Sign in with
Sign up | Sign in
Your question

Real world jitter numbers

Last response: in Home Audio
Share
Anonymous
May 1, 2005 7:18:01 PM

Archived from groups: rec.audio.pro (More info?)

I was looking at the manual for the RME ADI-2 which describes their
SteadyClock feature, and they're saying that a typical S/PDIF clock has ~10
ns of jitter while thier clock improves that to 2 ns. I'm finding it hard to
believe that this could have an audible impact. Even a 20 ns jitter is less
than 0.1% slip -per sample- at 44.1 KHz. This just doesn't sound like enough
to hear, but since I've never done any comparisions myself I can't say.

Is there really a difference in the sound at this level of precision?

Sean
Anonymous
May 2, 2005 11:31:39 AM

Archived from groups: rec.audio.pro (More info?)

Sean Conolly wrote:

> I was looking at the manual for the RME ADI-2 which describes their
> SteadyClock feature, and they're saying that a typical S/PDIF clock
> has ~10 ns of jitter while thier clock improves that to 2 ns. I'm
> finding it hard to believe that this could have an audible impact.

Taken in isolation, the jitter performance of a SP/DIF clock is is
sonically irrelevant as long as it isn't so bad that it causes data to
be received incorrectly. IME it takes a metric ton of jitter to cause
data to be received incorrectly with many modern digital appliances.

For example, if I send a SPDIF stream from a digital recorder to a PC,
the PC doesn't use the clock that is embedded in the SPDIF data to
reconstruct the analog signal. It's digital-out, digital-in. As long
as the numeric values of the data don't change, no harm done.

If converters in the PC are used later on to reconstruct analog
signals, they use a different clock - the PC interface's clock.

The most likely audible problem would arise if the audio interface
were used with an external converter and the external converter was
very weak-minded. Good DACs don't take the SPDIF embedded clock as
gospel, they do some processing to ensure that the conversion process
is as clean as possible. This processing is no longer complex to
implement or expensive.

I've seen even moderately-priced DACs reduce jittter to less than -110
dB when the SPDIF signal was so jittery that it sounded like it had
vibrato, before correction.

> Even a 20 ns jitter is less than 0.1% slip -per sample- at 44.1 KHz.

That could turn out to be something like -60 dB jitter, which might be
audible.
Anonymous
May 2, 2005 12:37:21 PM

Archived from groups: rec.audio.pro (More info?)

Sean Conolly wrote:
> I was looking at the manual for the RME ADI-2 which describes their
> SteadyClock feature, and they're saying that a typical S/PDIF clock
> has ~10 ns of jitter while thier clock improves that to 2 ns.

Where is the manual? I don't know exactly what jitter that this is
referring to. The core jitter of the actual converters have to to be
better then this, that is in the sub ns ranges for 24 bit converters.
Are you sure this isn't 100ps to 20ps? Maybe this is referring to some
unbufferd communication jitter between units.

>I'm
> finding it hard to believe that this could have an audible impact.
> Even a 20 ns jitter is less than 0.1% slip -per sample- at 44.1 KHz.

You don't need much core jitter to cause relatively large distortion
compared to the ideal spec of the converter, especially at 24 bits. If
this is just the jitter between systems sharing common clocks, then this
value is not so important, as it will be attenuated by jitter
attenuaters prior to the converters. I am not familiar with the specific
technical standard of S/PDIF clocks, but in general, if this spec is
just the transfer clock, not the converter clock itself, its spec,
within reason, might have little, or no effect on the distortion at all.
Usually, a jitter attenuater (phases locked loop usually) locks onto the
clock and produces a new clean clock. It just needs an adequate lock
range.

> This just doesn't sound like enough to hear, but since I've never
> done any comparisons myself I can't say.

To calculate the effect of basic jitter on the converter itself one can
approximate by:

dv/dt = 2.pi.f.Vo

We have the error due to jitter as

dv/V = 2.pi.f.V.dt

at 20khz, and a dt=tj=10ns

dv/V = relative error = 1.26m, or about -58db distortion, or about 10
bits.

1khz at 2ns => 12.5u => -98db, or around 16 bits. This is a bit low for
a 24 bit resolution converter. Typically, one might expect -110db
to -120db or better for an A/D chip itself.

>
> Is there really a difference in the sound at this level of precision?
>
> Sean

Who knows:-)

As far as the spec is concerned, I would need to see the manual to
understand what specific clock the spec was referring to. I am not a
specialist in audio PCM standards or products. My knowledge is more
general in this area.

From the numbers, if this is the real clock to the converters, it isn't
that good from a technical purest point of view. i.e An ideal 24 bit,
20Khz, requires an impossible jitter spec. A 10ns clock is certainly
less then 16 bits at 1khz. Whether or not this is audible is up to tose
golden ears boys.

If its not the actual clock, I don't know without further information.

Kevin Aylward
informationEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
Related resources
Anonymous
May 2, 2005 12:59:02 PM

Archived from groups: rec.audio.pro (More info?)

On Sun, 1 May 2005 15:18:01 -0400, "Sean Conolly"
<sjconolly_98@yaaho.com> wrote:

>I was looking at the manual for the RME ADI-2 which describes their
>SteadyClock feature, and they're saying that a typical S/PDIF clock has ~10
>ns of jitter while thier clock improves that to 2 ns. I'm finding it hard to
>believe that this could have an audible impact. Even a 20 ns jitter is less
>than 0.1% slip -per sample- at 44.1 KHz. This just doesn't sound like enough
>to hear, but since I've never done any comparisions myself I can't say.
>
>Is there really a difference in the sound at this level of precision?
>
>Sean
>
>
>
SPDIF is a serial data stream, and the jitter only has an effect on
the data recovery clock. Provided it is good enough to recover correct
bits - which clearly it does because it works, then it will present
clean data to the DAC buffer. It is the DAC's clock, not the SPDIF
stream, that determines the quality of the resulting audio.

As long as it isn't making loud crackling and crunching noises it is
doing its job just fine.

d

Pearce Consulting
http://www.pearce.uk.com
Anonymous
May 2, 2005 7:49:13 PM

Archived from groups: rec.audio.pro (More info?)

Check what Blers (Block Errors) mean and how many it takes before your
digital playback source runs into an Error 32 (mute) situation. This will
give you the history of perceived jitter in audio. There are some other
issues in imaging and depth, but almost all of these have been overcome by
better reclocking. Is 0.1% significant? Maybe not, but if you can get
better than you only have to decide whether the sound of what you're
producing is good enough or not. If it's all in the output product, then it
probably needs fixing. If it's fine all the way through, why worry?

Music and engineering are an art of the ear, not technical measurements.

--


Roger W. Norman
SirMusic Studio
http://blogs.salon.com/0004478/
"Sean Conolly" <sjconolly_98@yaaho.com> wrote in message
news:5Ffde.120970$f%4.14393@bignews1.bellsouth.net...
> I was looking at the manual for the RME ADI-2 which describes their
> SteadyClock feature, and they're saying that a typical S/PDIF clock has
~10
> ns of jitter while thier clock improves that to 2 ns. I'm finding it hard
to
> believe that this could have an audible impact. Even a 20 ns jitter is
less
> than 0.1% slip -per sample- at 44.1 KHz. This just doesn't sound like
enough
> to hear, but since I've never done any comparisions myself I can't say.
>
> Is there really a difference in the sound at this level of precision?
>
> Sean
>
>
>
>
Anonymous
May 3, 2005 4:26:21 AM

Archived from groups: rec.audio.pro (More info?)

On Mon, 02 May 2005 08:37:21 GMT, "Kevin Aylward"
<see_website@anasoft.co.uk> wrote:

<elegant analysis clipped for bandwidth>

I take back my earlier harsh words and apologize. You may
after all prove to be a good newsgroup citizen. Welcome,

Chris Hornbeck
Anonymous
May 3, 2005 4:26:22 AM

Archived from groups: rec.audio.pro (More info?)

Chris Hornbeck wrote:
> On Mon, 02 May 2005 08:37:21 GMT, "Kevin Aylward"
> <see_website@anasoft.co.uk> wrote:
>
> <elegant analysis clipped for bandwidth>
>
> I take back my earlier harsh words and apologize. You may
> after all prove to be a good newsgroup citizen. Welcome,

Indeed! Occasional bad manners and all I pay very close
attention and also hope for his continued presence.

Just a tip, Kevin. It's not necessasary to blow out
another's candle to make one's own glow brighter.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
Anonymous
May 3, 2005 2:25:13 PM

Archived from groups: rec.audio.pro (More info?)

Bob Cain wrote:
> Chris Hornbeck wrote:
>> On Mon, 02 May 2005 08:37:21 GMT, "Kevin Aylward"
>> <see_website@anasoft.co.uk> wrote:
>>
>> <elegant analysis clipped for bandwidth>
>>
>> I take back my earlier harsh words and apologize. You may
>> after all prove to be a good newsgroup citizen. Welcome,
>
> Indeed! Occasional bad manners and all I pay very close
> attention and also hope for his continued presence.
>
> Just a tip, Kevin. It's not necessasary to blow out
> another's candle to make one's own glow brighter.

I don't, as a rule.

Count the number of personal insults made by others to me, and the
number of personal insults made by me in retaliation.

Kevin Aylward
informationEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
Anonymous
May 3, 2005 2:33:10 PM

Archived from groups: rec.audio.pro (More info?)

Chris Hornbeck wrote:
> On Mon, 02 May 2005 08:37:21 GMT, "Kevin Aylward"
> <see_website@anasoft.co.uk> wrote:
>
> <elegant analysis clipped for bandwidth>
>
> I take back my earlier harsh words and apologize. You may
> after all prove to be a good newsgroup citizen. Welcome,
>
> Chris Hornbeck

Well, much appreciated.

Kevin Aylward
informationEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
Anonymous
May 3, 2005 9:36:48 PM

Archived from groups: rec.audio.pro (More info?)

Kevin Aylward wrote:

> Count the number of personal insults made by others to me, and the
> number of personal insults made by me in retaliation.

Not much of a stoic, eh? :-)


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein
!