Jitter and CD-R

[Guest]

Archived from groups: rec.audio.high-end (More info?)

Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
news:<Kc5Gc.24114$MB3.7322@attbi_s04>...
> On Sun, 04 Jul 2004 07:39:58 GMT, Stereophile_Editor@compuserve.com
> (John Atkinson) wrote:
>> if you doubt the articles in Stereophile, I suggest you read Barry
>> Blesser's compendium on digital audio in the October 1978 issue of
>> the Journal of the Audio Engineering Society, where the problems of
>> word-clock jitter were first described in an audio context (to the
>> best of my knowledge). A Stereophile article showing how different
>> analog signals can be reconstructed from identical bitstreams can
>> be found at http/www.stereophile.com/reference/1290jitter. This,
>> too, is based on an AES paper, this time by Stephen Harris, and a
>> third article can be found at
>> http/www.stereophile.com/reference/590jitter. A fourth article,
>> showing measured differences, goes up in the stereophile.com
>> archives on Monday.
>
> Hardly new knowledge. Jitter has been acknowledged as a problem in
> digital audio since the '60s - telecomms being several decades
> ahead of so-called 'high end' audio, as usual.

With respect Stewart, I think you are being a tad disingenuous here.
Yes, the problems of jitter were known in telecommunications work.
But it was only with the advent of practical digital audio in the
1970s, with the pioneering work of the BBC in the UK, Sony, NHK and
Denon in Japan, and 3M and the late Dr. Thomas Stockham at Soundstream
in the US, that digital theory became applicable to audio. Far from
the Blesser paper that I mentioned being "decades" behind, it was
published in 1978, just a handful of years after the first digital
audio devices were prototyped. Bob Stuart's work on jitter reduction
in Meridian players followed the Blesser paper by a decade, but that
was still just 5 years after the commercial launch of the CD medium.

>> You may think it unlikely, Mr. Seufert, but that doesn't mean it
>> doesn't exist. The problem is that there is no clock in the disc
>> data.
>
> Exactly! In a standalone player, the only clock is the free-running
> DAC clock, which also controls the data-reading servo. Hence, *if*
> that clock has vanishingly low phase noise, and *if* the power
> supplies are absolutely clean, there can be no jitter in the output
> signal, other than that of the A/D converter used to make the
> original digital master.

I really don't think we disagree, Stewart. I would say that my position
is that your word "if" conceals a multitude of design sins on the part
of product designers, sins that, according to the measurements
published in Stereophile and other review magazines, allow time-base
errors to propagate through to the recovered analog signal. I agree
with you that such design sins are poor engineering; I would merely
point out to you and Mr. Mr. Seufert that because something in theory
can be made to be perfect, that doesn't mean all real-world solutions
are also perfect.

<snip>

>> Around 12 years ago, this subject was examined by Ed Meitner and
>> Bob Gendron in an AES paper. To their surprise, they found -- and I
>> duplicated their work -- that riding on the DC power rail supplying
>> the ICs in a DAC was the audio signal described by the data that was
>> being processed. If you think about it, this is astonishing.
>
> It's not really astonishing, but it *is* a sign of poor system design.

I agree Stewart, But it is also not uncommon in cost-compromised
players with inadequate power supplies. Again, practical implementation
of theoretically perfect circuit topologies can leave a lot to be desired.

> See any current Meridian player for a fine example of how it should
> be done - or just hook up any old transport to a Benchmark DAC-1.

Yup. The Benchmark is an extraordinary product. Its designer, BTW,
claims that it will pass EEC RF emission standards with its cover
removed, which, if true, is a testament to the layout of its pcb.

> It's unfortunate that your august publication still makes the fatal
> error of assuming that a DAC which actually *is* sensitive to
> different transports is somehow 'superior', when the plain *fact*
> is that it's basically broken.

I am not sure that Stereophile has said this, at least not since the
early 1990s (if then). When you say "still makes" this error, Stewart,
are you aware of a recent instance?

John Atkinson
Editor, Stereophile

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

Stewart Pinkerton <patent3@dircon.co.uk> wrote:
> On 3 Jul 2004 14:55:52 GMT, Stereophile_Editor@Compuserve.com (John
> Atkinson) wrote:

> >Toni <post_to_usenet@hotmail.com> wrote in message
> > news:<xMfFc.13105$%_6.3916@attbi_s01>...
> >> En John Atkinson va escriure en 2 Jul 2004 00:12:33 GMT:
> >> >I wish that were the case. Stereophile has published quite a lot of
> >> > work showing, basically, that jitter propagates through a D/A system.
> >> > It can be low-pass filtered but not eliminated unless heroic
> >> > measures are employed.
> >>
> >> AFAIK, for jitter, you only need to read the CD into your laptop
> >> computer and then play it through the optical out to your
> >> amplifier.
> >
> >As I said, "heroic" measures. Not many audiophiles will tolerate a
> >delay of 60+ minutes before they can hear their music.

> Absolute rubbish if you know *anything* about the technology. A delay
> of 4 seconds is more than adequate for *total* buffering and
> reclocking. Ask Meridian, who do use this technique in their 800
> series. One might have hoped that the Editor of Stereophile would be
> aware of this. One might also have hoped that he'd be aware of
> asynchronous resampling, as used by the superb (and state of the art,
> and reviewed by Stereophile) Benchmark DAC-1..............

I assumed he meant that it takes 60+ minutes to read a CD into a
laptop and play it through a 'real' stereo system. Not that it
makes much sense that way; it takes me a few minutes at most to burn
a CD to my hard drive.

--

-S.
"We started to see evidence of the professional groupie in the early 80's.
Alarmingly, these girls bore a striking resemblance to Motley Crue." --
David Lee Roth

 

[Toni]

Archived from groups: rec.audio.high-end (More info?)

En Isaac Wingfield va escriure en Wed, 07 Jul 2004 07:08:24 GMT:

>In article <_CdGc.27466$IQ4.7481@attbi_s02>,
> Toni <post_to_usenet@hotmail.com> wrote:
>
>>The maniacs could even
>> replace the card's crystal by a high-quality custom-made one (not
>> very expensive, most electronics shops will order them for you).
>
>Curious as to what you mean by "high quality" in this context?
>
>What, exactly, is "low quality" about the stock one?

Hi Isaac

AFAIK, "low quality" cristals are those being more bulnerable to
microphonics and temperature changes. It depends mostly on
manufacturing and packaging conditions (that is cristal in
capsule packaging). Some people debate about different stability
of same frequency cristals on old big containers, normal
containers or micro-miniature packaging.

Small but constant frequency errors and long term variation with
temperature are probably not critical for this application, but
microphonics could cause more jitter than the one it is trying to
cure in the first place.

Toni

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

On Wed, 07 Jul 2004 07:10:04 GMT, Stereophile_Editor@compuserve.com
(John Atkinson) wrote:

>Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
>news:<Kc5Gc.24114$MB3.7322@attbi_s04>...
>> On Sun, 04 Jul 2004 07:39:58 GMT, Stereophile_Editor@compuserve.com
>> (John Atkinson) wrote:
>>> if you doubt the articles in Stereophile, I suggest you read Barry
>>> Blesser's compendium on digital audio in the October 1978 issue of
>>> the Journal of the Audio Engineering Society, where the problems of
>>> word-clock jitter were first described in an audio context (to the
>>> best of my knowledge). A Stereophile article showing how different
>>> analog signals can be reconstructed from identical bitstreams can
>>> be found at http/www.stereophile.com/reference/1290jitter. This,
>>> too, is based on an AES paper, this time by Stephen Harris, and a
>>> third article can be found at
>>> http/www.stereophile.com/reference/590jitter. A fourth article,
>>> showing measured differences, goes up in the stereophile.com
>>> archives on Monday.
>>
>> Hardly new knowledge. Jitter has been acknowledged as a problem in
>> digital audio since the '60s - telecomms being several decades
>> ahead of so-called 'high end' audio, as usual.
>
>With respect Stewart, I think you are being a tad disingenuous here.
>Yes, the problems of jitter were known in telecommunications work.
>But it was only with the advent of practical digital audio in the
>1970s, with the pioneering work of the BBC in the UK, Sony, NHK and
>Denon in Japan, and 3M and the late Dr. Thomas Stockham at Soundstream
>in the US, that digital theory became applicable to audio.

Hardly. Digital theory was *always* applicable to audio, and most of
telecomms *is* audio.................

> Far from
>the Blesser paper that I mentioned being "decades" behind, it was
>published in 1978, just a handful of years after the first digital
>audio devices were prototyped.

And well before CD was launched.

> Bob Stuart's work on jitter reduction
>in Meridian players followed the Blesser paper by a decade, but that
>was still just 5 years after the commercial launch of the CD medium.

Yes, and Bob Stuart has consistently been at least a decade ahead of
the so-called 'high end' brands in real engineering.

>>> You may think it unlikely, Mr. Seufert, but that doesn't mean it
>>> doesn't exist. The problem is that there is no clock in the disc
>>> data.
>>
>> Exactly! In a standalone player, the only clock is the free-running
>> DAC clock, which also controls the data-reading servo. Hence, *if*
>> that clock has vanishingly low phase noise, and *if* the power
>> supplies are absolutely clean, there can be no jitter in the output
>> signal, other than that of the A/D converter used to make the
>> original digital master.
>
>I really don't think we disagree, Stewart. I would say that my position
>is that your word "if" conceals a multitude of design sins on the part
>of product designers, sins that, according to the measurements
>published in Stereophile and other review magazines, allow time-base
>errors to propagate through to the recovered analog signal.

And this atrocious incompetence is signally most noticeable in
products which *you* place in Class A..............................

> I agree
>with you that such design sins are poor engineering; I would merely
>point out to you and Mr. Mr. Seufert that because something in theory
>can be made to be perfect, that doesn't mean all real-world solutions
>are also perfect.

Agreed - so why do you keep promoting grossly overpriced mediocrity
such as the Mark Levinson and Forsell CD devices? Not to mention your
earlier espousal of such ludicrously incompetent trash as the YBA
'blue laser' player?

>>> Around 12 years ago, this subject was examined by Ed Meitner and
>>> Bob Gendron in an AES paper. To their surprise, they found -- and I
>>> duplicated their work -- that riding on the DC power rail supplying
>>> the ICs in a DAC was the audio signal described by the data that was
>>> being processed. If you think about it, this is astonishing.
>>
>> It's not really astonishing, but it *is* a sign of poor system design.
>
>I agree Stewart, But it is also not uncommon in cost-compromised
>players with inadequate power supplies. Again, practical implementation
>of theoretically perfect circuit topologies can leave a lot to be desired.

However, both Meridian and Arcam seem able to provide essentially
immaculate performance without crippling price tags............

>> See any current Meridian player for a fine example of how it should
>> be done - or just hook up any old transport to a Benchmark DAC-1.
>
>Yup. The Benchmark is an extraordinary product. Its designer, BTW,
>claims that it will pass EEC RF emission standards with its cover
>removed, which, if true, is a testament to the layout of its pcb.

Indeed - so why is it not right at the top of your recommended list,
with about ten dollar signs?

>> It's unfortunate that your august publication still makes the fatal
>> error of assuming that a DAC which actually *is* sensitive to
>> different transports is somehow 'superior', when the plain *fact*
>> is that it's basically broken.
>
>I am not sure that Stereophile has said this, at least not since the
>early 1990s (if then). When you say "still makes" this error, Stewart,
>are you aware of a recent instance?

I confess that I have rather lost interest in S'pile since about 2001,
as it regrettably seemed to have fallen into the TAS morass, with
little of real interest to say to the inquiring mind................

If you have 'seen the light', and become more coherent (yuk, yuk),
then I'll be glad to reconsider.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
news:<cY4Hc.44459$Oq2.8565@attbi_s52>...
> On Wed, 07 Jul 2004 07:10:04 GMT, Stereophile_Editor@compuserve.com
> (John Atkinson) wrote:
> >Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
> >news:<Kc5Gc.24114$MB3.7322@attbi_s04>...
> >> It's unfortunate that your august publication still makes the fatal
> >> error of assuming that a DAC which actually *is* sensitive to
> >> different transports is somehow 'superior', when the plain *fact*
> >> is that it's basically broken.
> >
> >I am not sure that Stereophile has said this, at least not since the
> >early 1990s (if then). When you say "still makes" this error, Stewart,
> >are you aware of a recent instance?
>
> I confess that I have rather lost interest in [Stereophile] since about
> 2001, as it regrettably seemed to have fallen into the TAS morass,
> with little of real interest to say to the inquiring mind...

In other words, you cannot qote an instance of this "still" being said
in Stereophile. Oh well...

John Atkinson
Editor, Stereophile

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

In article <ivOFc.20695$%_6.4910@attbi_s01>,
Stereophile_Editor@compuserve.com (John Atkinson) wrote:

> In message <cc3j7f02icj@news2.newsguy.com>
> Timothy A. Seufert (tas@mindspring.com) wrote:
> > I'm afraid that I'm not inclined to give Stereophile articles much
> > credence when it comes to technical matters, having read many a howler
> > whenever said articles are brought to my attention.
>
> Arguing by credential is never very productive, Mr. Seufert.

To be sure, but if you wish to cite Stereophile as a serious technical
authority, you'd better be prepared for howls of derision.

> In the meantime, if you doubt the articles
> in Stereophile, I suggest you read Barry Blesser's compendium on
> digital
> audio in the October 1978 issue of the Journal of the Audio
> Engineering
> Society, where the problems of word-clock jitter were first described
> in an audio context (to the best of my knowledge).

I don't have access to that article (it may be on the AES web site, but
if it is, I can't get at it because the AES web site is down at the
moment). However, I assure you that I am well aware of jitter issues
through first-hand experience in a different but related field.

> A Stereophile article showing how different analog signals can be
> reconstructed from identical bitstreams can be found at
> http/www.stereophile.com/reference/1290jitter. This, too, is based
> on
> an AES paper, this time by Stephen Harris, and a third article can be
> found at http/www.stereophile.com/reference/590jitter. A fourth
> article,
> showing measured differences, goes up in the stereophile.com archives
> on
> Monday. Before you dismiss all this work as "howlers," I politely
> suggest you should read it.

That word clock jitter can affect the output of a DAC is not
controversial, no matter how much you try to present it as such.

I note that you frequently attempt to build a strawman of the "bits is
bits" crowd as a bunch of ignoramuses who believe such jitter is
completely inconsequential. No, we just think it's not an issue in
systems designed by competent engineers, and that jitter cannot
mystically propagate across barriers it shouldn't be able to cross.

Speaking of such barriers, what we have in fact been discussing is the
notion that jitter in the positions of pits and lands on a CD can affect
the analog output of common CD playback system designs, even when there
are no uncorrectable bit errors induced by pit/land position errors.
Would you care to address that topic?

> > The practice is that there is an oscillator feeding the DAC clock input.
> > In order for your claim to be true, somehow random variations away from
> > ideal placement of pit/land transitions on the disc must affect that
> > oscillator's jitter. This strikes me as more than a little unlikely.
>
> You may think it unlikely, Mr. Seufert, but that doesn't mean it
> doesn't
> exist. The problem is that there is no clock in the disc data.

Actually, the _feature_ is that there is no clock in the disc data.

> (The
> problem of an AES/EBU datastream, where there is a clock embedded in
> the data is somewehat different, but manifests itself in a similar
> manner in the DAC.) The CD player's crystal oscillator you mention
> must
> therefore control both the DAC and the disc rotation.

This is not a problem.

> The signal
> retrieved
> from the rotating disc is actually analog in nature.

Not in any sense that most engineers would think of as "analog". There
are two states, and binary information is encoded by the spatial
location of transitions between the two states. Absolute signal levels
do not matter so long as the contrast is sufficient to reliably
discriminate between the two states and avoid false edge transitions.
These are the characteristics typical of a digital signaling system, not
analog.

> Much processing
> is
> therefore required to reconstruct a digital bitstream to be fed to the
> DAC, including some kind of memory buffer.

But of course it requires a memory buffer; couldn't perform the
essential decoding and error correction steps without a buffer. And
that buffer also happens to be why jitter is not an issue during
playback.

> As I wrote, timing uncertainties in the raw data retrieval do appear
> to
> propagate through this system, resulting in measurable effects in the
> recovered analog signal. Things like buffers and PLLs low-pass filter
> the timing uncertainty but do not eliminate it, unfortunately.

Here we begin to see how you lead yourself astray. A PLL buffer does in
some ways act like a low-pass filter for jitter, but the buffer in a CD
player does not involve a PLL. In any reasonably well designed player,
that buffer's output and the DAC input are both clocked by a single
reference clock, not a clock synthesis or recovery circuit such as a
PLL. This means the jitter of the DAC word clock is simply unrelated to
timing uncertainties in the section of the player which decodes raw data
from the disc.

> That a CD-R carrying identical data to a CD can sound different is
> described in in article at http/www.stereophile.com/asweseeit/523.

This article is singularly unimpressive, being little more than a
collection of anecdotes. The one bit that is interesting mentions a
blind test but gives absolutely no details about the test design, so
that the reader cannot but wonder whether the test truly was blind...

> > The only remotely plausible explanation for such an effect that I've
> > ever seen put forth is power supply noise, but there are problems with
> > that idea too. The biggest being, why should there be any more power
> > supply noise than normal? The number of CMOS switching events in the
> > digital section of the player should not be any higher on average.
>
> Around 12 years ago, this subject was examined by Ed Meitner and Bob
> Gendron in an AES paper. To their surprise, they found -- and I
> duplicated their work -- that riding on the DC power rail supplying
> the
> ICs in a DAC was the audio signal described by the data that was being
> processed. If you think about it, this is astonishing. Check it out.

That would indeed be astonishing, but you are astonishingly failing to
consider a far simpler and more likely explanation, given what a DAC
_does_.

> Back to the subject of supposed technical "howlers" in Stereophile
> magazine. As editor, I try hard to keep the incidence of errors to an
> acceptably low level. When the subject has come up before on r.a.h-e
> or
> other newsgroups, I have therefore asked the posters to be specific.
> If,
> indeed, there is a factual error in the magazine, I need to be aware
> of
> it. There are 4 specific examples in the groups.google.com record:
>
> Example 1: A John Busenitz mentioned, as you have just done, Mr.
> Seufert,
> the purported high incidence of techical errors in Stereophile. It
> turned
> out that Mr. Busenitz was only able to cite one example of such an
> error,
> but it didn't actually appear in Stereophile. It was in a book written
> by
> Stereophile's one-time technical editor, Robert Harley.

Well then, let us examine Mr. Harley's writing _for your magazine_.

Here he makes more or less the same error you are making now:

"Jitter is most often introduced by mechanical imperfections in digital
audio-storage devices. A CD player's rotational servo, for example, can
introduce time-base errors (jitter) in the recovered signal if its speed
varies even by a tiny amount."

From: http/www.stereophile.com/reference/590jitter/index2.html

--
Tim

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

In article <cch49h01dvp@news1.newsguy.com>,
Toni <post_to_usenet@hotmail.com> wrote:

> En Isaac Wingfield va escriure en Wed, 07 Jul 2004 07:08:24 GMT:
>
> >In article <_CdGc.27466$IQ4.7481@attbi_s02>,
> > Toni <post_to_usenet@hotmail.com> wrote:
> >
> >>The maniacs could even
> >> replace the card's crystal by a high-quality custom-made one (not
> >> very expensive, most electronics shops will order them for you).
> >
> >Curious as to what you mean by "high quality" in this context?
> >
> >What, exactly, is "low quality" about the stock one?
>
> Hi Isaac
>
> AFAIK, "low quality" cristals are those being more bulnerable to
> microphonics and temperature changes. It depends mostly on
> manufacturing and packaging conditions (that is cristal in
> capsule packaging). Some people debate about different stability
> of same frequency cristals on old big containers, normal
> containers or micro-miniature packaging.
>
> Small but constant frequency errors and long term variation with
> temperature are probably not critical for this application, but
> microphonics could cause more jitter than the one it is trying to
> cure in the first place.

The things you describe affect essentially all crystals to just about
the same extent, assuming they are not actually broken.

In my experience, "high quality" mostly means, w.r.t. crystals,:

(1) the actual frequency of operation is closer to the specified
frequency. But note that even "non-precision" units are well within 50
parts per million, usually closer to ten ppm.

(2) the variation of frequency with temperature is less. But note that
even "ordinary" crystals are not likely to show enough drift to be
measured unless you have some pretty fancy lab equipment.

In almost every inside-a-computer or inside-consumer-electronics
application, ordering a "high quality" crystal without having a serious
understanding of exactly how the circuit was designed would be a waste
of money and might possible result in worse performance, not better. For
a "high quality" application, you need to know whether the crystal is in
a series or parallel configuration, and what the drive level and shunt
capacitance are, for starters.

Especially for "high quality" performance, the design of the oscillator
circuit that the crystal goes in is actually more important than the
crystal itself. Virtually every oscillator circuit used in consumer
audio applications will be of a type that simply cannot take advantage
of a crystal having higher precision or better temperature stability.
Most of the oscillators I've seen are actually pretty incompetently
designed and certainly could not take advantage of a superior crystal.
It's sometimes amazing that they work at all.

I know of exactly one situation where microphonics affected a crystal
oscillator in a way that was detrimental to the operation of the
equipment. The oscillator was the frequency control element for an FM
transmitter, and the microphonics were being caused by a snap-action
thermostat *inside* the crystal oven. That's a big "click" very close to
the crystal; nothing we could do from outside the unit had any noticable
effect on the oscillator. That particular crystal was physically much
larger, and operated at a significantly lower frequency than anything
I've seen in consumer gear. The large size and low frequency exacerbated
it's sensitivity to microphonics.

Isaac

 

[Guest]

Archived from groups: rec.audio.high-end (More info?)

In article <n%eHc.46717$XM6.41060@attbi_s53>,
Stereophile_Editor@compuserve.com (John Atkinson) wrote:

> Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
> news:<cY4Hc.44459$Oq2.8565@attbi_s52>...
> > On Wed, 07 Jul 2004 07:10:04 GMT, Stereophile_Editor@compuserve.com
> > (John Atkinson) wrote:
> > >Stewart Pinkerton <patent3@dircon.co.uk> wrote in message
> > >news:<Kc5Gc.24114$MB3.7322@attbi_s04>...
> > >> It's unfortunate that your august publication still makes the fatal
> > >> error of assuming that a DAC which actually *is* sensitive to
> > >> different transports is somehow 'superior', when the plain *fact*
> > >> is that it's basically broken.
> > >
> > >I am not sure that Stereophile has said this, at least not since the
> > >early 1990s (if then). When you say "still makes" this error, Stewart,
> > >are you aware of a recent instance?
> >
> > I confess that I have rather lost interest in [Stereophile] since about
> > 2001, as it regrettably seemed to have fallen into the TAS morass,
> > with little of real interest to say to the inquiring mind...
>
> In other words, you cannot qote an instance of this "still" being said
> in Stereophile. Oh well...

He missed the Arcam cd player on the July cover, too.

Stephen