Head-Fi.org › Forums › Equipment Forums › Computer Audio › Low-Jitter USB: Dan Lavry, Michael Goodman, Adaptive, Asynchronous
New Posts  All Forums:Forum Nav:

Low-Jitter USB: Dan Lavry, Michael Goodman, Adaptive, Asynchronous

post #1 of 166
Thread Starter 

If I had my druthers, there'd be an easy way to carry a good turntable setup and LP's around, as the best I've heard from analog sources still trumps (to my ears) the best I've so far heard from digital.  That said, I ended up abandoning vinyl a long time ago (I know, sacrilege!), as it simply didn't fit in reasonably with my lifestyle anymore.  Be that as it is, digital audio is something of great interest to me, and, more specifically lately, computer audio.

When Tyll Hertsens stopped by my office several years ago (on his way to New York) with a HeadRoom DAC he wanted me to hear (which was my first experience with a good USB DAC) I decided to begin my personal search for computer audio solutions with the goal of supplanting the spinning silver disc that had long ago displaced the spinning vinyl disc for me years before that.  It has been a fun journey, playing with a wide assortment of gear within a wide price range, that pointed me to where I am now.  (And I know for certain the journey isn't over, and may never end.)

One of my DACs of choice right now is the Lavry DA11 by Lavry Engineering.  Dan and Priscilla Lavry sent it to me on loan, and, quite honestly, all I was really interested in initially was the PIC feature (PIC = Playback Image Control)--a digital soundstage/crossfeed feature that Dan had developed.  What I didn't expect was to be so wowed by the DAC that I all but forgot about the PIC feature (although I do use the PIC feature for the mega-pan pseudo-stereo that was more common in the older stereo recordings, and that I find rather intolerable through headphones).  Priscilla asked what I thought of the DA11, and I asked how much I owed them for it, as I wasn't returning it.  In fact, I dig the DA11 so much I'm probably going to buy another one.

If you ever get a chance to meet Dan Lavry--and you're interested in digital audio--exchanging a few words with him is a must.  I've had the good fortune of spending hours with the man, in person and on the telephone, and, though I'm still a digital audio technical greenhorn, I feel like I've gone from zero to one (pah-dum-pum) from my discussions with him.

Some of the most interesting, accessible posts about digital audio that I've read anywhere have been here, and made by Dan Lavry.  Here are just some examples:

http://www.head-fi.org/forum/thread/449885/usb-to-spdif-converters-shoot-out-emu-0404-usb-vs-musiland-monitor-01-usd-vs-teralink-x-vs-m2tech-hiface/270#post_6170529

http://www.head-fi.org/forum/thread/449885/usb-to-spdif-converters-shoot-out-emu-0404-usb-vs-musiland-monitor-01-usd-vs-teralink-x-vs-m2tech-hiface/270#post_6171581

http://www.head-fi.org/forum/thread/449885/usb-to-spdif-converters-shoot-out-emu-0404-usb-vs-musiland-monitor-01-usd-vs-teralink-x-vs-m2tech-hiface/285#post_6173435

http://www.head-fi.org/forum/thread/449885/usb-to-spdif-converters-shoot-out-emu-0404-usb-vs-musiland-monitor-01-usd-vs-teralink-x-vs-m2tech-hiface/315#post_6175451

http://www.head-fi.org/forum/thread/432442/about-my-design-phillosophy

http://www.head-fi.org/forum/thread/424223/lavry-da10-inputs-which-is-better-xlr-optical-and-best-way-to-get-it-there-from-pc/15#post_5701270

And if you want to see his whole post history here, here's the link:

http://www.head-fi.org/forum.php?action=postsbyuser&id=109259

I mentioned "async" in the thread title, so if you were wondering when I was going to get to that, now's the time (but don't expect me to discuss the technical merits of it, as you can look it up "async USB" or "asynchronous USB" yourself).  As many of you are aware, the asynchronous USB code (developed by Gordon Rankin) is discussed at length on this forum and others (like Computer Audiophile).  Even those in the know rarely argue the merits of what Rankin's doing with async, which is to address jitter-prone USB.  But there seems to be a growing belief among some enthusiasts that async may be the only audiophile-class way to receive USB audio.  It sure had me convinced, and the Ayre QB-9 (for which Ayre licenses Rankin's async USB technology) was on my list of must-hear components, and has been almost universally well received by the audio press.  I've used and heard it, and it sounded excellent.

At CanJam @ RMAF 2009 last October, I brought my Lavry DA11 (among other gear), which I had hooked up to the optical output of my MacBook Pro.  I had the Ayre QB-9 on top of the DA11, hooked up to my MacBook Pro via USB (the QB-9's only input type).  It was only one rig (the rest of the rig being a Ray Samuels Audio Raptor, a heap of some of Nordost's best cabling and power conditioning gear, driving Sennheiser's HD800 with Moon Audio cabling, and other headphones as brought and borrowed by others), but many people listened to both DACs through it.  From my time with those DACs in that rig at RMAF, I observed only two people (of a good number of folks) who preferred the QB-9 to the DA11.  (Again, it was only one rig, so I'm not arguing this is how the outcome would be in other systems, only how it was with that rig then and there.)

NOTE:  I did not have Amarra (by Sonic Studio) running at the time.  Some day soon, I intend to compare the QB-9 and the Lavry DA11 both via USB, and through Amarra, now that I am running Amarra in my system.  (I am an Amarra fan, by the way, so maybe more on that another time.)

Another interesting guy I've met in my audio adventures is Michael Goodman of CEntrance (he goes by mgoodman here).  Several companies, including Lavry, license CEntrance software.  With all the talk of async USB, Michael one day posted the following here on Head-Fi:

http://www.head-fi.org/forum/thread/410565/the-lavry-da11-for-your-ears-only/120#post_6274515

This was in response to the general take by many computer audiophiles that "adaptive" is somehow the opposite of "async," and is, thus, inherently bad or inferior.  Based on what I'd read, I had all but accepted that as a truism myself.

A little bit of backstory on meeting Michael Goodman:  I met Michael when I placed a call to him, out of the blue, to ask him a question, and the conversation just kept going and going.  I like talking digital audio, and Michael is another guy who has much to say about it.  Well, actually, there are a lot of people who have much to say about digital audio, but Michael is one of the handful I've met that has much to say that I find both illuminating and interesting (like Dan Lavry).  We talked about many things over time, and it got to the point that he wanted me to hear his 24/96 USB microphone preamp (the MicPort Pro), which has a built-in headphone output for live mic monitoring.  As a DAC/amp, I told him I thought the MicPort Pro sounded good, not great, and gave him my opinions on where, as a headphone driving device, I thought it could use improvements.  Several calls later, and he told me he'd taken the feedback about the MicPort Pro, and was developing what would become the DACport.

Michael had the stones to show up as an exhibitor at CanJam @ RMAF 2009 with a couple of very rough prototypes of the DACport--fidgety prototypes that required the designer's touch to make function.  (Sadly, one of the prototypes was stolen at the show, but Michael and I both laughed about it, as there's no way that twitchy proto unit was going to be much fun for the thief.)  Anyway, the response to the DACport prototype demos was quite positive.  The DACport has since acquired the polish of a finished product, and was reviewed in the most recent issue of Stereophile.  Even John Atkinson (Stereophile's editor, and the reviewer of the DACport for that issue) seemed to initially approach it with some amount of reservation, as it is an adaptive mode USB audio device, not async.  His subjective review of the DACport, however, was glowing.

 

Just as interestingly, in my opinion--and the primary motivation behind this now very lengthy post--was that John's measurements of the diminutive DACport yielded one of the lowest (if not the lowest) jitter measurement he's published in a digital component review (that I can recall anyway, and I've been reading Stereophile for a long time).  Among the numbers was an estimated 91 picoseconds of jitter peak-peak, apparently below the Miller Analyzer's effective resolution limit--and, again, this coming from a device that is only the size of a partially-smoked Double Toro (and that draws its power entirely from the USB bus, which many would say is not the ideal power source, and power supply design being of absolutely critical importance in good DAC design).  In short, as far as jitter goes, I think John Atkinson's review and measurements helped bolster Michael Goodman's earlier point.

Some who know me here know that I am not really a measurements guy.  Yes, my portable CD player probably measures better in every respect than even an exceptional vinyl-playing setup--but I know which I prefer (vinyl).  Yes, many portable solid state amps probably measure better in most respects than my desktop tube amps, but, again, when it comes to driving most of my favorite non-IEM headphones, I know which I prefer (the tube amps).

So I'm not here to tell you I know what jitter sounds like, or to say that, between two DACs, I'd always prefer the sound of the one with lower jitter.  And I'm not here to debate about async versus adaptive, as what technical knowledge I have on such matters would be depleted in a blink.  I'll let the more learned debate the fine points of all that if they way want to.  I just thought it was interesting that it appears that Dan and Michael are right when they talk about there being more than one good way to deal with the likes of jitter, at a time when a lot of audiophiles were starting to believe (or already firmly believe) that when it comes to low-jitter USB there's only one solid solution.

As Dan Lavry said in one of his posts here (and as Michael Goodman has also similarly said to me in conversations):

"...Yet it matters very much WHERE that jitter is. It is only important to have the low jitter AT THE CONVERTER, right where the digital is converted to analog. That is the "conversion jitter" and that is the jitter that matters. Moving data around can tolerate 100 times the jitter level with no sonic impact. We call that "data transfer jitter". If we have say huge jitter on say the spdif cable, but we get to "clean it" before it gets to the critical circuitry, then we are doing fine..."

These guys, it seems (and others, too, I'm sure), are doing that, and not necessarily the way many of us budding computer audio enthusiasts were starting to believe they had to.  And, to me anyway, that's interesting news.

NOTE:  I haven't heard the production version of the DACport yet (I will very soon, though), so none of this is a commentary on how the DACport sounds.

 

(Of the several companies mentioned in this post, HeadRoom, Lavry Engineering, Moon Audio and Sennheiser are sponsors of Head-Fi.org, at the time of this post.)

post #2 of 166
Quote:
Originally Posted by jude View Post

...I feel like I've gone from zero to one (pah-dum-pum)...


I lol'd.

post #3 of 166

Quote:

But there seems to be a growing belief among some enthusiasts that async may be the only audiophile-class way to receive USB audio.

 

I just thought it was interesting that it appears that Dan and Michael are right when they talk about there being more than one good way to deal with the likes of jitter, at a time when a lot of audiophiles were starting to believe (or already firmly believe) that when it comes to low-jitter USB there's only one solid solution.


I couldn't agree more. There are so many myths in audio, and I fear the asynch myth has already been perpetuated to the point where there are probably a lot of audio enthusiasts who may never believe that synchronous USB can be an equally valid way of transferring audio data via USB regardless of the facts available to them.

 

In all the reading I have done here on Head-Fi, I've learned more about digital audio and DAC's in particular from the posts of a few no-nonsense engineers like Dan Lavry and Elias Gwinn than from all other posters combined. These guys don't rely on or spout the benefits of magical pixie dust, just scientific facts and solid engineering. I also appreciate that they have the professionalism not to belittle or challenge different manufacturers or design philosophies and instead focus their energy on explaining their own designs, the science behind them and why they chose to do the things they do in terms that can be understood by us laymen.

 

I think Head-Fi and this whole hobby would be more enjoyable if there were more people like these guys and fewer people designing and selling products whose only explanation as to why they design their products the way they do or advocate questionable tweaks is because "it obviously sounds better, and if you can't hear the difference your system isn't resolving enough" or the like.

 

Jude, I would be very interested in reading the paper from Michael Goodman regarding jitter and the different transfer modes when it is released, if it isn't already.


Edited by Bmac - 5/21/10 at 1:27pm
post #4 of 166

Of all the DACs that I heard from USB the ones that I found to work the best are Async DACs.  I haven't found one good DAC with USB that is good as any of the Async DACs in the market. 

 

When people are using computer transports (M2Tech) to DACs because the USB input of the DAC is worse than using Coax/Toslink we have a major problem in our hands.

 

I haven't heard of any of the DACs that jude said but I just haven't heard a good Sync USB DAC yet.  Until I hear a good Sync DAC I will say that Async is the way to go. 

post #5 of 166

One must be careful to attribute the better SQ to the right things.

 

Like all audio equipment designs, design and implementation are just as important as the technology and parts used.

 

It is not so surprising that its possible to design an Async USB interface in a DAC that does not sound as good as an Adaptive interface in another DAC.  This has both to do with the design and implementation of the USB interface and the DAC, as well as the parts selection.  Any of these can spoil the mix.

 

Many audiophiles believe that selecting the best D/A chip or selecting a particular technology for Firewire or USB is all that matters.  This is how they shop for components.  This could not be further from the truth.  Everything matters, from the power cord to the power supplies, to the types of capacitors that are used in the power decoupling.  Even the signal coupling capacitors in a S/PDIF interface matter.

 

However, given similar design skills and parts selection, the Async interface will always be slightly better than the Adaptive interface designed by the same designer.  It's simply because the technology is better.  It's physics.

 

And there are lots of Async designs out there such as M2Tech, Musiland and Tascam, and BTW, Gordon did not design the first one.  His is unique because it uses native drivers.  There are also quite a few 192kHz Async designs available now, all using custom drivers.

 

Steve N.

Empirical Audio

 

post #6 of 166

Great read Jude, thanks for the info and links.

post #7 of 166


 

Quote:
Originally Posted by audioengr View Post

One must be careful to attribute the better SQ to the right things.

 

Like all audio equipment designs, design and implementation are just as important as the technology and parts used.

 

It is not so surprising that its possible to design an Async USB interface in a DAC that does not sound as good as an Adaptive interface in another DAC.  This has both to do with the design and implementation of the USB interface and the DAC, as well as the parts selection.  Any of these can spoil the mix.

 

Many audiophiles believe that selecting the best D/A chip or selecting a particular technology for Firewire or USB is all that matters.  This is how they shop for components.  This could not be further from the truth.  Everything matters, from the power cord to the power supplies, to the types of capacitors that are used in the power decoupling.  Even the signal coupling capacitors in a S/PDIF interface matter.

 

However, given similar design skills and parts selection, the Async interface will always be slightly better than the Adaptive interface designed by the same designer.  It's simply because the technology is better.  It's physics.

 

And there are lots of Async designs out there such as M2Tech, Musiland and Tascam, and BTW, Gordon did not design the first one.  His is unique because it uses native drivers.  There are also quite a few 192kHz Async designs available now, all using custom drivers.

 

Steve N.

Empirical Audio

 


Couldn't agree more with what is being said here, but please, can we leave the power cord out of the game...

post #8 of 166

Michael Goodman @ Bmac: The paper is currently available as a design note, which you can read here:

 

http://centrance.com/products/dacport/

 

Click the "Design Philosophy" tab. I wanted to make it easier for you, so here is a slightly rephrased version of that note:

 

Jitter Management - Debunking the Asynchronous Myth

Some manufacturers may lead you to believe that Asynchronous USB transfers are superior to Adaptive USB transfers and that therefore you must believe in the asynchronous solution. This no more true than saying that you "must" hold the fork in your left hand. In fact, if you know what you are doing, you will feed yourself with either hand. The issue is really about good engineering practices.

x.gif

Background: There are two popular ways of sending digital data from the computer to a peripheral device. The argument comes down to jitter management and goes as follows: In Asynchronous mode the device is the clock "master", meaning that the device generates the clock and the computer follows it as a "slave". In Adaptive mode, the computer is the clock "master" and the device follows it as a "slave". The truth of the matter is that either method works fine, if correct design principles are followed. Jitter is the result of Irregularity of the data flow. If the data flow is irregular, then jitter is high, if the data flow is regular (well-controlled), then jitter is low.

 

Tricky Detail: Here is the tricky detail that often gets omitted in discussion: No matter which side is the source of the clock (Computer or DAC), the two devices are still joined together by the USB cable. The digital data on that USB cable is always irregular. It is irregular simply because the computer is involved. Computers do many things at once and end up sending data over USB in imprecise intervals, no matter who is the clock master on the bus. This irregularity causes jitter. So, since the computer is involved in the transmission there simply cannot be jitter-free traffic on the USB bus. Claiming to have a jitter-free solution is just like claiming to have a dust-free house. Irregularity always creeps in and needs to be actively managed. How well it is managed is the question.

x.gif

DACport jitter cleanup

And the Argument Breaks Down: Here is where the Asynchronous vs. Adaptive argument breaks down: In either of the two clocking schemes, jitter is present during the USB transmission. It's inevitable and also ok, if it is properly cleaned up prior to the D/A conversion, where it matters most. Neither clocking scheme is superior during transmission. Both schemes are capable of performing very well if you know how to reassemble the bits prior to Digital to Analog conversion. It doesn't matter that audio samples arrive from the computer in imprecise chunks of data, as long as they are straightened out before the DAC. The most important part is to make sure that samples arriving at the DAC are clocked accurately. DACport employs JitterGuard™, a proprietary two-stage clock management system that does just that - cleans up the jitter on the USB bus so that samples are virtually jitter-free at the D/A conversion point. The result is a natural responce, crisp definition, extra resolution and wide soundstage.

 

John Atkinson from Stereophile listened to DACport and found no problems with Adaptive, proving that the difference is just a marketing tactic -- a myth, an urban legend.


Edited by mgoodman - 5/23/10 at 4:25pm
post #9 of 166

There is an interesting article in this months Sound On Sound called    #Does Your Studio Need A Digital Master Clock .

I mention it because there is a section called Interface-Induced Jitter which explains things quite well to the lay person, (that's me )

post #10 of 166

So, it does not matter if you use asynch, synch or adaptive as all can successfully cope with jitter depending on the implementation.

 

Is it the case that adaptive was not that successful at first, then buffered synch and asynch showed the way and now adaptive has caught up?

 

I am generalising.

post #11 of 166



 

Quote:
Originally Posted by Prog Rock Man View Post

So, it does not matter if you use asynch, synch or adaptive as all can successfully cope with jitter depending on the implementation.

 

Is it the case that adaptive was not that successful at first, then buffered synch and asynch showed the way and now adaptive has caught up?

 

I am generalising.


Here is the history:

Adaptive was first available in products using PNP chips like the PCM270X, which have atrocious jitter, even when using good clocks.  This made adding USB to a product easy, but gave Adaptive a bad name.  Then there were toys like the Transit that used the TAS1020.  This also had atrocious jitter.  Then a handful of manufacturers licensed the CEntrance code to use the the TAS1020.  Some designers did a good job with this and used good clocks and the results were stellar.

 

In the mean-time, Gordon developed his version of async, also using the TAS1020 and proceeded to tout Adaptive for a couple of years until most everyone bought his argument.  He then licensed a couple of other manufacturers on his technology.

 

Even at that point in time, there were Adaptive solutions that met or beat the Async.  Implementation is everything.  Unfortunately, Stereophile did not get their hands on these solutions until now.  TAS did however, with the resulting Golden Ear awards. 

 

Its late in the game now and even I am changing to 192 Async, but not from Gordon.  Customers want to be future-proof, so they insist on 192 capability.

 

192 async comes with some baggage BTW:

 

1) requires 2 clocks, so more expensive

2) requires latest PC or Mac and USB2.0

3) requires low-latency I/O system

4) requires custom drivers (not a big deal IMO)

 

Steve N.

Empirical Audio
 

post #12 of 166

Being a Windows programmer I am amazed at the ability of today's top tier USB DAC vendors (software and/or hardware both) to deal with both the difficulties and moving target of the Wintel USB Endpoint. There has been a lot that has changed between XP and Win7 and it must be very difficult to decide on what to adopt in the capabilities of Win7 (Like HD Audio interface w/ 3ms data pump over HDMI) and still have to drag along legacy interface heuristics from XP (KMixer with 10ms data pump?).

 

mgoodman does an excellent job of demonstrating the aligning of data to clock in his description and diagramming and I think it helps paint the basic picture for people who are trying to embrace the concept and this thread as a whole (IMO) should be a sticky for noobs that are dealing with USB/DAC devices on here, as it takes quite a while to wrap your head around the various layers of what is going on when you click play in FooBar (or my case j.River Media Center w/Direct Sound) and data is sent to your DAC/AMP and into your eagerly awaiting eardrums.

 

The detail on how this all comes together is almost a fractal in both beauty and hardship as many problems have to be addressed and dealt with the re-assemble the data back to as complete a package as possible before the D/A conversion takes place.

 

One area which lends itself to the fractal analogy is the data stream and clock timing as it gets exceedingly more difficult to get those bits organized into packets again as there is a latency window within the clock/sync/stream transmission itself. What I'm referring to is the process of aligning the outbound stream of sync data (as side band information for synching data to clock) to the clock signaling. This dance of data, clock and sync data itself has inherent latency issues just getting the data pump (either Push or Pull method) out the port even before you try to reassemble it back using the proprietary techniques described in this thread.

What I'm saying is that we are going to be experiencing some form or another of jitter (I know we are talking pico-seconds here) for some time, under the method that Windows uses (only talking about Windows here) matures together with Intel's standards (and yes, AMD) and other methods evolve to allow the waveform to be handled as a true stream and not as a data packet synchronized to a clocking window.

post #13 of 166

Afaik, the gamma2 has such a de-jitter "stage" just in front of the DAC. Can be built from $100 to maybe 200 bucks depending on the configuration/features.

Also I see that those proprietary "technologies" are used for marketing..

 

@DannyBuoy: Isn't normal usb audio streaming a simple matter of chopping audio into 1ms chunks? The OS is doing most of the work. Also, most usb receivers are supported out of the box on many OSes. 

post #14 of 166

Analogous to chopping a rope of sausage into 1/4 slices and then assembling them back into the length of sausage you would have a hard time matching the slices together because of the damage done by slicing it up in the first place.

The point I was making is that there is leading edge and trailing edge sync issue by nature of taking a bit stream and dividing it up in slices to be reassembled by clock-sync technology. bits are lost (discarded) or padded (Null padding) according to the SDO/BClk data at the block interface.

This has nothing to do with support of the device at the Endpoint (USB Interface in this case) it was a comment about the Microsoft USB Driver which Microsoft developed following the Intel Standards for how to slicing that data so it can be passed out the USB interface and into the hands of people like mgoodman to try and reassemble based on the clock data sent along the stream along with the clock sync of the USB DAC.

I was saying that it still is a best effort no matter what because the USB Interface driver to begin with is not doing a 100% accurate job to begin with. I was hoping it would help people see that you are still bound by the operating systems limitations to commit 100% of its time to this duty and instead a push or pull technique is used along with buffers and blocks to give it its best effort in getting our music out the port with as little impact on it as it can.

Unfortunately the MS USB Driver is still at the 1.0 Standard even in Win7, I thought they would have jumped to the 2.0 Standard by now. (I am talking about the USB specification not the more familiar speed standard people are familiar with). That driver is what the plug and play devices like a DAC must connect with and it is a legacy standard for sure.

 

It takes an awful lot of time to write the type of firmware that does what is on these USB DACs and even though I have been a Windows programmer for 15 years, I would still not like to take it on as there are so many moving targets to try and keep together at that level. To over simplify; it is not difficult to take the USB data packets and forward them to the DAC, it is exceedingly difficult to do it with a desire to have it as precisely controlled as the best of them do.


Edited by DannyBuoy - 5/25/10 at 2:35pm
post #15 of 166

Argh! Now something else for me to stress about (darn OCD!). So when is MS going to fix their USB driver so it meets the latest 2.x spec??? I am surprised that this did not happen with Windows 7.  /sigh

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Computer Audio
Head-Fi.org › Forums › Equipment Forums › Computer Audio › Low-Jitter USB: Dan Lavry, Michael Goodman, Adaptive, Asynchronous