The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
USB cable impact sounds quality?
Old 19th January 2019
  #1
Gear Addict
 

USB cable impact sounds quality?

Recently I saw few shops selling high quality USB cables for audio interfaces. So i'm wondering if it impact the sound quality compared to a regular one ?

I saw this model for example : ' OYAIDE USB Class S NEO SERIES V2'
Old 19th January 2019
  #2
Another can o'worms we can open here? To be fair, there are a bunch of specialized USB cables made by many of the better companies. Shielding the power from the data lines is a basic first step. Kimber and AudioQuest make the expensive ones, Pangea is the lower cost yet well designed USB cable.
Old 19th January 2019
  #3
Lives for gear
IMO, one could make an argument for S/PDIF or AES where the receiver uses the transmitter clock as it’s own ADC clock. But for USB, I just don’t see it. If a packet of data goes through corrupted it will get fixed by the error correction information in the data packet. If that doesn’t work, USB will keep requesting a resend until it comes through correctly.

This didn’t stop someone from sharing with me on another thread that a USB cable made a difference. Who am I to say it didn’t? I’ve been wrong enough times that I don’t declare something can’t happen simply because I don’t understand it.

Trust your ears I say. If you can’t here a difference, don’t let Anyone tell you it’s there. And conversely, don’t let anyone tell you there’s no difference when you hear one.
Old 19th January 2019
  #4
Old 19th January 2019
  #5
Gear Addict
 
whippoorwill's Avatar
Does anyone know a usb cable that has a rugged build, maybe the pangea above?
I go through about one a year of cheap $5-10 usb cables, I think, because they always seem to fail. I use a laptop so I am always unplugging and replugging.
Old 19th January 2019
  #6
Lives for gear
 
Poinzy's Avatar
 

Boutique USB cable: 'I want to believe.'

Moving on...

I'd say the key to making your USB cables last is to pull them out by the plug, not by the casing. I just use Belkin cables when I can get them. I have some that are almost 20 years old, and have never had a problem with them. But I'm gentle with them.

I can usually tell just by looking at a cable whether construction is crappy or not.
Old 4 weeks ago
  #7
Lives for gear
 

Verified Member
Quote:
Originally Posted by u87allen View Post
... But for USB, I just don’t see it. If a packet of data goes through corrupted it will get fixed by the error correction information in the data packet. If that doesn’t work, USB will keep requesting a resend until it comes through correctly.
...
That is true, but some context is necessary. Errors in all four USB modes are always detected, and those that can be corrected on-the-fly are fixed. For those that cannot be fixed. Re-transmissions are one done only in command mode and bulk transfer mode (disk stuff), but not for the isochronous stuff like music/video.

For example, in isochronous mode, each 1024-byte data packet would have up to 7-milliseconds worth of music samples at 24/48k. You can readily see that any re-transmission could cause an out-of-order delivery or significant glitch/click that could only be prevented with a fairly deep buffer on one end to organize re-transmitted packet before being released to the converter (and some really smart session-aware endpoint daemons/agents). Deep buffers involve the dreaded latency that we're always trying to avoid.

The good news is that transmission errors are exceedingly rare. The baseline design of USB is to have better than one bit error in 10^12 bits... better than one bit per terabyte.

[ USB Protocol: Types of USB Packets and USB Transfers | EngineersGarage ]
[ USB in a NutShell - Chapter 4 - Endpoint Types ]
Old 4 weeks ago
  #8
Lives for gear
Quote:
Originally Posted by MediaGary View Post
That is true, but some context is necessary. Errors in all four USB modes are always detected, and those that can be corrected on-the-fly are fixed. For those that cannot be fixed. Re-transmissions are one done only in command mode and bulk transfer mode (disk stuff), but not for the isochronous stuff like music/video.

For example, in isochronous mode, each 1024-byte data packet would have up to 7-milliseconds worth of music samples at 24/48k. You can readily see that any re-transmission could cause an out-of-order delivery or significant glitch/click that could only be prevented with a fairly deep buffer on one end to organize re-transmitted packet before being released to the converter (and some really smart session-aware endpoint daemons/agents). Deep buffers involve the dreaded latency that we're always trying to avoid.

The good news is that transmission errors are exceedingly rare. The baseline design of USB is to have better than one bit error in 10^12 bits... better than one bit per terabyte.

[ USB Protocol: Types of USB Packets and USB Transfers | EngineersGarage ]
[ USB in a NutShell - Chapter 4 - Endpoint Types ]
Thanks for the info.

If what you're saying is true it would seem the error probability would be dependent on cable quality and length when streaming audio. Or at least beyond the threshold where the packet error correction could no longer recover the lost data.
Old 4 weeks ago
  #9
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Digital cables can make a difference but it has to do with analog factors such as grounding and immunity to RFI. It should make no difference with properly designed and built gear but plenty fails to isolate digital connections from analog circuitry and/or power supplies. I think cables are always a Band Aid on bad gear designs but there are plenty of those in the real world.
Old 4 weeks ago
  #10
Gear Nut
 
Digital Crush's Avatar
Vintage usb cables still sound better
Old 4 weeks ago
  #11
Lives for gear
 
IanBSC's Avatar
I compared a Shunyata Venom USB cable to my stock cables, and it definitely did sound different, more focused and precise with both my Mytek Manhattan and PS Audio Directstream DAC. My guess is that it has more to do with low-level electrical interference getting into the DAC than dropped packets or jitter. Obviously, it is irrelevant on the computer side, but there can be a sonic benefit to better converter isolation from the PC. That might have an impact on ADCs as well.

I've found that statements that "properly designed xxx shouldn't benefit from expensive yyy" are usually extremely optimistic. Even world class gear is more imperfect in the execution than most want to believe. But generally, I think it's not so much an issue of poorly designed equipment, but rather the opposite: that in a high level and resolving setup you can hear very subtle discrepancies.

Not the most crucial component, but in a sufficiently transparent high resolution system, it would certainly be audible.
Old 4 weeks ago
  #12
Run digital audio through USB (staying digital), take the resulting file and do a nul test with the source file.
It cancels completely, then move one !!

If it doesn't, then your system is not setup properly.
The most common thing would dither added at some stage (leaving the DAW or around the soundcard mixer) or src.

Otherwise, as Bob said, it's actual hardware issue.
Old 4 weeks ago
  #13
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
A data null doesn't mean anything because the problem lies in the analog circuitry or the clock (also analog...) in the D to A converter. There are very few people who really understand both analog and digital engineering design.
Old 4 weeks ago
  #14
Lives for gear
 
jaxman12's Avatar
I think it a lot more basic than people realize. With analog, the signal is effected by outside rf causing noise, improper grounding, signal degradation, etc. can effect the quality of the sound. Digital is sent in packets. Either the data packet is there or it is not. Snaps, crackles, and no signal are a sign of signal loss in a digital signal. So the question is " Is the quality of the sound effected? No, the signal is either there or it isn't. Outside interference does not add data packets.
Old 4 weeks ago
  #15
Lives for gear
Quote:
Originally Posted by jaxman12 View Post
I think it a lot more basic than people realize. With analog, the signal is effected by outside rf causing noise, improper grounding, signal degradation, etc. can effect the quality of the sound. Digital is sent in packets. Either the data packet is there or it is not. Snaps, crackles, and no signal are a sign of signal loss in a digital signal. So the question is " Is the quality of the sound effected? No, the signal is either there or it isn't. Outside interference does not add data packets.
I think many digital audio systems have the ability to do interpolation if a bit can’t be recovered, no? And if so, digital loss can occur before one hears clicks or pops.
Old 4 weeks ago
  #16
No need for that. It's trivial to use small buffers and send small packets around, each containing exact timing info and some sort of checksum. ASIO for example imposes a buffer - for good reason. Most engineers building hardware aren't exactly stupid. Nobody in the audio biz sends raw time dependent data over an USB cable. Assuming it (without any evidence at hand) is somewhat insulting toward engineers forced to the hardest exams you can find in the academic sector. They usually build systems having several magnitudes of greater complexity than a simple digital audio connection. All using impressive, hard to believe cleverness and creativity.

Given the budgets required, they use to mathematically prove the integrity of their systems against the information theorem, before even starting implementation. Several reliable and proven tricks exist that allow enforcing things like stability, authenticity and integrity. No black art btw, simple university schoolbook stuff.
Old 4 weeks ago
  #17
Lives for gear
Quote:
Originally Posted by FabienTDR View Post
Assuming it (without any evidence at hand) is somewhat insulting toward engineers forced to the hardest exams you can find in the academic sector.

.
Eh, if they can’t handle the idea they might be wrong or might not be engineering perfectly, I don’t think I’d want them on my engineering team.


If what mediagary says is correct, then the potential exists for data loss in a USB cable streaming audio.
Old 4 weeks ago
  #18
Again, it's trivial to exclude any problem, by design! This is taught and in broad use. You simply don't do such type of stupidity after have proved fourier, the sampling theorem and various other stuff from just any perspective you can imagine. You then clearly understand and know exactly where to find and how to prevent the problems. It's maybe difficult to imagine, but today's level of cleverness in anything technical is just stunning!

Kruger Dunning explain well how easy it is underestimating things you didn't learn in depth. I say this without any negative intention, it's just how it is, nobody's immune from it. Universities teach very deep, very solid stuff. Easy to underestimate. That's why humanity can stack things that high: The level of certainty and detail in technology has become incredible, almost surreal. Digital interconnection is a triviality.

In a practical sense: This thread is projecting the deficits of pigeons based or morse based communication (something everybody can imagine ) into modern realtime orientated audio systems (something taking years of studies to fully understand the tricks of the trade).

The greatest irony is finding this scepticism spread via the internet! Using a highly fragmented digital system made of gazillion cables of all sort and length, in total length, spanning several thousand times around the planet. And recalling these articles works, from north to south-pole, day and night, in any weather, without losing one bit of information. Someone must be wrong I guess!

Last edited by FabienTDR; 4 weeks ago at 10:54 PM..
Old 4 weeks ago
  #19
Lives for gear
Quote:
Originally Posted by FabienTDR View Post
Again, it's trivial to exclude any problem, by design! This is taught and in broad use. You simply don't do such type of stupidity after have proved fourier, the sampling theorem and various other stuff from just any perspective you can imagine. You then clearly understand and know exactly where to find and how to prevent the problems. It's maybe difficult to imagine, but today's level of cleverness in anything technical is just stunning!

Kruger Dunning explain well how easy it is underestimating things you didn't learn in depth. I say this without any negative intention, it's just how it is, nobody's immune from it. Universities teach very deep, very solid stuff. Easy to underestimate. That's why humanity can stack things that high: The level of certainty and detail in technology has become incredible, almost surreal. Digital interconnection is a triviality.

In a practical sense: This thread is projecting the deficits of pigeons based or morse based communication (something everybody can imagine ) into modern realtime orientated audio systems (something taking years of studies to fully understand).
Compromise is not stupid. It's engineering, and has to be done often times to arrive at practical solutions. And if one wants to stream audio with the lowest possible latency through long cables and hit a certain market price point, compromises, like tolerating some dropped bits, may need to be made.

You seem to be suggesting that this engineering is just so far beyond anyone here that we have can't even question it. I'm not with you on that one. I know many engineers, some with PhD's. And none of them are so far beyond me that I can't possibly comprehend what they are doing without going back to school for 4-8 years. So I'm going to keep questioning.
Old 4 weeks ago
  #20
Lives for gear
 
Paul Gold's Avatar
 

Verified Member
Quote:
Originally Posted by FabienTDR View Post
Again, it's trivial to exclude any problem, by design!
Does a box like this improve signal quality?

MUTEC - Professional A/V and High-End Equipment - MC-3+ USB
Old 4 weeks ago
  #21
Do you understand how fast modern USB is? Where is the bottleneck in low latency? USB can send the packet, including a checksum, check it and answer with a "resent the same package please", and repeat the process a few thousand times before time runs out. Even then, the device will easily be able to show an error state, return zero and warn the user of a connection problem.

pigeons! The last relict of such naive protocols in audio is midi. But modern systems are super reliable.
Old 4 weeks ago
  #22
Quote:
Originally Posted by Paul Gold View Post
Does a box like this improve signal quality?

MUTEC - Professional A/V and High-End Equipment - MC-3+ USB
Yes, but has nothing to do with digital interconnection or cables. Clocking is a AD/DA problem, a very analogue problem. Crane song have a nice jitter audio demos on their website, this effect is real (but unrelated to digital itself, clocking is an analogue task).
Old 4 weeks ago
  #23
Lives for gear
Quote:
Originally Posted by FabienTDR View Post
Do you understand how fast modern USB is? Where is the bottleneck in low latency? USB can send the packet, including a checksum, check it and answer with a "resent the same package please", and repeat the process a few thousand times before time runs out. Even then, the device will easily be able to show an error state, return zero and warn the user of a connection problem.

pigeons! The last relict of such naive protocols in audio is midi. But modern systems are super reliable.
Actually, did some reading on this. Apparently there is no mechanism for re-transfer in isochronous mode, which is what USB audio uses for data transfer.
Old 4 weeks ago
  #24
Lives for gear
 

Verified Member
Quote:
Originally Posted by u87allen View Post
Actually, did some reading on this. Apparently there is no mechanism for re-transfer in isochronous mode, which is what USB audio uses for data transfer.
Correct. Now, consider that a USB connection is *not* a cellphone connection. In contrast, it has an excellent signal-to-noise ratio. That's how it achieves the high data rate with stability. Remember, to qualify at a minimal level of a single bit per 10^12 bits connection is to say that an error is possible...but not likely.

Mr. @FabienTDR makes point that are well taken. We blithely use the public internet which is using the same CRC error checking and checksum, and 10b/8b byte encoding, and Reed-Solomon encoding, and a bunch of other mechanisms to ensure that the data send it exactly the data received. There are no "half-bits" or "weak bits" that get interpolated.

Once upon a time, we in the local area network (Ethernet, Token-Ring, FibreChannel, ESCON, etc) would shake our heads at the pains the wide area network designers (T1, OC-x, SONET, etc) had to go through to ensure that the data got there in a reliable fashion. Today, much of that data integrity discipline is a normal part of our Thunderbolt, SATA and USB connections.

Relax and enjoy the data.
Old 4 weeks ago
  #25
Lives for gear
Quote:
Originally Posted by MediaGary View Post
Correct. Now, consider that a USB connection is *not* a cellphone connection. In contrast, it has an excellent signal-to-noise ratio. That's how it achieves the high data rate with stability. Remember, to qualify at a minimal level of a single bit per 10^12 bits connection is to say that an error is possible...but not likely.

Mr. @FabienTDR makes point that are well taken. We blithely use the public internet which is using the same CRC error checking and checksum, and 10b/8b byte encoding, and Reed-Solomon encoding, and a bunch of other mechanisms to ensure that the data send it exactly the data received. There are no "half-bits" or "weak bits" that get interpolated.

Once upon a time, we in the local area network (Ethernet, Token-Ring, FibreChannel, ESCON, etc) would shake our heads at the pains the wide area network designers (T1, OC-x, SONET, etc) had to go through to ensure that the data got there in a reliable fashion. Today, much of that data integrity discipline is a normal part of our Thunderbolt, SATA and USB connections.

Relax and enjoy the data.
Where does the 10^12 number come from? I've definitely experienced USB audio connections that have dropped enough bits to cause a click or even a freeze in communications. I don't think we can guarantee any USB connection anywhere will have this level of reliability. If the 10^12 number were real and consistent I think we'd never have the need for packet retransmission. CRC, or even parity, would be sufficient, as the probability a 1000 bit packet would have two errors would be something like 10^18.

Last edited by u87allen; 4 weeks ago at 03:51 AM..
Old 4 weeks ago
  #26
Lives for gear
 

Verified Member
Quote:
Originally Posted by u87allen View Post
Where does the 10^12 number come from? I've definitely experienced USB audio connections that have dropped enough bits to cause a click or even a freeze in communications. I don't think we can guarantee any USB connection anywhere will have this level of reliability. If the 10^12 number were real and consistent I think we'd never have the need for packet retransmission. CRC, or even parity, would be sufficient, as the probability a 1000 bit packet would have two errors would be something like 10^18.
Lots of stuff falls out of a search for USB BER ... Here's one ( https://www.google.com/url?sa=t&sour...deiTxw9AOnihzJ )

The USB.org site has lots of stuff.
Old 4 weeks ago
  #27
Lives for gear
Quote:
Originally Posted by MediaGary View Post
Lots of stuff falls out of a search for USB BER ... Here's one ( https://www.google.com/url?sa=t&sour...deiTxw9AOnihzJ )

The USB.org site has lots of stuff.

got a couple of hits for USB 3.0. And yes, they mention that 10^12 number before error correction. But didn’t see anything for usb 2 as far as bit error rate, which is what like 90% of people are using for streaming audio I think.
Old 4 weeks ago
  #28
Lives for gear
 

TL;DR
Transfer errors aren't the reason for getting super expensive USB cable.

It's true, that vast* majority of current USB interfaces use isochronous transfers, I wrote about that couple times on GS mostly in the latency thread. And that this type of transfers doesn't support built-in re-transmission of data packets, however there are 16 bit CRCs. So it's possibly to know exactly, how many packets gets possibly garbled during transfer.
I've worked on firmware and played with couple of different XMOS development boards for USB interfaces, I also had access to HW USB analyzer. It's not really problem to implement a counter for such errors in firmware, so devs and designers can easily find, whether that's happening or not.
RME USB interfaces has such error counter available for users. If you open their interface control panel, you can easily find that.

It can happen of course, but in practice with normal devices, which adheres to specs, it's not really any issue with that. You can also provoke errors, with intentionally broken cable, some weird impedance mismatch on data lines, intentionally make worse CMRR on balanced data lines etc. but more often than not, if someone experience that, it's some issue with upstream hub or hardware problem with host controller. Naturally all those issues affects all transfer types, so likely there would be also problems right from the start with enumeration and control transfers, you'd experience random disconnects etc.

You can test normal setup for days or even weeks and you'd really hard pressed to find any transfer related errors. Of course, as already mentioned, you can always try digital loopback and bit compare audio.

Frankly if that would be regularly happening under normal circumstances, it wouldn't be possible to use isochronous packets at all. Nobody wants error in transfers, be it audio, video or other latency sensitive application.
So from my point of view, you can't really address feelings about better soundstage, tighter bass with some tellurium, cryogenically treated cable to data errors.

Michal



* USB 3.0 is bit more problematic and sensitive to proper connections, termination and interfacing than 2.0 and requires shorter cable, because of its higher bandwidth and frequencies. Also initially there was compatibility issues with isochronous transfers among different controllers. In that case, bulk transfers can be more suitable than isochronous, even for audio applications.
Check for example page 100 at RME MADIFace XT manual
https://www.rme-audio.de/download/madiface_xt_e.pdf
It's a general thing and also the reason, why some isochronous USB 3.0 setups has problems (audible hard clicks and dropouts, not some audio homeopathy) and when you switch it back to 2.0 mode, it's fine. I've experienced that on MOTU, Zoom for example with certain controllers.
Old 4 weeks ago
  #29
Lives for gear
 
SASMastering's Avatar
edited :

Last edited by SASMastering; 4 weeks ago at 11:35 AM.. Reason: not really relevant to sound quality
Old 4 weeks ago
  #30
Lives for gear
 

Verified Member
The error detection and recovery description narrowly pertinent to this discussion is on page 19, and pages 79-83 in the USB spec document.

https://www.google.com/url?sa=t&sour...wo1YlIKPxnafoJ

When I get to a laptop, I'll clean up the links with an edit.. It's a 600+ page PDF. Bit error rate stuff for USB 2.0 is elsewhere. I'll track it down when I have some time.

EDIT: A link to the USB Serial Bus Specification design document is below.
[http://sdphca.ucsd.edu/lab_equip_manuals/usb_20.pdf ]
This document is available in multiple places on the web. Pages 19, 47, and 79-83 are part of the answer. I'll look for the USB 2.0 Bit Error Rate stuff later.

EDIT2: Here's a link to a USB 2.0 clock and data recovery circuit that's the guts of the physical link to the cable. It specifies its error rate to be better than 10^12 in the bold script abstract.
[https://www.cs.ccu.edu.tw/~wildwolf/...idat2011-2.pdf ]

Last edited by MediaGary; 4 weeks ago at 02:02 PM.. Reason: Updated USB Design PDF link
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
frankthefunky1 / Music Computers
9
light767 / Connectors, cables, stands and accessories
16
MikeMello / Connectors, cables, stands and accessories
10
DeyBwah / So much gear, so little time
0

Forum Jump
Forum Jump