Looking for a treatise on jitter relationship to stereo imaging

[davide256]

As above. I'm feeling lazy about doing the math. Anyone seen an article with decent science behind it?

 

Possible calculation model example

 

two different instruments on stage physically separated 50 feet.

 

Assume base frequency of 2khz for each, with harmonics of 4khz, 8khz and 16khz but with balance differences in spectrum making

both unique for sound signature.

 

If well miked and played back through a "perfect" stereo system, does jitter destroy imaging? I suspect that jitter affects sound signature

( Guarneri, Stradivarius, or country fiddle?) but that image is more affected by wow/flutter, pattern vs random jitter.

[esldude]

Just listen and jitter will indeed completely destroy the image. Put in a jitter reducer and the image comes back. Clear as day.

 

Math, science and any evidence this really happens?????? Take some good advice and don't waste your holiday weekend looking for or thinking about it. It is a fruitless search.

[sandyk]

Just listen and jitter will indeed completely destroy the image. Put in a jitter reducer and the image comes back. Clear as day.....

.

 

Dennis

And the device you used to verify this statement was ???

Perhaps it was meant to be a smart ass reply ?

Alex

[4est]

Smart ass reply? Ya think?!?

Dennis

And the device you used to verify this statement was ???

Perhaps it was meant to be a smart ass reply ?

Alex

[esldude]

https://davidkessner.wordpress.com/2011/10/27/a-pragmatic-approach-to-jitter-in-digital-audio/

 

Rather back of the envelope or napkin calculation in the above little blog.

 

If you were to move the speaker or your head back and forth, 2.5 times the width of a gold atom that would be 1 picosecond of jitter. So the apparent image shift would be sub-molecular. 135 picometers I think he said (I haven't checked his calculations).

 

Brownian motion of the air is several orders of magnitude larger than this. It appears our ear drums are evolved to work very near, but just above the point where they would 'hear' brownian motion. It makes sense hearing such motion would be of no survival use. As the smaller effects of jitter are well below this point I think it safe to say 1 picosecond of jitter were it coupled through to speaker cones and the air would have no chance of being audible.

 

Such tests as have been done indicate jitter isn't audible until some 4 or 5 orders of magnitude more than this with music. And not much less with a high frequency test tone. That seems to fit roughly with the physics of things without thoroughly calculating it out. (left as an exercise for the reader if you will).

 

I take it this is something along the lines of what the OP wanted to see.

[sandyk]

https://www.google.com.au/?gws_rd=ssl#q=Jitter+causes+wideband+noise

[whoozwah]

I don't know that I've ever heard jitter or its effects on music. Or if I have I wasn't aware of it or it was so minute as to be not worth the trouble to fix. Feels like decorating behind the christmas tree. Or dusting off the unused top shelf of a closet in a guest bedroom.

[davide256]

https://davidkessner.wordpress.com/2011/10/27/a-pragmatic-approach-to-jitter-in-digital-audio/

 

Rather back of the envelope or napkin calculation in the above little blog.

 

If you were to move the speaker or your head back and forth, 2.5 times the width of a gold atom that would be 1 picosecond of jitter. So the apparent image shift would be sub-molecular. 135 picometers I think he said (I haven't checked his calculations).

 

Brownian motion of the air is several orders of magnitude larger than this. It appears our ear drums are evolved to work very near, but just above the point where they would 'hear' brownian motion. It makes sense hearing such motion would be of no survival use. As the smaller effects of jitter are well below this point I think it safe to say 1 picosecond of jitter were it coupled through to speaker cones and the air would have no chance of being audible.

 

Such tests as have been done indicate jitter isn't audible until some 4 or 5 orders of magnitude more than this with music. And not much less with a high frequency test tone. That seems to fit roughly with the physics of things without thoroughly calculating it out. (left as an exercise for the reader if you will).

 

I take it this is something along the lines of what the OP wanted to see.

 

Survival is evolution's shaper of our senses. See the threat, hear the threat, identify the threat, know where the threat is. So it makes perfect sense to me that if we can't see the threat but we can hear the threat, evolution would fine tune our hearing to identify where the threat is based on sound alone.

 

Appreciate the article, they were looking at this only in a monophonic sense so will keep looking for a differential analysis of left/right channel jitter impact on identifying spatial location.

[davide256]

Found this on the Apogee site, no math related to imaging but some assertions on the relationship between jitter and imaging

 

What is jitter? | Apogee KnowledgeBase

[esldude]

Survival is evolution's shaper of our senses. See the threat, hear the threat, identify the threat, know where the threat is. So it makes perfect sense to me that if we can't see the threat but we can hear the threat, evolution would fine tune our hearing to identify where the threat is based on sound alone.

 

Appreciate the article, they were looking at this only in a monophonic sense so will keep looking for a differential analysis of left/right channel jitter impact on identifying spatial location.

 

So how can you hear it in stereo if the effect in mono is well below brownian air motion? If both ears can't hear low levels of jitter it can't effect imaging. If want to look in stereo then imagine it moves both speakers the picometer and back. Same difference, the effect is well below even being trivial.

[sandyk]

Certain types of Jitter manifest themselves as wide band low level noise. The "veil" that many talk about removing.

[esldude]

Certain types of Jitter manifest themselves as wide band low level noise. The "veil" that many talk about removing.

Yes this is true. But what level is the noise? The idea of removing veils from reduced jitter has been sold long and hard. That doesn't makes it so.

[Fitzcaraldo215]

Yes this is true. But what level is the noise? The idea of removing veils from reduced jitter has been sold long and hard. That doesn't makes it so.

 

I honestly do not know why this jitter thing keeps cropping up in computer audio when most DACs are asynchronous USB. Does anyone still believe there is a lot of jitter that way? I do not, but I am willing to be shown.

 

Of course, USB DACs have usually very low residual jitter internally irregardless of the interconnect method. Some may be slightly better than others in that regard, but I think we are unable to isolate the jitter itself as a major cause of any sonic differences. There are too many other differences between DACs to say for sure.

 

Other interconnect methods, like spdif and especially HDMI, often have much more measured jitter than asynch USB. They usually suffer from the jitter-causing two master clock and recovery of the player clock from the data stream issues. That is because they use one way communication, and do not have the ability for the DAC to signal the player as in asynch USB. Depending on the implementation, particularly with HDMI, the contribution of jitter may vary with the degree to which other sophisticated jitter suppression was designed in.

 

Older, proprietary asynchronous schemes were used via HDMI on certain, specific now defunct Sony and Pioneer player-AVR combos. They were shown to have much lower measured jitter in that asynch mode in HiFi News and Record Review. Anecdotally, most listeners thought they sounded better that way.

[esldude]

Jitter as it appears to computer audiophiles:

[Jud]

Jitter as it appears to Dennis:

 

 

 

 

 

 

 

[Richard Dale]

Yes this is true. But what level is the noise? The idea of removing veils from reduced jitter has been sold long and hard. That doesn't makes it so.

 

What is this 'so' thing that you speak of? Are you the guardian of 'so-ness'?

[sandyk]

Yes this is true. But what level is the noise? The idea of removing veils from reduced jitter has been sold long and hard. That doesn't makes it so.

 

Dennis

I really feel sorry for people like yourself who are unable to hear these things, such as what happens when you power a Mac Mini with a Linear PSU such as John Swenson's JS2, and then the further improvement when followed by a USB Regen powered by something better than the supplied Meanwell 7.5V SMPS, or better still according to numerous reports in this forum, the Intona USB isolator.You will undoubtedly even make excuses for why you haven't tried too hard to obtain a loan of one based on some measurements posted elsewhere using only the supplied SMPS, or your own ingrained beliefs, and refuse to try these things for yourself, despite a very good warrantee from Uptone Audio, or the fact that many people would happily buy yours if you didn't like it, at probably what it cost you, to avoid having to possibly wait on a queue to purchase one.

Even a hard nosed E.E. like John Swenson could hear an improvement when he put his Fan Filter in line to filter out the sharp rise and fall time PWM fan speed pulses that were radiated by the fan itself as RF/EMI.

 

Sorry Dennis, but measurements don't always give the full story , even with amplifiers or preamplifiers with virtually identical specifications that SEEM to indicate that they should be completely transparent according to present test methodology.

Some preamplifiers and amplifiers excel in the areas of soundstage and the perception of image height, despite having virtually identical near perfect measurements.

Alex

[jabbr]

As above. I'm feeling lazy about doing the math. Anyone seen an article with decent science behind it?

 

This is a good start: https://www.by-rutgers.nl/PDFiles/Audio%20Jitter.pdf

and:

https://www.by-rutgers.nl/rutgerS-Clock.html

[esldude]

I wonder if the OP listened to the files I used in the 8th generation poll. If you haven't read that thread don't. It contains the answers. I could give you links to the files. Two are the original digital file, and one was played and recorded for 8 generations. You get added jitter of the DAC, and jitter of the ADC each time through. 8 times twice the jitter. If jitter messes with the image this should do it. Picking the odd sounding file of the three should be a piece of cake.

 

One of the recordings I used was a minimalist recorded Wilson audio track of a piano. No processing done to the file by Wilson or me. Should be just the ticket to hear the image widening, depth destroying digital devil demon jitter.

 

BTW, both DAC and ADC in that experiment are rated at below 250 picoseconds jitter. Low enough, but hardly super low state of the art.

[jabbr]

I wonder if the OP listened to the files I used in the 8th generation poll. If you haven't read that thread don't. It contains the answers. I could give you links to the files. Two are the original digital file, and one was played and recorded for 8 generations. You get added jitter of the DAC, and jitter of the ADC each time through. 8 times twice the jitter. If jitter messes with the image this should do it. Picking the odd sounding file of the three should be a piece of cake.

 

One of the recordings I used was a minimalist recorded Wilson audio track of a piano. No processing done to the file by Wilson or me. Should be just the ticket to hear the image widening, depth destroying digital devil demon jitter.

 

BTW, both DAC and ADC in that experiment are rated at below 250 picoseconds jitter. Low enough, but hardly super low state of the art.

 

There is jitter and there is jitter. As Rutgers notes, the specific measurement that seems to affect audio the most is the close-in phase noise.

 

From Google: http://www-tcad.stanford.edu/tcad/pubs/device/SPIE04_navid.pdf

[esldude]

There is jitter and there is jitter. As Rutgers notes, the specific measurement that seems to affect audio the most is the close-in phase noise.

 

From Google: http://www-tcad.stanford.edu/tcad/pubs/device/SPIE04_navid.pdf

 

Curious that you say "it seems to affect audio the most", and then link to a pdf which does not contain the word audio, or sound or sonics. The paper seems to be looking at what noise causes the most effect with higher frequency oscillators which might apply to some of these clocks we derive timing from while yet not saying anything about its effect upon audio results. The noise in the clock inaccuracies isn't linked to any direct audio effect in this pdf.

[jabbr]

Start with the Rutgers articles, the last link has additional math and explains "close-in phase error". You wanted information on the relationship between jitter and noise? I've seen a number of different jitter measurements so it's good to be more precise and know what the best thing is to optimize. Rutgers publishes the circuit he used to make his measurements so these really are good resources.

 

Also realize that really low noise and accurate clock oscillators were not developed for audiophiles rather much better funded applications... High speed networks, RF communications, phased array radar etc

[Superdad]

I honestly do not know why this jitter thing keeps cropping up in computer audio when most DACs are asynchronous USB. Does anyone still believe there is a lot of jitter that way? I do not, but I am willing to be shown.

 

Well I can't show you, but I can tell you something that makes it evident.

 

In working on a greatly enhanced REGEN (which will have true galvanic isolation in addition to about 6 other improvements), we compared the already decently low-jitter Crystek C-3391 (standard on the USB REGEN for the past year), with the 9 times the price Crystek CCHD-575--an ultra-low-phase-noise clock worthy of being the master audio clock in a SotA DAC. And even as a USB clock, the difference was plainly heard in 20 seconds. I don't take kindly to large increases in build cost if it does not result in worthwhile, audible improvement, but it took only a heartbeat to decide that this mattered.

 

 

[bTW, Fitzcaraldo215, I am well aware of your objectivist skepticism and calls for measurements over at WBF. Such is not near as easy at it might seem. Measuring close-in phase noise is not a simple matter--and a typical AP system won't do either. John (Swenson) is developing a unique and modest-cost phase-noise test probe system based on the software for the John Miles TimePod/Symmetricom/Microsemi 3120A (a rare and $20K item these days)--see

 

But the fact is, despite ridicule from the meter-bound crowd, it is often faster and more efficient to try something based on a thought, and to listen to it to decide if it is worthwhile. Measurements are helpful, but they only go so far in telling you what the sonic result will be. Ask ANY engineer at ANY of today's well-known high-end audio firms and they will tell you they do both in refining a design.

 

Do tell Amir I said hello. I remain in self-imposed exile from that other place (life is too short for the hostility), and recently reaffirmed my decision when I read--in the MicroRendu-bashing thread--Amir claim that he "spent extensive amount of time" with John Swenson, and called him a "hobbyist." He met John and I for 15 minutes outside the elevators at RMAF, and he mostly went on about a meaningless measurement technique. Such a guy. ]

 

--Alex C.

[jabbr]

Look at the Rutgers measurement circuit -- I'm seriously considering building it for my home lab and to measure a circuit I am toying with -- the question is as to what the useful limit if lowering close in phase noise is-- the application comes up because as the clock rate goes up so does the close in phase noise. So moving from DSD256 to DSD512 to DSD1024 at some point as the close in phase error goes up, increasing the rate will have negative returns...

[jabbr]

 

In working on a greatly enhanced REGEN (which will have true galvanic isolation in addition to about 6 other improvements), we compared the already decently low-jitter Crystek C-3391 (standard on the USB REGEN for the past year), with the 9 times the price Crystek CCHD-575--an ultra-low-phase-noise clock worthy of being the master audio clock in a SotA DAC.

 

The Crystek CCHD - 575 is a very nice part for the price ~$20 - wouldn't describe as SOTA for A DAC-- expect to see it in decent boards.

 

For comparison, the dual clock oscillator which Andrea mori is offering as a DIYaudio group buy is targeted at $60. That's with two SC cut crystals from Laptech and a Driscoll oscillator circuit.

 

Now I understand you need to run a business, but we aren't looking to save just as they now say, a Tubman:)

 

 

And even as a USB clock, the difference was plainly heard in 20 seconds. I don't take kindly to large increases in build cost if it does not result in worthwhile, audible improvement, but it took only a heartbeat to decide that this mattered.

 

 

Ok so actually for a USB application in which we really don't know the parameters by which SQ is affected by clocks, I think the CCHD-575 is perfect -- for example, as opposed to DACs, we don't have information to say that close in phase noise, is, for example, the critical parameter.