Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[zetterstroem]

it's nice to learn something new...

hope that someday i will be able to listen to it....

and maybe kill some more of the myths about digital audio !

[Tomas Danko]

bobkatz wrote on Sat, 23 April 2005 11:29

But this discussion does get down to questions of "is less jitter always better sounding?" And I certainly have experienced audible tradeoffs... relating to the fact that if you reduce the jitter to the lowest you can make it, there will be residual jitter with a particular spectrum that translates to a certain noise floor or distortion floor within the DAC. Which if it hits the ear's sensitivity region could sound worse than if you add a carefully controlled film of white noise to cover that up. Doesn't matter if that white noise was created in the time or frequency domain. Think about it. BK

I'm thinking about it...

Apogee Big Ben, perhaps? This would be the only theoretical reasoning behind their claims, as far as I'm concerned. It's a pity they would probably never let us know, even if this was their approach. And what a pity if it turns out it's not an Apogee exclusive... as they would want to.

Then again, once could debate how much of an 'improvement' this really is.

Sincerely,

Tomas Danko

[Joe Crawford]

Looks like its my turn to eat crow.  A totally new, at least to me, definition of the term “spread spectrum” can be found here: http://en.wikipedia.org/wiki/Spread_spectrum.  Since the term is now used to define a method for spreading the noise spectrum generated by a digital clock, that in order to be correct, even the IEEE will have to start adding the “radio” or “clock” qualifiers to it (http://www.isssta2004.org/).

Occasionally some of the most inovative solutions to problems can come from taking new technology (such as spread spectrum clocking designed to hide PC generated EMI from regulatory testing methods) and applying it to a totally new area (such as covering up the residuals of audio clock jitter).

Daniel, I appologise and wish you the best of luck... hope it sells like crazy...

Joe Crawford

[danlavry]

Tomas Danko wrote on Sat, 23 April 2005 14:42

bobkatz wrote on Sat, 23 April 2005 11:29 But this discussion does get down to questions of "is less jitter always better sounding?" And I certainly have experienced audible tradeoffs... relating to the fact that if you reduce the jitter to the lowest you can make it, there will be residual jitter with a particular spectrum that translates to a certain noise floor or distortion floor within the DAC. Which if it hits the ear's sensitivity region could sound worse than if you add a carefully controlled film of white noise to cover that up. Doesn't matter if that white noise was created in the time or frequency domain. Think about it. BK I'm thinking about it... Apogee Big Ben, perhaps? This would be the only theoretical reasoning behind their claims, as far as I'm concerned. It's a pity they would probably never let us know, even if this was their approach. And what a pity if it turns out it's not an Apogee exclusive... as they would want to. Then again, once could debate how much of an 'improvement' this really is. Sincerely, Tomas Danko

So what is the idea here? Instead of fixing a problem of some jitter at some particular spectrum, one can hide it by raising the level of white noise jitter?
Is it no analogous to taking a poor AD, say with some bad idle tones at around -90dB, and adding a "fix" - enough white noise to make the dynamic range -80dB?

Are you are suggesting that an external clock box, adding jitter to almost all the AD devices on the market (different makes, models and manufacturers)is a good thing? Are you suggesting that almost all AD's have that spectraly offensive type jitter that calls on an external device to add jitter?

As a designer, I can assure you that an internal crystal is a better way to have a spectraly flat jitter. Using an external clock leaves you more opened to non spectraly flat jitter (such as the effect of signals picked up by a cable)...
 
Take a look at the site below. A picture is worth many words...
Look for "Jitter: internal vs. external"

  http://www.lavryengineering.com/forum/phpBB2/viewforum.php?f =1&sid=67ea28f2845fd1cfadd51d2e51191cf0

Regards
Dan Lavry
www.lavryengineering.com

[Tomas Danko]

Dear Mr. Lavry,

I'm not suggesting anything, but merely discussing if by chance this method would be the 'super-secret' one found in the Apogee Big Ben. Until they make it public, we will never know, and I doubt it that they ever will publish this information.


For whatever it is worth, I am firmly seated on your side of the reasoning (both the clocking-issue as well as the overkill sample rate-ditto) and have already followed the related threads for ages by now.

Sincerely,

Tomas Danko

[danlavry]

Tomas Danko wrote on Mon, 25 April 2005 19:47

Dear Mr. Lavry, I'm not suggesting anything, but merely discussing if by chance this method would be the 'super-secret' one found in the Apogee Big Ben. Until they make it public, we will never know, and I doubt it that they ever will publish this information. For whatever it is worth, I am firmly seated on your side of the reasoning (both the clocking-issue as well as the overkill sample rate-ditto) and have already followed the related threads for ages by now. Sincerely, Tomas Danko

Hi Tom,

Thanks for the comments. I am curious to hear your comments about the jitter thread (jitter, internal vs external...) at:

http://www.lavryengineering.com/forum/phpBB2/viewforum.php?f =1&sid=6acfbf0f60559d11c998c2cdc5abc673

I did my best to say it all in one post, and I hope the "picture" does help...

Regards
Dan Lavry
www.lavryengineering.com

[zetterstroem]

hi dan...

this is exactly why i was a bit confused.... and the reason i started this topic

however what daniel says makes sense too...

although we were talking about d/a's... if that makes a difference?

and thanks for the clear illustration on your website... now i can get even a drummer to understand!!  

is there a book on the way??

[danlavry]

zetterstroem wrote on Tue, 26 April 2005 08:15

hi dan... this is exactly why i was a bit confused.... and the reason i started this topic however what daniel says makes sense too... although we were talking about d/a's... if that makes a difference? and thanks for the clear illustration on your website... now i can get even a drummer to understand!!   is there a book on the way??

Yes, a DA is different. We are talking about the use of external vs internal clock for AD, not DA.
In the case of a DA, the data (such as AES or SPDIF) comes with the clock, on the same cable, so there is no external DA clocking, bu definition. The clock needs to be separated from the data by the DA itself.
The jitter of the clock (separated out from the data stream) is important, and the common way to "clean the jitter" is to use a PLL (phase lock loop circuit). There are a few other ways to do a better job.

Regards
Dan Lavry
www.lavryengineering.com

[Tomas Danko]

danlavry wrote on Mon, 25 April 2005 23:12

Tomas Danko wrote on Mon, 25 April 2005 19:47 Dear Mr. Lavry, I'm not suggesting anything, but merely discussing if by chance this method would be the 'super-secret' one found in the Apogee Big Ben. Until they make it public, we will never know, and I doubt it that they ever will publish this information. For whatever it is worth, I am firmly seated on your side of the reasoning (both the clocking-issue as well as the overkill sample rate-ditto) and have already followed the related threads for ages by now. Sincerely, Tomas Danko Hi Tom, Thanks for the comments. I am curious to hear your comments about the jitter thread (jitter, internal vs external...) at:  http://www.lavryengineering.com/forum/phpBB2/viewforum.php?f =1&sid=6acfbf0f60559d11c998c2cdc5abc673 I did my best to say it all in one post, and I hope the "picture" does help...

Hi Mr. Lavry,

To me, the sensible way to approach this on a practical level is to first ask the question 'how bad are the internal clocks of today?'. Can they really be so bad, they need to be 'fixed' by an external clocking device? Now, that would mean they are very bad.

Or perhaps the bad clocking designs is only to be found in less expensive converters, and that leads to the second sensible question: Why spend $1,500 to 'fix' something that costed much less? Instead of spending $1,500 - $2,000 to buy something that doesn't need the 'cure'.

If these questions stated above are valid, I don't see the point to even justify an expensive external clocking device. However, let's pretend that professional converters are lacking in stable clocking and that one has to cough up more than $3,500 to solve the 'problem'. Given this scenario, I'm willing to extend the reasoning further...


I don't think the skin-effect would yield a lesser rise time severe enough to disturb the eye-pattern of the clocking circuit in the converter. I do however believe that funky cable termination can work wonders to sabotage an otherwise healthy connection. In the end, any kind of IC or OP-amp between the master- and slave-clock that has to be at both ends of the cable surely will be a bigger culprit unless the RF interference and ground loops are present (although they shouldn't be in a professional environment).

It's ridiculous to even try and claim less jitter with external clocking, all things considered.

Trying to se the forest among all the trees, or however I should put it, while reading all the correspondence from Apogee at REP it seems to me they fessed up saying that their external clock would not improve a normally-functioning converters internal clock. So that's one discussion not needed to be taken further, in my opinion.

This leaves the topic at hand down to psychoacoustics, and taste. Those are topics better discussed elsewhere, from what I understand with the purely technical focus you are trying to maintain here.


Regardless, I am starting to wonder if this is a take on noise shaping in the time-domain. As far as ordinary noise-shaping goes, even though it's supposed to shift the focus away from the areas where our ears are the most sensitive, I rarely like the result as much as plain dither only. This is my subjective opinion. Maybe one can noise-shape the jitter artifacts in a similar way, shifting the problem away from the midrange frequencies and so on, and maybe this is what's going on here. If so, it will have to be as subjective a result as different noise-shapers, and the mystery would be solved.

Best regards,

Tomas Danko - Sticking to 24/96, all converters running on internal clocking, dither applied last in the chain and no noise-shaping whatsoever. YMMV

[Johnny B]

danlavry wrote on Tue, 26 April 2005 16:38

"[T]he common way to "clean the jitter" is to use a PLL (phase lock loop circuit). There are a few other ways to do a better job.

Dan, will you be kind enough to explain and clarify your statement?

[danlavry]

Johnny B wrote on Tue, 26 April 2005 19:00

danlavry wrote on Tue, 26 April 2005 16:38 "[T]he common way to "clean the jitter" is to use a PLL (phase lock loop circuit). There are a few other ways to do a better job. Dan, will you be kind enough to explain and clarify your statement?

One solution is to insert a SRC in series. The impact of jitter on a sample rate conversion is different than a direct the impact on a DA. It is NOT TRUE that src frees you from jitter problems. With an SRC, The DA side clock has very little jitter (it is a fixed crystal clock), but the jitter on the input side of the SRC does have an impact. It impacts the data itself, which is dependent on the RELATIVE timing between the SRC input and output. But a good SRC may be an improvement - I will leave that to the ears...

Well, I did not want to advertise my gear, but you asked - My CrystalLock technology is far superior to PLL's, SRC's and the other solutions. Check my website for description of it, under support, DA924 Manuel, page 11 ...

Regards
Dan Lavry
www.lavryengineering.com

[Johnny B]

Thanks Dan.

I've heard that other people have some problems with PLL too.

[Nika Aldrich]

Tomas Danko wrote on Sat, 23 April 2005 09:42

I'm thinking about it... Apogee Big Ben, perhaps?

I don't think so.  I'm pretty sure that Weiss' box is intended to mask jitter going to the D/A by adding additional jitter of a white noise variety - again, think of it like dither for jitter, though I hate to go down that path.

Big Ben (I believe) attempts to reshape the existing jitter to conform to a curve that is more harmonically pleasing.  We could think of Weiss as pure dither and Apogee as noise-shaping.  While the frequency response of jitter does not affect us quite like the frequency response of distortion in audio, there are certain frequencies that we are less prone to hear, and if certain spurious components are present we may hear them more noticeably.

The trick of each of these boxes seems to be to treat the jitter.  One does so by masking the spurious components.  The other by reshaping the floor altogether.

FWIW, Dan's "Crystallock" technology is much more akin to Apogee's solution - reshaping the existing jitter so that it is less audibly offensive.  Dan uses very lengthy buffers and PLLs to "reshape" the jitter such that it has extremely high amplitude at very low frequencies where it is less likely to cause audible distortion, thereby removing the jitter at the higher frequencies where it will cause more audible distortion.

Nika

[Nika Aldrich]

Tomas Danko wrote on Tue, 26 April 2005 11:57

To me, the sensible way to approach this on a practical level is to first ask the question 'how bad are the internal clocks of today?'. Can they really be so bad, they need to be 'fixed' by an external clocking device? Now, that would mean they are very bad.

I think it is certainly possible that the clocking in internal devices measures better in an overall amplitude measurement.  An overall amplitude measurement, however, does not necessarily indicate how the jitter will manifest itself upon listening.  It is certainly possible to have two boxes - one with an overall jitter of, say, 1ns and another with an overall jitter amplitude of, say, 10ns, but the latter to sound significantly better.

If internal clocks have lower amplitude but the frequency distribution of the jitter is more audibly destructive then external boxes may still have a place?  Yes?

Nika

[Nika Aldrich]

danlavry wrote on Mon, 25 April 2005 18:12

Hi Tom, Thanks for the comments. I am curious to hear your comments about the jitter thread (jitter, internal vs external...) at:  http://www.lavryengineering.com/forum/phpBB2/viewforum.php?f =1&sid=6acfbf0f60559d11c998c2cdc5abc673 I did my best to say it all in one post, and I hope the "picture" does help... Regards Dan Lavry www.lavryengineering.com

Dan,

Yes, very helpful.  I saw the various things in your picture that can add jitter to the clock signal, such as the clock output drivers, the PLLs, and the cables.  Am I to take it from this post that one should use cables that impart less opportunity for jitter on the clock signal if they NEED to clock to an external device for some reason (such as a D/A converter)?

I see that Apogee has a line of "Ultra Low Jitter" cables.  Do you suppose that those might address this need?

Nika