Did We Overhype USB Audio And Overlook Possible Pitfalls?

[The Computer Audiophile]

Hi Guys - The title of this thread "Did We Overhype USB Audio And Overlook Possible Pitfalls?" is meant to be read as a question, not a conclusion or suggestion.

 

 

A little background:

 

Remember the days of adaptive USB audio and 29/96 limitations? I remember thinking how great this was considering we could then play high resolution rather than being stuck at 48 kHz and lower. Then I remember the wonderful introduction of asynchronous USB and even higher resolution playback up through 24/192. Wow, life was good, especially when compared to where we had previously been with USB audio. Then came DSD via DoP. Another neat step in the USB audio evolution. More recently we are in a sample rate arms race of DSD256/DSD512 and PCM 32/384 or higher. I'm not suggesting this arms race it's beneficial, especially when so many members of the CA community really enjoy sample rate and format conversion prior to sending the audio to their DACs. But, in this evolution to higher and higher rates, it seems like very few people and still fewer manufacturers are talking about other areas of USB audio improvement. Companies like AudioQuest (Jitterbug) and UpTone Audio (Regen) have developed products to improve the USB audio experience. Many people will never go back to listening without one of these devices. Also, some manufacturers refuse to put USB inputs on their components (Berkeley Audio Design) while others recommend using digital inputs other than USB (Schiit, Hegel).

 

 

I've been thinking long and hard about USB audio for several months and even talked to some well respected audio engineers about the topic. I've been listening to many DACs using different inputs and trying to judge whether I can conclude anything with respect to the quality of an input and have this judgement be relevant for anything other than the exact system I'm using at the time. I haven't been able to come to any conclusion stating that one input method is always better than another or than USB is inferior to Ethernet or AES etc... But, I have listened to DACs in my system that certainly sound worse via their USB inputs. I raised this finding with a few engineers in the industry and was surprised to hear them either agree with me, take it a step further and says USB is too flawed, or provide me with solid objective information for using USB in other ways. One example of this came when talking to a manufacturer who has offered USB inputs on its DACs for many years and now offers Ethernet inputs (in addition to USB). This manufacture thinks Ethernet is capable of much better audio than USB. Sure that's one manufacturer's opinion and not indicative of anything else, but it's someone whose opinion I highly respect. Another manufacturer I talked to at CES told me something very interesting about his tests with USB inputs. He said when he physically removes the USB input / receiver board from the DAC and places it on top of the chassis for testing, the DAC measures better. This physical separation of USB from the DAC was objectively a better solution. He also said his customers weren't interested in another box or different solution, so he was sticking to the current integrated USB design. He ideal solution would be to have a USB to AES converter to send AES into his DACs. This is exactly what Berkeley Audio Design has been saying since it released the Alpha USB years ago.

 

 

That said, I'd like to discuss the question of whether or not we (me included) overhyped USB and/or overlooked some of its pitfalls, in addition to other people's thoughts about anything USB audio related.

 

Again, I'm making not judgement or conclusion about USB audio and in no way suggest that people render an opinion based on my questions. It's simply a topic I'm interested in discussing and think we could all learn quite a bit from members of the CA community.

[The Computer Audiophile]

I look at how much trouble and expense people here are going to to improve this superior technology and my head spins. If MQA is great and requires USB I'll go back but for now my Chromecast/Toslink to Remedy configuration is satisfying and cost effective for 99% of my listening (and sounds better than my USB ). Money not spent on $400 USB cables I can think of better uses for.

MQA doesn't require USB.

[The Computer Audiophile]

For full implementation how does MQA identify the DAC used absent two way communication with the DAC? Id be happy if you're right about that.

Meridian has always shown MQA using Ethernet. Other manufacturers who don't even support USB or Ethernet input on the DAC will have full implementations of MQA.

[The Computer Audiophile]

No idea where it is all going, but it sure is a great time to be alive and an audio hobbyist!

Agree!

[The Computer Audiophile]

Optical or wireless transmission have the inherent advantage of electrical isolation. I wonder if as much time and as many resources were placed into developing either of these, whether they would be ahead of where USB is now?

I really wish optical would be perfected.

[The Computer Audiophile]

I2S is is a trivial clock+data serial interface only suited for short links, typically on the same PCB or over a very short cable, due to the lack of any error correction. For transmission from a computer to a different device, you really want something using differential signalling and at least some amount of error correction.

There is no error correction in USB audio.

[The Computer Audiophile]

USB packets have checksums to detect any single or double bit errors, and since it uses differential signalling, it is more robust against interference than I2S to begin with.

I'll repeat, USB audio has no error correction (as told to me by the Godfather of USB audio, Gordon Rankin).

[The Computer Audiophile]

I just checked the USB 2.0 spec. It has CRCs for each packet. The USB audio layer doesn't add any additional error correction.

Isochronous USB (used for adaptive and asynchronous audio) can identify errors but has no error correction.

 

Seriously, ask Gordon or John Swenson or someone who has worked with USB audio extensively.

[The Computer Audiophile]

Correct, but he was mentioning CRC, this is a 'Check', not a correction.

Yes, but he mention there was no "additional" correction. Thus, implying (after first stating) there is correction to begin with.

[The Computer Audiophile]

Async USB Audio has error detection in the from of a CRC attached to the packet. But there is no error correction, meaning the ability to retransmit the packet like in Ethernet etc. So if there is an error in a malformed or corrupt packet, it is simply discarded from the buffer.

I believe that's what I've been saying as well.

 

 

 

I don't need Gordon to tell me that I have read the RFC myself and worked in the IT industry for years.

Working in the IT industry can mean anything from a janitor at IBM to the President of Salesforce. Neither position is likely to know much about the specific USB audio standard. Thus, I'm not sure where you are going with your statement. Plus, I hate to act as if I know this stuff too well, which is why I recommended asking Gordon or John who may know more about implementing USB audio at a high level that anyone on the planet.

 

One more thing, did you take my comments, directed at someone else, personally for some reason?

 

 

 

 

What people never seem to tell you is that packet errors are unbelievably rare in transmitting packet data. USB can be troublesome because of the noise from PC itself but that almost never translates to errors in packet transmission. What happens most times is the manufacture of the DAC has problems dealing with noise that can come via the USB bus and makes it's way into their DAC. It's how they deal with that noise is what makes the USB implementation work better for some DAC over others.

 

Good, we agree on this as well. I only raised the error correction issue because a previous comment implied that USB audio has error correction.

[The Computer Audiophile]

I meant no offence at all, sorry if my post was misconstrued. I realize you were responding to someone and really was just putting that out for general knowledge. But it probably could have been worded better, sorry about that. For me, being in the IT industry was running networks, server farms and coding applications. Anyway, just trying to get along that USB is pretty robust, more so that pretty much anything we have ever had in the audio world before. And done right can sound pretty spectacular.

Ah, all cool here :~)

[The Computer Audiophile]

USB is packet based just like Ethernet, but USB audio is more like UDP than TCP.

Interesting you mention UDP. Most people think Ethernet means TCP/IP and all audio over Ethernet must use TCP. However, some home audio uses UDP. That said UDP is more than fine for home audio with a decent network.

[The Computer Audiophile]

Hi John:

 

You didn't misinterpret. I misspoke. I meant the Isochronous firmware such as Gordon Rankin's "Wavestream" product. Don't know what made me conflate that with the drivers except perhaps the fact that I had been up all night when I wrote that. USB 3.1 has isochronous operation as part of the spec. While USB 1.0 through 2.0 use a "speak-when-spoken-to" protocol requiring something akin to "Wavestream" on order for isochronous or device initiated communications towards the host, USB 3.0 and 3.1 was designed to allow this without having to add third-party protocols. Sorry for the confusion. I'll go stand in the corner with a dunce cap on my head now....

Somewhat related - I had a meeting with the USB Implementers Forum (USB-IF) at CES and they said Microsoft would be enabling native USB audio drivers for Class 2 components as soon as the USB stack is rewritten.

[The Computer Audiophile]

What's the roadmap for that, or when?

They made it sound like it was happening in the not too distant future. Maybe six months is my guess.

[The Computer Audiophile]

Not necessarily, it could mean there's no correction in addition to the Cyclic Redundancy Check.

Uh OK.

[The Computer Audiophile]

Hey Jud, I spent an inordinate amount of time optimizing the AES interface between my PC and Pacific Microsonics Model Two - I ended up with Weiss and RME interfaces. Anyone who thinks AES is anywhere near perfect should think again - there are massive impedance mis-matches to contend with. Not a good idea in digital audio.

 

No, the only digital interface I have full faith in is properly-terminated BNC. I simply can't understand why this isn't the norm in digital audio. (Well apart from the fact that recording studios had loads of balanced audio cables lying around - not even 110 ohms - so AES in, their infinite wisdom, knocked up the AES/EBU standard.)

 

"Bring back BNC!", that's what I say.

 

Mani.

Some manufacturers prefer the much higher voltage of AES over the impedance of BNC.

[The Computer Audiophile]

My RME interface allowed me to choose between an AES voltage of 3V vs. 7V. I actually preferred using the lower 3V - it produced a smoother, less fatiguing sound.

 

I've spoken to a number of engineers I respect who believe digital audio should be treated in exactly the same way as RF electronics, i.e. using properly-terminated BNC connectors. But apparently, in the early days of digital audio this would have been overly expensive for studios (I don't really understand why, but that's what I've been told). They had loads of balanced XLR audio cables lying around, so the AES/EBU standard was born to utilize these. No need to worry that the audio cables weren't exactly 110 ohms, let alone the massive impedance mismatches of the XLR connectors. The AES guys seemed (still seem?) to think digital audio worked in the kHz range and not the MHz-GHz range.

 

Take all this with a pinch of salt. I'm no RF or audio engineer - it's just what I've been told. But it does seem to fit perfectly with my experience of digital audio.

 

Mani.

Very interesting. Thanks Mani.