[opaqueice]

People who download files from Stereophile to hear jitter would need to know that their own system didn't produce any jitter, wouldn't they?  Otherwise they'd be listening to digital files containing jitter, on equipment that was adding jitter of its own while they listened.  Has this been brought up and I missed it?

Presumably stereophile added lots of jitter, more than would already be present on a decent system.

Incidentally, I just realized there's a much easier way to analyze the effects of jitter.  No need to take any Fourier transforms:

if S(t) is the signal you start with, the jittered signal is S_j(t) = S(t + j(t)).  But j(t) is generally very small compared to t, so we can Taylor expand:

S_j(t) = S(t) + j(t) S'(t) + ...  (here S'(t) means the derivative of S(t) with respect to t)

So S_j(t) - S(t) = f j(t) S(t)  if S(t) is close to a pure tone with frequency f and we drop the (very small) higher terms. 

So that's it - the amplitude of the side bands is simply f j(t) (which is the result I had before), and the location of the sidebands is f-f_j and f+f_j (if f_j is where most of the jitter power is concentrated).  If the jitter has a white spectrum, as in your example, the "side bands" get smeared all up and down the spectrum, and you just get white noise.  Very simple, actually.

I still don't see how to match the amplitude in those plots, though.

[bpape]

It was a joke Brian.  I'll agree that we don't necessarily have to know WHY we do or don't hear a difference - but that is kind of relevant to try to reproduce identical tracks with and without it for a group of people to listen to - don't ya think?

Bryan

As long as the group of people are together in one room listening to the same system, sure.  One problem seems to be, unlike garden variety noise, the effects of jitter are epiphenomenal, kinda like harmonic distortion (you're not really hearing anything called jitter, jitter doesn't sound like anything, it is supposed to affect all frequencies all the time, you're hearing its effect, and most likely hearing it more by its absence, etc).  But then again at these low levels, every change becomes epiphenomenal I guess.

The idea (I think....) was to have a pair of tracks that could be sent to a lot of different folks in different places to listen to on different systems.   

Bryan

[art]

I am typing this up off-line, in between all the day-to-day hassles of trying to run a business.

Let's assume that some of you guys d/l'ed the phase noise-jitter s/w.

I will take an 11 MHz clock, with the following phase noise number, which are probably pretty close to what you would find in a typical CD player.

10 Hz = -100 dBc, 100 Hz = -130 dBc,  1 kHz = -135 dBc, the rest = -140 dBc.

So, we get around 1 pSec. Not bad. Easy to believe numbers.

Many years ago, I made some measurements on what happens if you stick the clock into the digital filter chip, and distribute the clock from it. (About the same thing as implementing it internal to that chip.) I am doing this form memory, but I recall the phase noise increased (depending on the actual unit) somewhere in the range of 13- 20 dB.

OK...........plug in some new numbers......off the top of my head:

10 Hz = 90 dBc, 100 Hz = -115 dBc, 1 kHz = -100 dBc, and -120 dBc above that point.

So, we have gone from around 1 pSec to around 5 pSec. What this example can not show is all the 60 Hz (and related harmonic) sidebands, plus some amount of data-correlated noise. Still, it gives you an idea of how much improvement you might typically find by sticking a better clock in your CD player.

Translation: Some. Enough to justify the cost? Depends on your point of view. For me, it costs virtually nothing, so I do. If it costs $300-400 to have someone do it for you, well, you have to decide for yourself.

In any case, jitter should not be a major problem in a CD player. Worry about it only if you don't have other areas to address.

But, when we go to SPDIF, things change a lot. Typical RX chip data sheet says jitter is (whether you believe them or not......well, let's see if we should) around 200 pSec, for the popular ones.

OK, again from memory, but I found that the phase noise was at least 40 dB worse than the clock it was derived from. Maybe more. I have seen 60 dB. But let's stick with around 40 dB degradation.

New numbers to plug in:

100 Hz = -55 dBc, 100 Hz = -85 dBc, 1 kHz = -95 dBc, -100 dBc above that.

That gives us less than 180 pSec.

OK, I pulled those data points out of my head, based on my experience. Can we juggle them some to get down to 100 pSec range? Let's see what that takes.

Hey, only takes 5 dB reduction to get down to 100 pSec.

Of course, this only shows jitter from random noise sources. In real life, the jitter in SPDIF has a strong degree of data correlation. So, it is more audible. You simply can not ignore that fact.

Now, Ethan maintains that anything under 2 nSec is not audible. How much phase noise would you need then? Well, just raise the close-in noise to around -30 dBC. Trust me, that level is really high. Like I said, you can't make a clock with that much jitter unless you really try hard.

So, if Ethan is right, then the clock makes absolutely no difference in any digital audio. Only one of us can be right. So which one is it?

Well, I know that 99% of the folks here believe clock jitter is a problem. So how can Ethan be right?

There is a way! And I will show you.

(This is called comparing apples to apples, and not bananas.)

Let's plug in some new phase noise numbers:

10 Hz = -90 dBc, 100 Hz = -115 dBc, 1 kHz and up is -120 dBc.

Now we set the carrier frequency to 44 kHz. Which is..................drum roll.............the sampling frequency.........and the jitter is................more drum roll........

<2 nSec. Hey, those phase numbers don't like much unlike what I would call a modest oscillator.

So, to a guy like Ethan who lives and works in 44 kHz world, and the guys who build the 11.2896 MHz clocks, we aren't talking the same language. I kinda sorta suspected this all along. Now I have proof.

In a sense, we are both right. I know from my years of designing oscillators, clocks VCOs and PLLs how low this stuff has to be to inaudible. Not all that far from someone who isn't an engineer, and speaks a different language believes from his listening experience.

So, now I guess we can quibble over the last few picoseconds at clock frequency, not sampling frequency.

(Anyone remember what I said about needing to be down to single digit pSec range?)

Pat

[WGH]

The idea (I think....) was to have a pair of tracks that could be sent to a lot of different folks in different places to listen to on different systems.   

Bryan

But on what media? Ideally it should be on reel-to-reel or vinyl, that way you are only listening to the jitter test, not the jitter produced by your equipment too.

As in quantum physics, where observing the state of a quark changes its state, and only partial information can be determined, if the test is on digital media, the very act of observing the test (listening) will change the data, and the results will be different for everyone.

Wayne

[bpape]

You're preaching to the choir.  I'm not agreeing with doing it on CD - just saying that's what was trying to be done.  Part of my 'joke' was identifying things like what you just mentioned and someone else previously did also.  There is no agreement on how to do it, what it impacts, etc.

IF SPDIF is truly the issue, then one could distribute .wav files and play back from PC via a USB D/A converter that doesn't have SPDIF in the chain.  Just a thought.  There are lots of other ways.

Bryan

[WGH]

Part of my 'joke' was identifying things like what you just mentioned and someone else previously did also.

Bryan

Got it. I missed a page trying to keep up.

[Norman Tracy]

The audible effect of jitter in digital audio is a subject which has been around and will come around for as long as we listen via PWM and other digitally encoded means. Before HTML I learned firsthand about flame wars on the ASCII Usenet discussing this subject. I summarized in a circa 1996 article on the ACG website here:

http://www.audiocraftersguild.com/AandE/npt.on.jitter2.htm

Rereading my article 12 years on and all but the specific IC part numbers referenced still stand. The references to the AES papers by Hawksford, Harris, and Dunn remain the bridge for many of us from the information theory of the PhD to the application in the real world by engineers.
   
Part of the challenge here is the width and depth of the subject. Asking what is the “sonic signature of jitter and how to conquer it?” is exactly like asking “what is the sonic signature of distortion and how do I get it out of my system?”  That discussion could go on and on branching off into threads on distortion in speakers, line level amps, power amps then onto sub threads on IMD, THD, TIMD, SID, weighting of harmonic orders, and on and on. This thread has already been there done that currently standing at 226 posts across 23 pages! Likewise when getting into jitter we quickly need to zoom down from the 50,000 foot view into the details. For example where and by what means are we measuring the jitter? The Stereophile measurements mentioned earlier in this thread come from an instrument set which deduces the jitter levels based on playing a engineered test signal known to provoke signal correlated jitter in digital audio systems then digitizing and analyzing the device under test (DUT) analog output via FFT. This test rig yields numbers in the circa 150 pS range for state of the art devices. For systems designers and hi-fi magazine evaluators this is a great test because it encompasses the entire device from data input to analog outputs. The danger for confusion to the layperson is that with little research one can learn that for $20 an oscillator with a total jitter of less than 30 pS (example from the Fox FXO-HC53 series) can be had. Here the anti-high-end audio pundits will chime in howling at what a big rip off Brand X is because it costs $VW,XYZ and look at that oscillator which is better for only $AB! Oh the scandal, those audiophiles are such a bunch of delusional lemmings. Wrong again bozo, you are not comparing apples to apples or even apples to oranges but rather apples to puppy dogs. In the one case an entire system is the DUT and the instrument set yields numbers only comparable to itself. The case of the oscillator module has the manufacturer measuring the jitter directly at the chip’s output pin under lab bench optimized ideal steady state circumstances using industry standard methods.

Jumping to another node in the DAC subsystem (regardless of if the subsystem is part of a player with a disk drive or stand alone DAC) a strong variable is the architecture of the specific DAC chip used. As digital over sampling filters have been incorporated into the DAC chips and the IC designers have moved through RsquaredR to single bit to hybrid multi-bit sigma-delta architectures the actual DAC IC has jitter attenuation properties. The point I am driving toward is that the first order math referred to in opaqueice’s posts and the figures posted from Pohlemann’s book by Ethan Winer are modified in implementation specific ways correlated to the exact chip set the designer selected and how well cared for it is by the power supplies and PCB layout. Early DAC chips like the Burr-Brown PCM-58 would behave like the Mathcad models I am guessing were used to generate the Pohlmann figures posted (note ‘Simulations showing the spectrum..”’ disclaimer in the figure title). Soon the IC designers had better than 16 bit resolution licking the output linearity issue (at least until 24 bit came along) and turned their attention to how bit and word clock jitter were affecting their devices. With Moore’s law providing plenty of gates on ever smaller dies the rather clever IC designers incorporated conversion schemes yielding “low sensitivity to jitter”. Opaqueice alluded to this with his disclaimer in post #209 “the analysis above assumes the DAC does nothing to attenuate jitter”.
 
Alas this leaves the bewildered audiophile still sitting in front of his system wondering “what is the sonic signature of jitter and how to conquer it”? As touched on by the 200+ and growing posts to this thread the conquer it side of that question is an ongoing battle involving both technical complexities and marketplace realities. The sonic signature side of the question is in my experience that jitter makes the music fuzzy. Polishing off the fuzz is like any exercise in buffing in that how far to go will be dictated by taste, personal aesthetic criteria, and the resources available.

[AphileEarlyAdopter]

Folks,
This is going nowhere. So lets forget about jitter and lets share our personal experiences on one aspect of our system.
Do you have a digital connection in your system ?
If yes, have you tried, different digital cables and was there any difference to the sound of your system?

Here's my input on the subject.
Yes, different digital cables produce audible sound differences in my system.

So lets go from the phenomenon to understanding and analysing the mechanism. If we all don't recognise the phenomenon, it is waste to go into analysing the cause, the parameters and steps to recreate it.

[miklorsmith]

Actually, I don't have a definite opinion of the effects of jitter.  What I do have is that there is a significant disparity in the quality of output of digital devices.  I'm not enough of an engineer to track where those benefits reside but I hear them.

[jhm731]

Folks,
This is going nowhere. So lets forget about jitter and lets share our personal experiences on one aspect of our system.
Do you have a digital connection in your system ?
If yes, have you tried, different digital cables and was there any difference to the sound of your system?

Here's my input on the subject.
Yes, different digital cables produce audible sound differences in my system.

So lets go from the phenomenon to understanding and analysing the mechanism. If we all don't recognise the phenomenon, it is waste to go into analysing the cause, the parameters and steps to recreate it.

I agree.

Here's a simple test of the effects of jitter. Try comparing the coax and optical digital output of an SB3.
Here are jitter measurments from the SD forum:

"I just purchased a brand new SB3, and a colleague subjected the digital outputs to a quick jitter measurement using Audio Precision equipment. Results: SB3 digital coax out jitter: ~99ps, SB3 digital optical out jitter: ~892ps"

Your cable differences are probably a result of poor impedance matching, which Pat has covered in Lab circle.

[opaqueice]

Here's a simple test of the effects of jitter. Try comparing the coax and optical digital output of an SB3.
Here are jitter measurments from the SD forum:

"I just purchased a brand new SB3, and a colleague subjected the digital outputs to a quick jitter measurement using Audio Precision equipment. Results: SB3 digital coax out jitter: ~99ps, SB3 digital optical out jitter: ~892ps"

I've tried that experiment going into my Benchmark DAC1.  There is no difference.

Although that's hardly surprising, since (at least if you believe Benchmark's published plots) the DAC1 is totally immune to jitter.

[marklivia]

In regards to the digital cable making a difference, i disagree totally that it's just impedence mismatching. How can one cable improve every system it's been used in, as is the case w/ mine? This very pricey item utterly transformed the output of my transport and dac. I see it as getting rid of jitter. What else could a cable do that could make digital sound really real? (I hate to use the phrase"analog")

[jhm731]

In regards to the digital cable making a difference, i disagree totally that it's just impedence mismatching. How can one cable improve every system it's been used in, as is the case w/ mine? This very pricey item utterly transformed the output of my transport and dac. I see it as getting rid of jitter. What else could a cable do that could make digital sound really real? (I hate to use the phrase"analog")

Audiophiles love plug and play upgrades.

Until you determine the impedence match of your "very pricey item" that "utterly transformed the output of my transport and dac," verses what you were using before, you will never know what made your "digital sound really real."

BTW, Hiend cable manufacturers love guys like you.

[BrianM]

I've only compared two digital coaxial cables into my DAC, and I thought I heard a difference, though I'm unable to exactly describe what it was (other than "better"), and certainly don't know if it had anything to do with jitter.  I switched from a Belden 1694a (2 ft length) to a VenHaus (1.5 meter length).  Both have the same impedance; the longer length is supposed to mitigate reflections caused by the RCA jack.  Since I heard an improvement...maybe it does!

[Geardaddy]

The audible effect of jitter in digital audio is a subject which has been around and will come around for as long as we listen via PWM and other digitally encoded means......

Part of the challenge here is the width and depth of the subject. Asking what is the “sonic signature of jitter and how to conquer it?” is exactly like asking “what is the sonic signature of distortion and how do I get it out of my system?”
 
Alas this leaves the bewildered audiophile still sitting in front of his system wondering “what is the sonic signature of jitter and how to conquer it”? ....

The sonic signature side of the question is in my experience that jitter makes the music fuzzy. Polishing off the fuzz is like any exercise in buffing in that how far to go will be dictated by taste, personal aesthetic criteria, and the resources available.

Bingo.  A layman's definition of jitter is simply digital distortion, and that distortion (the part that is audible...) manifests as "fuzz" or possibly Harley's more elaborate word picture:

"loss of space & depth; softening of the bass; hardening of timbre; a glassy sound on initial transients (most noticeably on the leading edge of upper-register piano attacks); a metallic sheen overlaying the treble; and an overall flattening of the soundstage & homogenization of instrumental images with the stage." "" Dunno

I am not an engineer but simply a pragmatic end user interested in taming jitter (I am referring to the epiphenomenon here....agreed).  Again, in my layman's mind, that involves an accurate reconstitution of digitized bits and removing all sources of error in that reconstitution.  Part of what motivated me to start this thread is industry-related verbage pertaining to jitter.  Proponents of computer centered, CD-less  audio are claiming superiority in part due to their ability to deal with jitter, the absence of moving parts, etc... (see http://www.positive-feedback.com/Issue30/cjdiaries.htm).

So, as was suggested earlier, what anti-jitter technologies or tweaks have you all used that appeared to transform your digital front end with or without measurements.....     

[Ethan Winer]

Folks,

Well, there's not much more to add now, and I do agree this stuff all falls under tweaks. I still plan to make a set of Waves files for an "audibility test," because I believe this is the core issue. I know it angers - or at least irritates - some people to be told they can't possibly be hearing a change when they swap digital cables etc. I'd like to ask for a show of hands among those who claim to hear a change by swapping digital cables if they have absorption at the reflection points in their room. But I don't want to be accused of spinning this into a plug for acoustic treatment so I won't ask.

When I get a chance I'll create a series of files as described earlier, with obnoxious noise under various types of music, and I'll start a new thread for comment.

Thanks guys, I feel this discussion has been highly illuminating. And thanks also for keeping things civil.

--Ethan

[Double Ugly]

I know it angers - or at least irritates - some people to be told they can't possibly be hearing a change when they swap digital cables etc.

As it should.

[HiFiSoundGuy]

  If you REALLY want to HEAR a DIFFERENCE you only have to try these products from http://www.machinadynamica.com  Teleportation tweak! CodeName Turquoise! The ALL NEW Intelligent Chip! These products are NOT SNAKE OIL!

[BrianM]

I know it angers - or at least irritates - some people to be told they can't possibly be hearing a change when they swap digital cables etc.

As it should.

Seriously?  Maybe it's a little irritating at the most, but I don't think people should be taking this dispute that seriously.  If one agrees these are tiny tweaks, one should probably have the humility to receive reminders about psychoacoustics with relative equanimity.  If I hear a change swapping digital cables and am told I couldn't possibly have for what approaches a scientific reason, my reaction is to remember to maybe try and obsess a little less about what I'm sure I'm hearing.  From tiny tweaks.  Altogether a healthy thing, probably.

[Geardaddy]

Ethan, I am still interested in your audibility tests to smoke out any psychoacoustic elements.  It would also be fun to apply that same principle to digital formats at some point.  Anyway, I appreciate everyone's input.  This thread has been great....