Asynchronicity: A USB Audio Primer

Take it easy Chris! - I LIKE research, and I LIKE high end audio. You completely misinterpreted my comment! <br />

<br />

You like to challenge people who make assertions by asking for the research to support it. In this case I was responding in a positive way with the "bada bing" comment! The writer responded with the research you were requesting - you probably were not expecting that! I did read the research and felt it added to the discussion. <br />

<br />

I think you need to take a look at how much of a control freak you are, unless you just want writers that tell you and your little sponsors how great you are!<br />

<br />

cheers<br />

Eric

Eric - Your response was exactly what I was expecting. It also says much more about you than it does me. Personal attacks like calling me a control freak are very sophomoric, especially when you include statements like, <i>"... you and your little sponsors ..."</i>. This is classic playground terminology that only takes away from any point you were trying to make. I take pride in the fact I am not a control freak and I encourage all comments whether they agree with me or not. In fact I even added a sentence about that in the article. Take a look around the site here, you'll see many comments that disagree with what I say and add much to the discussion. A one-sided site ads nothing and would be a disservice to this wonderful hobby and hobbyists. I am no preacher and the readers of CA are certainly not in the CA choir. <br />

<br />

Asking for research is certainly not challenging people. It's actually trying to get more information in front of those who seek it instead of relying on hearsay. I am actually very happy the links to the research were provided and I expected the links to be provided. Your comment about reading the research and feeling that, <i>"it added to the discussion"</i> was close to something I would have said in college if asked a question about research I really did not read.<br />

<br />

Saying you responded, <i>"in a positive way with the "bada bing" comment"</i> is a little misleading. I appeared to me as more of a "gotcha" like you thought I was burned by the fact that research links were posted. It also appeared to me like you rubber stamped the fact that links were provided without reading the material and doing more research which you are a big fan of. More research into this topic including counter arguments to the papers provided clearly leads one to conclude there is no consensus on this topic. I asked for the link from the reader because I wanted to make sure he was not just saying something he had heard through the grapevine. We all know forums are full of ideas that live on forever often repeated without any knowledge of the origination of the idea.<br />

<br />

Thank you Chris<br />

<br />

It helps clarifying what we are talking about when discussing USB DACs.<br />

A part from Wavelength, DCS and Ayre, are you aware of other manufacturers having developped asynchronous USB interfaces, or being in the process of developping it?<br />

<br />

Thanks

Hi Fyper - NObody I know has developed an asynchronous USB DAC at this level. The EMU 0404 is technically asynchronous via USB, but it's design and implementation are 180 degrees different than Ayre, Wavelength, and dCS. There is a reason the EMU is only $200 even though it's one of the only async USB DACs in the world.<br />

http://www.bd-design.nl/contents/en-us/d168.html<br />

<br />

Which I think existed earlier than you heard about async USB ...<br />

I worked on the software myself in 2006.<br />

<br />

I like to have this as an example that it is not (or at least not always) the big audio companies who are innovative, but the power of high qualified individuals is.<br />

You could call this one obsolete already (16/44.1 only), but there will be more ...<br />

<br />

Interesting Peter. It appears to be using Bulk mode similar to FireWire. What chip does it use?

Peter,<br />

<br />

Bernt is a good friend of mine and yes this would be considered an asynchronous device. But it is not an ISO Asynchronous device and therefore would require drivers.<br />

<br />

I would not call 16/44.1 obsolete a number of my customers think the dynamics of the NOS style dacs sound better than current 24/32 bit units.<br />

<br />

Thanks<br />

Gordon

Chris ... you comment on using USB Probe - is this a piece of software I can download from somewhere??<br />

<br />

Eloise

AD1865.<br />

<br />

I don't consider it similar to Firewire. It uses an internal buffer of 64KB (and uses USB 1.1 IIRC).

Good to hear. All I know is that I told Bert to find you at the last RMAF and talk a bit about things. I don't think that really happened, you being quite busy.<br />

<br />

It is true that it requires its own drivers, and for that matter its own player software. Not a good thing by itself.<br />

<br />

<strong>a number of my customers think the dynamics of the NOS style dacs sound better than current 24/32 bit units.</strong><br />

<br />

At least I try to keep that up. (but for 24/192) :-)<br />

<br />

Peter

Peter,<br />

<br />

Yes he did stop by at RMAF and we talked for a couple of minutes but it was mass hysteria at the room.<br />

<br />

BTW you mentioned that small companies are not the driving force here. I would have to put WA as one of the driving forces and I am not a large company. Actually if you look at computer audio none of the large companies are really prepared for it. They are all using technology from some other company.<br />

<br />

In regards to the 24/32 @192.... got me I think a number of people are ripping their cd collection and thinking it's 16/44.1 why do anything more than that.<br />

<br />

Also the AD1865 dac chip is an 18 bit part used by Audio Note also. He could just write a ASIO wrapper for .NET and be compatible with a lot of applications.<br />

<br />

Thanks<br />

Gordon

Thanks. Hoping that Chris allows for a little more offtopic :<br />

<br />

It seems that I suggested that only the big companies can get these kind of things done, but I actually wanted to say just the opposite (also still having in mind the slaughter of A.J. and T.F. at me stating similar before) : the real progress is made by the small companies and even individuals. There is logic in this I think;<br />

<br />

Small companies who give all their heart to the product, and individuals just the same, may work 1000s (yes thousands) of hours to get something done, which would not be allowed by the big companies who just have to pay for that, possibly by hireing expensice contractors. The easy example is myself. I must be somewhere at 6000-7000 hours on the development of XX, and against my normal hourly tariff even my own company (ehh, owned by me) would not have paid for such an investment. Not with a selling price of EUR 72, and not with a selling price of EUR 999. It would require sales amounts of over 10,000 and 750 units respectively, and neither is feaseable (on beforehand, when decisions for investments must be taken).<br />

<br />

Another kind of example would be the development of a loudspeaker. So, talking about Bert, I must have spent 500 hours at least on helping him with let sounding a horn not as a horn. Again against my tariff he sure wouldn't have spent that. Helping as a friend does wonders here, and friends are never big companies (unless something would be in it for myself afterall).<br />

<br />

So, the real innovative stuff not only really happens with individuals (who just need to be very good at their job, like electrical engineering), but it also is logic it goes like that.<br />

Of course these things happened at the large audio companies in the past, and it just may these days. But economy is not really challenging right now.<br />

<br />

Peter<br />

<br />

<i>Chris ... you comment on using USB Probe - is this a piece of software I can download from somewhere??</i><br />

<br />

This is where I get it:<br />

<br />

http://developer.apple.com/hardwaredrivers/download/usbdebug.html

I noticed BD design lists using any USB cable as a feature. Probably the case though the I imagine the esoteric cable companies will disagree. I also look forward to more ethernet DACs.

But of course it is true that just a random cable can be used, since this way of using USB is just about data (and not about audio !).<br />

<br />

Do you care about the USB cable when you, say, download pictures from your digital camera ?<br />

<br />

Peter

While the innovation comes from the small companies I would suggest that it won't take long for the big companies to get involved if they see a market. Perhaps 2010 will be an interesting year for USB audio.<br />

<br />

Hi Peter - nice to see you about!<br />

<br />

Best<br />

Andrew

It will be interesting to see which way the specialist HiFi manufacturers go.<br />

<br />

So far we've seen Bryston and Audio Research go the route of basic USB connection which can easily be improved on. Ayre go the route of licensing WaveLength's async USB technology and soon will have Naim launching their new (first) DAC on which any form of dedicated computer interface is conspicuous by it's absence, preferring to support optical or SPDIF either direct to the motherboard or via third-party interface.<br />

<br />

Eloise

It's hard to generalize and implementation is everything but if we just look at the DACs on the CASH List the interfaces that have provided sonic superiority are:<br />

<br />

Berkeley Audio Design Alpha DAC – AES<br />

Weiss Engineering Minerva – Firewire<br />

Devilsound DAC – only Adaptive USB offered<br />

Bryston BDA-1 – AES<br />

Benchmark Media Systems DAC1 HDR – AES<br />

Wavelength Audio Proton – only Asynchronous USB offered<br />

<br />

IMO the excellent digital interface cards and software drivers with the Lynx AES16 and Juli@ give them the slight advantage. The CA reviews of these DACs seem to confirm. What I find surprising are the FW DAC owners that use their DACs such as a Weiss DAC2 or FireFace 800 as input to the Berkeley DAC. What is noticeably absent in the above is toslink, which is almost universally said to be an inferior interface in terms of sonics but which many manufacturers seem compelled to implement for CD transports and legacy systems. And obviously the best USB has to offer is excellent 24/96 but nothing higher.

The question of whether or not you need to do format conversion and where the clock resides in the chain seems very important but often gets confused in these comparisons. A lot of apples and oranges discussions. <br />

<br />

A comparison I would pay money to hear is Lynx -> BADA (or Bryston or whatever) compared to Ayre USB DAC compared to Firewire directly to Weiss Minerva/DAC2 compared to Firewire to Metric Halo ULN-2 or 8. No interface conversions, but still apples and oranges in that the DACs are entirely different. One of the things I am curious about is, independent of the DAC, does USB have "a sound"? Does Firewire? <br />

<br />

Does the Lynx card output AES natively? <br />

<br />

When feeding a BADA or Bryston DAC, is the Lynx calling the shots (clock wise), or is the DAC the master clock.?<br />

<br />

Hi Dan - When using a lynx with the Alpha DAC or Bryston BDA-1 there is no master clock. The Lynx and each DAC would user their clocks. This is because there is no word clock in or output on the Alpha or BDA-1. The dCS Paganini I am listening to right now has word clock in and out and the Puccini u-Clock has four word clock outputs. Thus I am clocking the Lynx and the Paganini via the U-Clock (master clock in this configuration) when under 96 kHz.

For completeness, could an explanation of the use of Bulk USB in audio be given? I have in my mind some idea about this being used as the main USB protocol for external disks but I think it is used in audio too just can't remember in what products. Does it simplify USB communication, not needing asynchronous communication by using a big buffer & reclocking the data out?

JKeny,<br />

<br />

Consider the amount of time you are investigating this why not just go to USB.org and look it up.<br />

<br />

Bulk cannot be used in Audio without a driver. If you use either of the bulk or interrupt endpoints you would have to write a driver.<br />

<br />

It actually complicates things using bulk and interrupt because there is no guarantee that the packets will reach the endpoint in a continuous time period.<br />

<br />

Only Isosynchronous USB is guaranteed to support Audio and has the highest priority on the buss.<br />

<br />

Thanks<br />

Gordon

Just confused by reading various post regarding asyn usb implementation. I think it is implemented in some control feedback to adjust USB to send faster or slower to match with the DAC clock. But what that means is that the incoming data rate is constantly being adjusted and always slightly different from DAC clock. Is it being resampled? What would DAC do to handle this problem. I'm thinking of CD player, the CDP is constantly adjusting the CD spinning rate to have a constant data stream. This process may create a lot of jitter.<br />

<br />

Another question, if DAC clock is slower than the speed of player software. With Asyn USB, the USB transmission follow the DAC clock. So we have a player software keep sending data with its own speed, which is faster than the USB transmission rate. What the PC is going to do to handle it?<br />

<br />

I think if the player software also slave to the USB transmission speed, it should be fine. But how to make a player software slave to USB transmission speed? Is there such a player.<br />

<br />

If we can test whether wavelength DAC is bit perfect, it is going to help understand whether the implementation slave the audio play and USB from end to end. If not, then there would be some resampling by PC which may impact audio quality. <br />

<br />

Anybody test whether Wavelength is bit perfect with USB interface?

<i>"Anybody test whether Wavelength is bit perfect with USB interface?"</i><br />

<br />

I'm not sure you are asking the right question here. There is no way to test such a thing because the output of all DACs is analog. Thus, the bits are "gone."

<strong>if DAC clock is slower than the speed of player software. With Asyn USB, the USB transmission follow the DAC clock. So we have a player software keep sending data with its own speed, which is faster than the USB transmission rate. What the PC is going to do to handle it?</strong><br />

<br />

Either async or isync, the player sees a buffer. When the buffer is half empty (or whatever is decided for) the player receives a message. Next the player becomes active and fills the buffer. This goes on and on.<br />

With the async that needs a driver (see Gordon's post) the buffer is at the DAC side (in de DAC box).<br />

<br />

Peter