jitter

[th00ht]

I've hear various accounts on DAC jitter. As by design spdif does not include a clock signal (the clock is derived from the data itself) it seems that jitter could occur when an external DAC is used on either transporter or SB. I've read different accounts that this jitter might result in audible artifacts.
Has any of you experienced negative results of using external DACs caused by jitter?

[cliveb]

Originally Posted by th00ht

I've hear various accounts on DAC jitter. As by design spdif does not include a clock signal (the clock is derived from the data itself) it seems that jitter could occur when an external DAC is used on either transporter or SB. I've read different accounts that this jitter might result in audible artifacts.
Has any of you experienced negative results of using external DACs caused by jitter?

Nice one. Posting this in the Audiophiles forum is the equivalent of shouting "FIRE!" in a crowded theatre. I just hope you don't get trampled by the resulting stampede :-)

[DCtoDaylight]

Originally Posted by th00ht

by design spdif does not include a clock signal (the clock is derived from the data itself)

This isn't true, SPDIF does include a clock signal. The problem stems from the fact that the data and clock are combined, they aren't sent separately. This allows the data to have an influence on the clock recovery circuit, and potentially induce jitter.

I've haven't experienced any issues that I can say were caused by jitter, but then the DAC I own has a very good clock recovery circuit.

[Patrick Dixon]

Any clock recovery circuit, is likely to introduce 'jitter' to the recovered clock, because that's the nature of synchronising one clock to another. The recovered clock has to 'follow' the S/PDIF signal, and so it will have to periodically adjust itself to keep synchronised.

You can hear the negative effects of jitter, but they can be down to a better or worse clock source in the 'transport', as well as the S/PDIF signal path and clock recovery circuit. You'll probably only know that you're hearing the effects of jitter though, once you hear something with less.

[ar-t]

OK, before the loonies come out of the woodwork......

Yes, the clock is sent with the data, but it has to be extracted. Extracting it means that you need a phase-lock loop to do it. All PLLs will exhibit jitter. How much depends on a lot of factors.

In the case of SPDIF, because it was not intended for anything other than testing, the clock contains a lot of data-correlated artifacts. This type of jitter is worse, in terms of sonics, than Gaussian noise.

Jitter has 2 parameters that are important: amount (usually expressed in pSec, nSec, whatever, or unit intervals) and its spectral content. You see jitter numbers thrown around in the same manner as you will see THD numbers. Without the harmonic distribution of the distortion, that number is not all that helpful. (Some harmonics are especially nasty, while others can be pleasant.)

So, not only do jitter numbers do little, other than provide yet another misunderstood venue for forum food fight fodder, there is not much discussion on what it sounds like.

The best way that I can explain how jitter sounds, well.....as someone pointed out, you know when you hear something with less. You do not actually hear the jitter itself. What you do hear it mucks things up in a subtle manner. The 2 things that you will notice when the jitter is reduced is that the bass is tighter and more defined, and that the top end is cleaner and smoother.

Some DACs have additional circuitry designed to minimise these effects. Sometimes they work well, other times they work better on paper than in practice. An example of such is the blind trust some designers place in ASRCs. They will all "remove" the jitter, but do so by moving it around. (Please don't ask me to explain how they work.) The less successful ones perform poorly because the chip designers took too many shortcuts in the implementation of the mathematical functions inside the chip that do the hard work. They end up spreading the jitter out all of the place. I can think of one very widely used ASRC that is used in a lot of cost-effective "one box, lots of functions for not a lot of money!" units.

Pat

[pfarrell]

ar-t wrote:
> All PLLs will exhibit jitter. How much depends on a lot of factors.

The real question is "does it matter at all?"

Or perhaps "do any audiophile level DACs exhibit enough to care?"


> In the case of SPDIF, because it was not intended for anything other
> than testing, the clock contains a lot of data-correlated artifacts.
> This type of jitter is worse, in terms of sonics, than Gaussian noise.

I can see data correlated errors being worse that random, but that
doesn't address the real quesion: "does jitter matter?"

--
Pat Farrell
http://www.pfarrell.com/

[ar-t]

Listen to a system with lots of jitter, and one with much lower jitter. If you can't hear the difference, then I guess is doesn't matter to you.

Some folks are happy listening to MP3s on iPods. Does that matter? Not to them, but maybe to others.........

Pat

[pfarrell]

ar-t wrote:
> Listen to a system with lots of jitter, and one with much lower jitter.
> If you can't hear the difference, then I guess is doesn't matter to
> you.

This is silly. Where am I supposed to find said systems with "lots of"
or "lower" jitter?

If you are on this list, you most likely have a SB3 or Duet or
Transporter. Any of them particularly terrible? I would expect that TP
to be fairly good, but what metric is used? For years, I ran the output
of a SB through a Benchmark DAC-1. The Benchmark folks claim that their
DAC is immune to jitter. I suspect that their claim is marketing, but
how can one tell?

I've read that the "jitter" measurement that JA uses at Stereophile is
meaningless, yet he devotes column inches to it each time he gets a
transport or DAC.

What metric of jitter is important?
What devices that we are likely to have are examples, good and bad, of
this measured thing?


"When you can measure what you are speaking about, and express it in
numbers, you know something about it; but when you cannot express it in
numbers, your knowledge is of a meager and unsatisfactory kind. It may
be the beginning of knowledge, but you have scarcely, in your thoughts,
advanced to the stage of science."
Popular Lectures and Addresses, Lord (William Thompson) Kelvin.

I tried for five years to get a PhD in Software Engineering based on
measurements of software, preferably measurements of quality or even
quantity. I failed. There are no measurements of software that mean
anything. Nothing like power, frequency, pressure, etc. that real
engineers use to design products.

I hear people talk about jitter, and use "less jitter" to justify huge
expenditures of hard earned cash. But I've seen zero science or
engineering to justify what is good and what is bad.

Without some science, I don't believe in jitter, and I don't believe it
matters.

--
Pat Farrell
http://www.pfarrell.com/

[ar-t]

Originally Posted by pfarrell

ar-t wrote:
> Listen to a system with lots of jitter, and one with much lower jitter.
> If you can't hear the difference, then I guess is doesn't matter to
> you.

This is silly. Where am I supposed to find said systems with "lots of"
or "lower" jitter?

Listen to a decent CD player, that has a SPDIF output. Listen to its analogue output, and then a typical outboard DAC. That is how.

If you are on this list, you most likely have a SB3 or Duet or
Transporter. Any of them particularly terrible? I would expect that TP
to be fairly good, but what metric is used? For years, I ran the output
of a SB through a Benchmark DAC-1. The Benchmark folks claim that their
DAC is immune to jitter. I suspect that their claim is marketing, but
how can one tell?

Immune...........well, that is their explanation. They use circuits that get rid of most of the jitter inherent in the SPDIF chain.

I've read that the "jitter" measurement that JA uses at Stereophile is
meaningless, yet he devotes column inches to it each time he gets a
transport or DAC.

You expect me to defend him? Dream on, bub.

What metric of jitter is important?
What devices that we are likely to have are examples, good and bad, of
this measured thing?

Like THD, it depends on amount and spectral distribution. It can not be answered in 25 words or less. Especially to skeptics.

"When you can measure what you are speaking about, and express it in
numbers, you know something about it; but when you cannot express it in
numbers, your knowledge is of a meager and unsatisfactory kind. It may
be the beginning of knowledge, but you have scarcely, in your thoughts,
advanced to the stage of science."
Popular Lectures and Addresses, Lord (William Thompson) Kelvin.

I tried for five years to get a PhD in Software Engineering based on
measurements of software, preferably measurements of quality or even
quantity. I failed. There are no measurements of software that mean
anything. Nothing like power, frequency, pressure, etc. that real
engineers use to design products.

I hear people talk about jitter, and use "less jitter" to justify huge
expenditures of hard earned cash. But I've seen zero science or
engineering to justify what is good and what is bad.

I can measure it. I heard it before I measured it, which is why we set about to measure it.

Without some science, I don't believe in jitter, and I don't believe it
matters.

So, you don't believe in science and already have your mind made up. Pointless to continue.

Pat

[pfarrell]

ar-t wrote:
> Listen to a decent CD player, that has a SPDIF output. Listen to its
> analogue output, and then a typical outboard DAC. That is how.

Why would I have a "decent CD player"? I got rid of all audio CD players
within a year of getting my first SqueezeBox.

And a demonstration does not example proof. A counter example can
dispute a theory, but one demonstration means nothing.

> Like THD, it depends on amount and spectral distribution. It can not be
> answered in 25 words or less. Especially to skeptics.

So use more than 25 words.

This topic comes up periodically, and the only answer I see is "trust me"

I don't trust anyone on this topic. I want science or engineering.
Preferably both.


> I can measure it. I heard it before I measured it, which is why we set
> about to measure it.

What do you measure? What values are important? What confidence do you
have for the limits of where it matters? Are the absolute differences
important? or only relative differences (the way dB are defined)?

> So, you don't believe in science and already have your mind made up.
> Pointless to continue.

No, I believe in science and measurement. I was a PhD student for five
years. That is all about science. And I studied engineering for four
years as an undergraduate. I don't believe in voodoo. I don't believe
that software engineering is engineering, but Computer Science is science.

Audio technology is normally designed. If its a real problem, then
engineers need measurements to know how to design against it. Real
engineers do things like say "I can have this amount of THD in the
circuit for $20, or one half that for $30, or one tenth that for $300"
so that cost effective tradeoffs can be made.

What I don't see is any science between the claims that jitter is
important or specific values that are good, bad or indifferent.

Its not that my mind is made up, but I've not read a single credible
source that says "here is the science on the topic of jitter"

That S/PDIF is less than optimal is not very interesting. It was
designed by Sony and Phillips as a cheap, mass market connection.

In practical terms, S/PDIF is the same as AES/EBU, just over different
media. If jitter is so terrible in S/PDIF, then it would be just as
terrible in AES/EBU. If its terrible in professional circles, why hasn't
it been designed out of the standard in the past 20+ years.

So far, I've seen zero evidence that anyone's understanding of jitter is
more than "a meager and unsatisfactory kind" and its a long way from
"advanced to the stage of science."

Lord Kelvin's positions are part of why we as a world have named our
temperature scale with his last name.

Show me the science.

--
Pat Farrell
http://www.pfarrell.com/