I couldn't detect any audible or measurable advantage over the Lindy cables, not even with a USB analyzer.
Such cable material is available from several companies starting at approximately €8/meter without a plug.

 

[antcollinet]

Is there a cable that causes (audible) jitter and reclocking issues in all DACs (so really a cable issue, assuming it isn't broken) or do you mean that some DACs have a poor USB receiver and/or PLL or internal layout issue that can cause certain (audible) issues ?

There is the crux... the reaction of the USB receiver to incoming data that could potentially cause issues... is that a DAC receiver issue or cable issue ?

Are there any modern USB DAC interfaces that are not asynchronous?

Certainly the vast majority are. And all those that are - don't have any issues with jitter or re-clocking - they are simply not part of the process.

So you are saying the cable can't cause jitter or reclocking issues? Are you sure? Because I know it can.

Then you need to start relearning what you think you know at least as regards modern designs of USB DAC interfaces.

And as I point out above - even for interfaces where jitter is relevant - it is rejected by even halfway competent modern DAC implementations to well below the level of audibility. Jitter has been a solved problem for decades.

 

[somebodyelse]

So you are saying the cable can't cause jitter or reclocking issues? Are you sure? Because I know it can.

Please explain how the cable can cause either jitter or reclocking issues given that the cable doesn't have a role in either clocking or reclocking.

 

[SanojS]

Please explain how the cable can cause either jitter or reclocking issues given that the cable doesn't have a role in either clocking or reclocking.

Yes, cable doesn't do clocking/reclocking and I never said so, I said it can CAUSE ISSUES with reclocking in DAC stage, that affects the receiving end from properly reclock the signal. It can be because of significant noise pollution through the ground for example, or if jitter is severe enough that the DAC isn't able to completely recover the correct clocking sequence.
Jitter is the result of improper timing between packets and can be the result from impedance mismatches, sub-par shielding causing interference, and signal degradation.

Happy now?

 

[SanojS]

Are there any modern USB DAC interfaces that are not asynchronous?

Certainly the vast majority are. And all those that are - don't have any issues with jitter or re-clocking - they are simply not part of the process.




Then you need to start relearning what you think you know at least as regards modern designs of USB DAC interfaces.

And as I point out above - even for interfaces where jitter is relevant - it is rejected by even halfway competent modern DAC implementations to well below the level of audibility. Jitter has been a solved problem for decades.

I know plenty, thank you.
Yes, modern devices have been built to rectify most common issues from improper cables/signal interference and jitter. But if you look on the subject of this thread, you will see what a cable test should be like. That is what I was replying on - what issues a bad cable can cause. This is still true, if you wanna compare cables, you have to take into consideration such things.

Please note that many many audiophiles run old sensitive DACs. I have a friend with such an R2R DAC that requires signal cleaning before doing its job. So yes, I know what I'm talking about. Stop making presumptions about other people's knowledge.

 

[TLEDDY]

“…re-frame the discussion for those reasonable folks who have their opinions but are open to new information”

Sorry - despite the excellent experiment design, True Believers will denigrate the findings.

Question showing my ignorance: Null testing of analog measurements are conclusive - is it possible to null test the digital output, perhaps post DAC?