Do USB Audio Cables Make A Difference?

[b4nt]

Livin in a world where some folks actually believe a cable carrying a digital signal can sound different than the next.

As I wrote earlier, bits may flip. But this shall be audible. Yesterday, my conferencing mic/speaker crackled again, a lot.
- either you get crackles, bits have flipped, swap the USB cable
- or none, just your tracks, your short USB cable is fine

I beleive bunch of bits will flip for any bad USB cable, connector or cablings.

In the telecom world, it is very common to have transmission errors due to bad equipments, optical patches, cables and wirings. It is also common to have a very minimal but acceptable error rate on links, depending on transmission rates and technologies. A single bit flip for 10^9 or 10^12 transmitted can be acceptable there.

I do not think there is any error check on HDMI for video. Neither any retransmit. My video here is always fine, with huge amounts of bits sent every days. If I would have errors due to my HDMI cable, I beleive I would notice it on my display.

If one would like to test his USB cable, he would need to transmit bits over it, and evaluate the error rate, not with audio, but with a link tester.

And for the sound itself, I wouldn't know either how this could happen. I dont change my mains cables nor my USB cable to tweak my audio.

 

[mansr]

It is also common to have a very minimal but acceptable error rate on links, depending on transmission rates and technologies. A single bit flip for 10^9 or 10^12 transmitted can be acceptable there.

USB 2.0 recommends targeting a bit error rate of 1e-12 or better.

I do not think there is any error check on HDMI for video. Neither any retransmit. My video here is always fine, with huge amounts of bits sent every days. If I would have errors due to my HDMI cable, I beleive I would notice it on my display.

The HDMI video signal uses 8b/10b or (for high resolutions) 128b/132b encoding which provides some error correction and detection. A high error rate generally manifests as sparkles in the image. An occasional error is unlikely to be noticed visually, especially when viewing video.

 

[b4nt]

USB 2.0 recommends targeting a bit error rate of 1e-12 or better.

For the cable or for the link, which includes the two receivers/transmitters? I have sensitive speakers, and I'm myself sensitive to clics, pops and noise. If I would get bit flips, repetitive bit flips, depending indeed which bits flip, I think I would notice them, on quiet vocal, flute or piano tracks.

If bits flip for a loud distorded pop track, not sure all can be audible. Maybe in some quieter or vocals tracks parts.

According to my ears, I never had to check and swap my almost basic and cheap USB cables. I had noises, but due to EMI, mains, and PSUs, and due to a bad RCA cable, plus due to my phone being to close, that's all, and that was solved.

My conferencing mic/speaker sometime crackles. I do not care, but this is audible and loud.

The HDMI video signal uses 8b/10b or (for high resolutions) 128b/132b encoding which provides some error correction and detection. A high error rate generally manifests as sparkles in the image. An occasional error is unlikely to be noticed visually, especially when viewing video.

I'm not that often watching videos on my PC, maybe 1/50th of time. And I never noticed any visual effect poping in my viewing area over years I'm using HDMI on PC, with basic graphics. Sometime a pixel or two may have flashed? I noticed nothing.

 

[DonH56]

Think of how hard it is to see a single or even a few pixels flashing, and the analogy carries through to digital audio. A erroneous bit or three is probably unnoticeable given the bit rate even without retries (error correction would catch most low-bit errors anyway). Audible distortion tends to be caused by large and sustained errors generated by a broken cable, too long a cable run, or similar. The degradation is not usually gradual, like a rising analog noise floor, instead it goes from clean and clear sound to audible glitches with little transition band. That is why so many say it either works or it doesn't. FWIWFM, that tends to be true with high-speed data links like PCIe and SAS/SATA, USB, etc. Testing such serial links, the BER goes from miniscule (1e-15 or so) to very high (1e-3 or less) with little change in impairments. From none or maybe one or two errors in minutes or hours to thousands in seconds with almost nothing between.

 

[b4nt]

In the past, I had to test 34Mbps optical links and equipments. If I remember well, we where expecting not more than 1 or 2 errored bit (or maybe seconds with errors) over 24 hours or so. Where we noticed more, it was due to bad levels, dirty fibers, or hardware, and it was much more, with errors every minute to 15 minutes. In rare cases only, we had more than two errors/events over 24 hours, that was less easy to solve.

I'm using DAC + USB since 6 years now, very often during hours. Never had to worry about USB data transmission. Others maybe.

 

[dc655321]

Think of how hard it is to see a single or even a few pixels flashing, and the analogy carries through to digital audio. A erroneous bit or three is probably unnoticeable given the bit rate even without retries (error correction would catch most low-bit errors anyway.

An audio BER of even a few ppm is not subtle.
https://www.audiosciencereview.com/...dio-cables-make-a-difference.1887/post-771406

 

[b4nt]

An audio BER of even a few ppm is not subtle.
https://www.audiosciencereview.com/...dio-cables-make-a-difference.1887/post-771406

So I'm almost sure I would notice a single flipping bit in a 8 hours listening session

As I wrote earlier, it might depend on which kind of track you listen to (like if you have or not clicks in that track itself). Or which bit is flipping. Would be randomly spread over audio plus quiet parts in tracks, more or less repetitive, and shall end noticeable.

 

[b4nt]

An audio BER of even a few ppm is not subtle.
https://www.audiosciencereview.com/...dio-cables-make-a-difference.1887/post-771406

I have this for you, discrete pop in a pop track, unaudible for many (attachment D.zip): https://www.audiosciencereview.com/...-khadas-tone-board-1.24596/page-8#post-833718

 

[DonH56]

An audio BER of even a few ppm is not subtle.
https://www.audiosciencereview.com/...dio-cables-make-a-difference.1887/post-771406

A "few ppm" is on the order of 1e-6 BER and is pretty gross (though audibility would much depend upon which bits and context). And that would have to be after error correction and such. At 44 kS/s (~704 kbits/s per channel) that is an error about every 0.7 seconds. If the error rate were that high I would expect the link to be near failure.

 

[b4nt]

A "few ppm" is on the order of 1e-6 BER and is pretty gross (though audibility would much depend upon which bits and context). And that would have to be after error correction and such. At 44 kS/s (~704 kbits/s per channel) that is an error about every 0.7 seconds. If the error rate were that high I would expect the link to be near failure.

Yes, but with that 10ppm sample, we can clearly hear both the repetition and the effects. If I would get such a loud clic once or twice a day, or even occasionnaly, I would notice it.

 

[sjeesjie]

I have this USB cable with ferrite endings on both end. That’s to end RFI right? Would that make any difference?

 

[Julf]

I have this USB cable with ferrite endings on both end. That’s to end RFI right? Would that make any difference?

Typically used on laser printers and other devices causing a lot of HF noise. The cable protects other devices, not the device itself.

 

[Katji]

And that led to many people thinking they are helpful for USB audio, because DACs typically have USB B connectors, typically used for printers. Probably why so many USB audio cables have them.

 

[sonci99]

What about the quality of usb ports, do they matter?
Especially for dacs that have no power supply but feed from the usb cable, I noticed that even the volume is different when connecting the khadas tone to laptop or to the desktop..

 

[SIY]

What about the quality of usb ports, do they matter?
Especially for dacs that have no power supply but feed from the usb cable, I noticed that even the volume is different when connecting the khadas tone to laptop or to the desktop..

Might the audio configuration in one of devices be set incorrectly?

 

[CuteStudio]

Especially for dacs that have no power supply but feed from the usb cable

I had a terrible buzzing hum from a DAC (The Apple USB-C one), and it was caused by a dodgy PSU feeding the Raspberry Pi power.

So yes, in my experience the PSU is critical: I've ordered a USB extender and I'll make a separate PSU, so I'll break the +ve wire, and feed a clean 5V from a 7812 (EDIT: I mean a 7805!) (fed from a 9V supply I have knocking about) feeding an OSCON cap into the output, where I'll plug the DAC into.

So the DAC will get a totally clean supply.

 

[Julf]

I had a terrible buzzing hum from a DAC (The Apple USB-C one), and it was caused by a dodgy PSU feeding the Raspberry Pi power.

So yes, in my experience the PSU is critical: I've ordered a USB extender and I'll make a separate PSU, so I'll break the +ve wire, and feed a clean 5V from a 7812 (fed from a 9V supply I have knocking about) feeding an OSCON cap into the output, where I'll plug the DAC into.

So the DAC will get a totally clean supply.

So that DAC seems to have poor power supply stabilization/filtering. I would not use a linear supply to feed digital circuits.

 

[solderdude]

and feed a clean 5V from a 7812 (fed from a 9V supply I have knocking about)

I would use an 7805 instead
It also won't help against groundloops that are more common than 'clean' 5V.
The ground connection is where the gremlins usually pass through.

 

[CuteStudio]

I would use an 7805 instead

Haha, yes of course, I must have been thinking about 12 when I typed that in!
For the Raspberry Pi there is no ground as the PSU (and the clean 7805+Cap supply) will be fed from double isolated adapters.

Ironically the buzzing PSU was an official Pi PSU, and was also double isolated I think - no earth pin connection to the output.

 

[CuteStudio]

So that DAC seems to have poor power supply stabilization/filtering. I would not use a linear supply to feed digital circuits.

I think all filtering is simply a dB noise rejection ratio, some are better than others, and in general (but slightly off topic) I find RF gets through anything . So even a great DAC may only have a 60dB PSU noise rejection..
The Apple DAC is only a few mm big, so great filtering is neither expected or designed

This does remind me though - I had some small amplified speakers chargeable from USB, and they buzzed in cheap USB sockets too, some USB sources are very dirty.

I'm not sure what's different about a linear or a switched supply for a DAC. They both charge of a capacitor, which is the ultimate arbiter of stability on the fine grain, neither a switched or a linear can react so fast.
But if there is a reason please say, I'm not a PSU designer!

Also of course, this is not a power amplifier of bass box - this is a low power DAC taking very few mW, so the source of energy into the Cap merely has to be quiet and stable: which is in general what a 7805 or similar gives (quieter than any switched supply - by definition), and even an RC filter after would be an option.