Jump to content
IGNORED

New Dichotomy - Async vs. Non-async DACs


cfmsp

Recommended Posts

Gordon, That's a very interesting analysis of the Musiland DAC but is it not just repeating what JS said above?

 

Yes it only works on PC, at the moment, but s one of the most stable drivers I or Fred (fmak) has experienced - basically anything can be run on the same PC without pops or clicks!

 

Now if your friend Charles would just clear up his misleading posts with regard to the jitter spec of the AP SYS2722, some clarity might prevail? Is the Sys2722 noise floor 1000ps as he stated on Diyhifi in relation to the Musiland or is it 120ps as he stated in relation to his Ayre QB-9.

 

A bit of a discrepancy, I'm sure you'll agree!

 

The M2Tech HiFace is another USB asynch device that seems to provide great sound in a value for money package.

 

Link to comment

Please keep in mind that async USB-SPDIF adaptors are not async USB DACs as JS pointed out clearly.

 

"From theoretical grounds the best way to get the computer data to a DAC chip is to have low jitter oscillators right next to the DAC chip generating the timing for the chip and that clock also being used to control the rate at which data comes from the computer so it matches the rate at which data goes into the DAC chips."

 

IMHO, the best async audio-playback solution is to have a single (zero-jitter) clock as a reference and an ability to control data flow from a computer. It doesn't matter whether to use USB, Firewire or Ethernet, regardless of different engineering effforts and costs.

 

Link to comment

bordin,

Problem is you can't have just one clock if you want to cover 16/44 to 24/192 - you need two low jitter clocks & be able to switch between them. It was said elsewhere, that all these asynch, adaptive, ethernet, firewire, etc are all just transport mechanisms & once implement properly, don't particularly matter once the audio clock is local & low jitter.

 

Link to comment

 

"Problem is you can't have just one clock if you want to cover 16/44 to 24/192 - you need two low jitter clocks & be able to switch between them."

 

Agreed, but I think when people refer to 'single' clock, they mean that only the DAC's clock is in play, i.e. the DAC's clock is NOT trying to track a clock upstream.

 

" It was said elsewhere, that all these asynch, adaptive, ethernet, firewire, etc are all just transport mechanisms & once implement properly, don't particularly matter once the audio clock is local & low jitter."

 

Agreed so long as he condition stated above holds. I thought that adaptive USB required an upstream clock?

 

clay

 

 

 

Link to comment

Jkeny,

 

Gordon, That's a very interesting analysis of the Musiland DAC but is it not just repeating what JS said above?

 

Actually what I was saying was that this is no big feat. John did not mention how they did it and that was what I was clearing up.

 

Yes it only works on PC, at the moment, but s one of the most stable drivers I or Fred (fmak) has experienced - basically anything can be run on the same PC without pops or clicks!

 

So??? I would hope this would be the basis for any audio device.

 

Now if your friend Charles would just clear up his misleading posts with regard to the jitter spec of the AP SYS2722, some clarity might prevail? Is the Sys2722 noise floor 1000ps as he stated on Diyhifi in relation to the Musiland or is it 120ps as he stated in relation to his Ayre QB-9.

 

JKeny, if you really do not know how the testing works, then why even comment on it. First off who would you believe Charlie or the guys on Diyhifi? I mean at least Charlie owns one of these and has used it. Also you really don't even understand what Charlie was saying because the reference 120ps did not have anything to do with AP device. It had to do with the Miller Audio Test Set. This device owned by Stereophile has a minimum reading of 120ps. The Miller on one had gives a numerical readout whereas the AP, Prism and others do not because it is really impossible to give a consistent number.

 

So what Charlie was saying was that the QB-9 had less than the measurable value of 120ps of jitter that the Miller Test set could test. Therefore JA would test it on the AP27xx series and show the FFT results of this test.

 

The M2Tech HiFace is another USB asynch device that seems to provide great sound in a value for money package.

 

But since it uses PLL instead of two fixed oscillators and outputs via SPDIF the jitter will be significantly higher than a well implemented Async USB DAC. Most really good PLL's and even excellent ones will have more than 50x the jitter that a fixed oscillator will have. Some of this jitter does not matter since it will be high frequency crap easily filtered out. But then with adding SPDIF into the quotient the jitter will raise again because of the way SPDIF works.

 

For another $3 dollars they could make the unit work a lot better. For another $15 they could make the device stellar.

 

Thanks

Gordon

 

 

Link to comment

JKeny,

 

Problem is you can't have just one clock if you want to cover 16/44 to 24/192 - you need two low jitter clocks & be able to switch between them. It was said elsewhere, that all these asynch, adaptive, ethernet, firewire, etc are all just transport mechanisms & once implement properly, don't particularly matter once the audio clock is local & low jitter.

 

Not true... if the interface is not asynchronous then the clocks have to move and that creates jitter. Any Adaptive and any Firewire using native device drivers would fall into this category as well as any SPDIF.

 

Thanks

Gordon

 

Link to comment

 

then we're in total agreement.

 

I don't quite get 'bulk mode' as it relates to Firewire, and it's certainly not as sexy sounding as 'asynchronous', but as far as I'm concerned, I don't care what the 'mode' is called, all I know is that my DAC uses Firewire and it's own local clock (when set to 'internal' clock) and the results are fabulous.

 

As near as I can tell, there is no official async Firewire protocol (per Gordon), but that some (and definitely not all) of the pro audio manufacturers write Firewire drivers to control the flow of data from source WITHOUT reliance on upstream clocking info. That's what I care about - high quality clock(s) as close to the DAC chip as possible, clocking the digital data exclusively. I read that from Bob Katz when I first started researching computer audio, and nothing I've heard/read since has strayed me for a second since.

 

Is there a more precise way to say that than Asynchronous Firewire?

 

thanks

clay

 

 

 

 

Link to comment

Gordon,

So first off, do you agree with Charles Hansen's statement that the noise floor of the AP SYS2722 is 1000ps? Can we get some clarity on this, please??

 

Here's what he said on AA

JA has switched to using an AP 2722 to do the spectral analysis for his jitter plots. The noise floor of the hardware is probably around 20 dB (10x) lower than the National Instruments card. But the software won't spit out a number.

 

So he has a choice of giving an accurate graph of the jitter spectrum or else using software that only works with cards that are so noisy that they corrupt the measurement. He has apparently chosen to forgo the number in favor of the accurate graph.

 

It's plain to me what he is saying about the better noise floor of the AP2722 & that more accurate graphs are produced from it.

 

So what Charlie was saying was that the QB-9 had less than the measurable value of 120ps of jitter that the Miller Test set could test. Therefore JA would test it on the AP27xx series and show the FFT results of this test.

Surely the limit of the jitter measured with the AP2722 is 1000ps and above - about 20dB higher than 100ps. How can you say anything about the Ayre QB-9 being less than 120ps measured with this equipment or are we being misled?

 

I was looking for clarity on this but you don't seem to have provided any

But since it uses PLL instead of two fixed oscillators and outputs via SPDIF the jitter will be significantly higher than a well implemented Async USB DAC.

 

The M2Tech HiFace has two fixed oscillators!!

 

Link to comment

JKeny,

 

Again you don't know what you are talking about. Noise floor is measured in dB not ps.

 

I think the problem here is not what Charlie wrote.... but what he did not write. He assumed you knew something about the testing equipment and how units were tested.

 

The Ap can test SPDIF jitter to 1000ps PP and Data. This has nothing to do with testing a DAC as it does not have any SPDIF. So the best way to test a unit like this is with side bands. The problem is the testing that Stereophile uses is based on Juliann Dunn's SPDIF test methods. Since we are not using SPDIF this testing actually does not apply but still gives a good indication of the capabilities of the equipment. Remember the Ap measurement tools do not spit out a number. Why don't you read up a bit on how this works. There are a ton of documents and several papers by Mr. Dunn that would be worth reading:

 

http://www.tnt-audio.com/clinica/jitter2_e.html

 

As far as misleading... believe me the QB-9 has significantly less jitter than 120ps, more like less than a quarter of that.

 

Thanks

Gordon

 

Link to comment

Gordon,

Rather than trying to score points, (jitter is also known as phase noise & plotted in units of dBc/sqrt(Hz) or should that be dBc/Hz, who cares :)) - why don't you answer the question. I will re-quote again here what I said a number of posts back:

Charles Hansen stated on DiyHiFi about the AP2722 that http://www.diyhifi.org/forums/viewtopic.php?p=40486#p40486 "AP only specs that model to have a jitter measurement floor of 1000 psec"

 

As you prefer his more exact wording - Do you agree with his statement? Please answer this directly for the sake of clarity for all readers here who look at the Stereophile published jitter spectrum graphs! You have avoided answering this question, so far.

 

Please then explain how equipment of this stated limitation can show jitter spectra for anything below 1000ps? Please don't try to obfuscate a simple question.

 

Let's keep the personal attacks out of our posts from now on. I may make basic nomenclature mistakes every now & then but it doesn't negate my question which you have now avoided more than once. Your use of technical exactitude is a cloaking mechanism for avoiding a straight answer. And the use of your clichéd statement "I think the problem here is not what Charlie wrote.... but what he did not write" is a primitive avoidance technique.

 

So come on for your own credibility, here & elsewhere, either you or Charles answer the question.

 

BTW, are you the spokesman for Charles or just doing point duty on this foray? Why doesn't he answer directly himself?

 

Link to comment

Let me try - nobody can say I have an axe to grind.

 

The Audio Precision system can make all sorts of measurements in audio systems. One measurement it can make is jitter of digital streams that are fed to it as SPDIF signals. The test limit there is 1 nS, over the audio band, as detailed in the product literature.

 

http://ap.com/downloads/file/143

 

Since both one box CD and DVD solutions as well as some external DAC solutions don't use SPDIF as a data transport scheme, a different approach is needed and used by some (like Stereophile) to measure the *results* of jitter induced conversion distortion. This is what you see in those graphs in Stereophile reviews. The test is described here:

 

http://stereophile.com/features/1208jitter/

 

It so happens that John Atkinson of Stereophile uses an AP SYS2722 as his audio spectrum analyzer for this test. It's coincidental that the two test regimes use the same base piece of test equipment.

 

 

 

 

 

Link to comment

I've just caught up on this thread.

Legitimate question - how can a device with a lower limit of 1ns give meaningful data beneath that?

 

I suppose I should say at the outset I'm not invested in the question. But I am concerned by the tone. It is not what anyone likes to see.

 

@Gordon - it's up to you but you might want to consider a more generous approach in your replies. It doesn't reflect well on your undoubted expertise and knowledge.

 

@jkenny - it's up to you but you might want to consider a less confrontational tone in your posts.

 

This isn't US Foreign Policy since the 50's and it's impact on current geo political stability we are debating here - just some measurements which may or may not be valid and may or may not translate into a dac you like the sound of.

 

 

 

Best Wishes

Andrew

Link to comment

I've just caught up on this thread.

Legitimate question - how can a device with a lower limit of 1ns give meaningful data beneath that?

 

I suppose I should say at the outset I'm not invested in the question. But I am concerned by the tone. It is not what anyone likes to see.

 

@Gordon - it's up to you but you might want to consider a more generous approach in your replies. It doesn't reflect well on your undoubted expertise and knowledge.

 

@jkenny - it's up to you but you might want to consider a less confrontational tone in your posts.

 

This isn't US Foreign Policy since the 50's and it's impact on current geo political stability we are debating here - just some measurements which may or may not be valid and may or may not translate into a dac you like the sound of.

 

 

 

Best Wishes

Andrew

Link to comment

Andrew,

I will take your advice - I guess I'm just frustrated asking the same question since Tuesday. I'm not frustrated that there hasn't been an answer from Charles (I don't expect instant replies) but rather that there have been replies from Gordon that do nothing to further the understanding or throw any light on it, the opposite, in fact

 

Anyway, that's just me & my frustration!

 

Link to comment

JKeny,

 

"AP only specs that model to have a jitter measurement floor of 1000 psec" As you prefer his more exact wording - Do you agree with his statement?

 

No, again as CG pointed out this is the specification they have for SPDIF JITTER not for computational jitter from the FFT analyzer window.

 

Ok here's the math... the Miller unit that JA use to use was a 16 bit board 90dB SN ratio AD converter and Miller Claimed 120ps. Now this is far fetched because even though with the best Windowing (Prism 7) this is more like 260ps. Anyways... Paul Miller (he designed the Miller QC Suite used by JA and other magazines) went onto to say that using a 24 bit AD like the one in the AD27xx or their new 24 bit board would yield a respectable 400fs. Now again that is pretty far fetched and I would put that number at more like 1ps.

 

Opps sorry seems Charlie and I are both wrong it's 130ps not 120ps for the Miller 16.

 

The QC Suite measures audio jitter in a different fashion to the AP. Our 16bit measurements are good to ~130psec (limited by the 16bit data pattern) while our 24bit jitter measurements are good to ~0psec.

 

Zero of course is just silly as there never would be anything close to 0, there always will be something.

 

What I meant about what Charlie did not write is exactly what I am trying to explain. But if you don't listen you will not understand anything but what you want to. Which so far seems to be incorrect.

 

It would be easier for you to read the Julian Dunn information and learn how to calculate it get an AP, Standford or Prism unit and start testing yourself.

 

and Yes I know all about Phase Noise as I deal with it all the time.

 

Thanks

Gordon

 

Link to comment

the post from cg.

 

The 1 nS threshold for the Audio Precision device in question is for measurement of the SPDIF data stream, one would use this to measure the jitter present at the output of a transport by plugging the SPDIF cable into the AP and measuring.

 

The jitter (expressed as noise) spectrum analysis done by John Atkinson (on the same test machine, it is a multi purpose audio testing device) is entirely unrelated to testing SPDIF jitter levels as above. DAC jitter level is observed from the noise spectrum of the analog output of the DAC when the input to the DAC is a test signal designed specifically to show jitter expressed as noise in the DACs analog output. This test procedure has nothing to do with the 1 nS threshold of the test above.

 

If you are interested in learning to understand how JA measures jitter for tests published in Stereophile, please go to the Stereophile web page, articles, and read JAs most recent article on measuring jitter.

 

SO/ROON/HQPe: DSD 512-Sonore opticalModuleDeluxe-Signature Rendu optical with Well Tempered Clock--DIY DSC-2 DAC with SC Pure Clock--DIY Purifi Amplifier-Focus Audio FS888 speakers-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Purple Fuses, Spacetime system clarifiers.  ISOAcoustics Oreas footers.                                                       

                                                                                           SONORE computer audio

Link to comment

In the time it took for me to write and edit the following, there were two good or better posts on the same subject. Don't read this - save your brain energy for something else. If I knew how to just delete the post, I would...

 

In this case, the Audio Precision test system is not used for its however limited ability to directly measure jitter in a data stream. Instead the audio spectrum analyzer functionality is used. So, that 1 ns measurement limit doesn't apply. Any other suitable audio spectrum analyzer could have been used, but here Audio Precision was used. Hey - it's a really good audio spectrum analyzer and if it's sitting on your test bench, it's a good choice.

 

The test Stereophile uses is described in that article previously linked to. It was developed by Julius Dunn, a really bright engineer who passed away a few years ago. What's done is that a CD is played in the case of a stand alone player, or a digital track is played in the case of a separate DAC, that has a single high purity tone precisely at 1/4 of the sampling rate used for the medium. On playback, this generates a tone pattern at the analog output. A perfect DAC system will present a predictable pattern based entirely on the mathematics of the conversion process. One with non-linearities or jitter will present a different pattern, with sidebands and spurii. That's what the Audio Precision test system, used as an analog spectrum analyzer, looks at.

 

That make some sense?

 

These jitter numbers by themselves are kind of meaningless as far as audio reproduction goes, unless they are vanishingly small. Jitter as stated by itself doesn't express the spectral content of the jitter nor does it separate distinct tones from noise. How these emerge at the DAC output can produce really different audible effects. At least this J-Test is a little more qualitative than just a number. Plus, it shows what actually is produced at the analog output of a digital source. In almost all cases the digital signal is converted to analog along the way for amplification and application to your loudspeakers or headphones, so this represents what you'll hear.

 

Link to comment

Here is a link to John Atkinson's Stereophile article on measuring jitter. I would suggest that anyone who is trying to understand the measurments published by Stereophile should read this:

 

http://stereophile.com/features/1208jitter/

 

SO/ROON/HQPe: DSD 512-Sonore opticalModuleDeluxe-Signature Rendu optical with Well Tempered Clock--DIY DSC-2 DAC with SC Pure Clock--DIY Purifi Amplifier-Focus Audio FS888 speakers-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Purple Fuses, Spacetime system clarifiers.  ISOAcoustics Oreas footers.                                                       

                                                                                           SONORE computer audio

Link to comment

Thanks all for your explanations & links (barrows, the last link doesn't work). I've tried to search that Stereophile site before for jitter articles but it shows too all the jitter measurement hits - about 10 pages of them.

 

Apologies, Gordon, if I was impatient with you but I was frustrated by your answers & felt you were avoiding the question with smart answers.

 

Link to comment

I just clicked on the link and it worked fine? Took me right to: "A Case of the Jitters".

 

SO/ROON/HQPe: DSD 512-Sonore opticalModuleDeluxe-Signature Rendu optical with Well Tempered Clock--DIY DSC-2 DAC with SC Pure Clock--DIY Purifi Amplifier-Focus Audio FS888 speakers-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Purple Fuses, Spacetime system clarifiers.  ISOAcoustics Oreas footers.                                                       

                                                                                           SONORE computer audio

Link to comment

Jkeny,

 

I was not trying to frustrate you with what I was saying but to make you realize that if you read all this stuff on 18 different sites that maybe you are falling into the biggest problem today with Computer Audio and that is misinformation.

 

Look I talk to Charlie probably 10 to 15 times a week. I could have pointed this out, but really it's silly considering his health to be running around all over the place putting out internet fires.

 

The big deal here is that you have to realize jitter is only one facet of the puzzle. You could have great jitter and still have a poor sounding unit. Heck you could have async USB and still have poor jitter. Nothing is a silver bullet, everything has to be considered.

 

I suggest you keep an open mind. I am not trying to steer you wrong, neither is Charlie. But there are a lot of people out there who will.

 

Thanks

Gordon

 

Link to comment

I'm well aware Gordon that "correct implementation" of the "correct technology" is the "true way" - that's a nice general statement that we can all agree on. The arguments start when we try to define what those 3 terms in quotations mean. I think you are mis-reading me, I don't blindly absorb information (or mis-information) from various sites I evaluate it as best I can and I do keep an open mind.

 

My background isn't e'ee so I can make stupid mistakes in nomenclature but I think I know the principles & pitfalls in the audio minefield. (BTW, the reason you know I'm on different sites is because I use the same name across all of them so I'm easily recognised - i bet there are a lot like me but using different nicks).

 

I didn't know Charles was unwell & I will be more cognisant of this in the future. Send him my regards even though we have had a few spats in the past.

 

I'm not going to go into it here but what got my goat was the reaction of many posters on DiyHiFi to my thread which wanted to look at the technology (& it's implementation) behind the Musiland DAC - it was a shameful case of technological closed mindedness. Now there is more of a rational view as expressed in John Swenson & your post.

 

As my nick says in other forums "There is a crack in everything, that's how the light gets in"

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...