Why does source matter when I'm only using a Digital Output?

Oct 16, 2006 at 1:43 PM Thread Starter Post #1 of 19

KrooLism

500+ Head-Fier
Joined
Sep 23, 2006
Posts
773
Likes
13
I mean... what's the difference between a $3000 CD player and a $200 DVD player if I am using a digital output?

I mean, isn't digital just ones and zeros? Isn't it up to the DAC to do most of the work?

How does a good source's digital signal differ from a not so good source?
 
Oct 16, 2006 at 2:58 PM Post #3 of 19
Oct 16, 2006 at 5:28 PM Post #4 of 19
Quote:

Originally Posted by KrooLism
I mean... what's the difference between a $3000 CD player and a $200 DVD player if I am using a digital output?



$2800

While many will talk about jitter there is little hard evidence that jitter at the level found in consumer grade DVD/CD players has any audible effect.

see...
http://www.regonaudio.com/Jitter.html

"Potentially" may well be the operative word in most cases. Certainly, jitter of sufficient amount can be audible. The Stereophile Test CD NO.2 provides a demonstration involving 11 kHz jittered by 10 nanoseconds at 4 kHz frequency. There is an audible roughness to the sound compared with an unjittered tone. But consider what is involved here: Fifty times the 200 picosecond threshold we just discussed; jitter of fixed frequency, which creates sidebands of definite frequencies, 7 kHz and 15 kHz, the 7 kHz one at a region of high hearing sensitivity; and a sustained pure tone at high level as a signal. No wonder the result is audible! The relevance of the jitter levels of contemporary high-quality CD players to audible performance is less obvious. The evidence adduced in the press tends to be of the anecdotal variety. And, since jitter is only one of a considerable number of potential electronic failings of CD players, it is not easy on an anecdotal basis to isolate the sonic effect of jitter alone.

From Dunn's 1999 paper

A recent paper Eric Benjamin and Benjamin Gannon - "Theoretical and Audible Effects of Jitter on Digital Audio Quality", AES, 1998. describes practical research that found the lowest jitter level at which the jitter made a noticeable difference to be about 10ns rms. This was with a
high level test sine tone at 17kHz. With music none of their subjects found jitter
below 20ns rms to be audible.


There may be other differences between sources but Jitter is unlikely to be one worth worrying about.
 
Oct 16, 2006 at 9:07 PM Post #5 of 19
Quote:

Originally Posted by hciman77
$2800

While many will talk about jitter there is little hard evidence that jitter at the level found in consumer grade DVD/CD players has any audible effect.

see...
http://www.regonaudio.com/Jitter.html

"Potentially" may well be the operative word in most cases. Certainly, jitter of sufficient amount can be audible. The Stereophile Test CD NO.2 provides a demonstration involving 11 kHz jittered by 10 nanoseconds at 4 kHz frequency. There is an audible roughness to the sound compared with an unjittered tone. But consider what is involved here: Fifty times the 200 picosecond threshold we just discussed; jitter of fixed frequency, which creates sidebands of definite frequencies, 7 kHz and 15 kHz, the 7 kHz one at a region of high hearing sensitivity; and a sustained pure tone at high level as a signal. No wonder the result is audible! The relevance of the jitter levels of contemporary high-quality CD players to audible performance is less obvious. The evidence adduced in the press tends to be of the anecdotal variety. And, since jitter is only one of a considerable number of potential electronic failings of CD players, it is not easy on an anecdotal basis to isolate the sonic effect of jitter alone.

From Dunn's 1999 paper

A recent paper Eric Benjamin and Benjamin Gannon - "Theoretical and Audible Effects of Jitter on Digital Audio Quality", AES, 1998. describes practical research that found the lowest jitter level at which the jitter made a noticeable difference to be about 10ns rms. This was with a
high level test sine tone at 17kHz. With music none of their subjects found jitter
below 20ns rms to be audible.


There may be other differences between sources but Jitter is unlikely to be one worth worrying about.




Hmmm, well I can quote from experience on the "clock mod" that was installed on my CD transport. Didn't cost much, but certainly has an obvious positive impact upon how well my system sounds.
Still, if you choose not to believe my comments because it's counter to your suggestion, that's your call.
 
Oct 16, 2006 at 10:32 PM Post #6 of 19
Quote:

Originally Posted by hciman77
"Potentially" may well be the operative word in most cases. Certainly, jitter of sufficient amount can be audible. The Stereophile Test CD NO.2 provides a demonstration involving 11 kHz jittered by 10 nanoseconds at 4 kHz frequency. There is an audible roughness to the sound compared with an unjittered tone. But consider what is involved here: Fifty times the 200 picosecond threshold we just discussed; jitter of fixed frequency, which creates sidebands of definite frequencies, 7 kHz and 15 kHz, the 7 kHz one at a region of high hearing sensitivity; and a sustained pure tone at high level as a signal. No wonder the result is audible!



While that's apparently an extreme example of jitter to make it obvious even with test tones, generally I consider music to have the much higher potential to reveal jitter than test tones, as every other form of distortion.

Indeed it's hard to isolate jitter as cause for sonic effects. All I can say is that every digital transport I've tried has had clear sonic impact in my system, and this with two different DACs (Theta Pro basic II and Bel Canto DAC2) -- I even doubt that the concerned transports have considerably different jitter rates (although the jitter patterns may absolutely differ).
.
 
Oct 16, 2006 at 11:58 PM Post #7 of 19
Quote:

Originally Posted by hciman77
$2800

see...
http://www.regonaudio.com/Jitter.html

.




I love the part about ones and zero's being abstractions. How many times have people posted on this board about bit perfect its just 1 and 0's, all the same. Do we think there are little 1's and zero's floating thru the wires? We are talking electricity not 1 ands zero's.
 
Oct 17, 2006 at 12:27 AM Post #8 of 19
Quote:

Originally Posted by hciman77
From Dunn's 1999 paper

A recent paper Eric Benjamin and Benjamin Gannon - "Theoretical and Audible Effects of Jitter on Digital Audio Quality", AES, 1998. describes practical research that found the lowest jitter level at which the jitter made a noticeable difference to be about 10ns rms. This was with a
high level test sine tone at 17kHz. With music none of their subjects found jitter
below 20ns rms to be audible.



Very interesting and in sharp contrast to his earlier publications, where drew the line three orders of magnitude lower. From his 1992 AES paper in section 3.3 on the Audibiity of sampling jitter.

Quote:

This plot can be used as a specification for allowable sampling jitter in Nyquist sampled systems.

At 20 Khz the peak to peak sampling jitter must be less than 20 ps, increasing at 6db per octave for lower frequencies until approximately 500 Hz where the limit is 1 ns. Below 200 Hz the jitter may up to 500 ns in amplitude before the sidebands could become audible.


Quite a difference. Do you have a URL to the full 1999 paper?


Cheers

Thomas
 
Oct 17, 2006 at 8:46 AM Post #9 of 19
Quote:

Originally Posted by regal
I love the part about ones and zero's being abstractions. How many times have people posted on this board about bit perfect its just 1 and 0's, all the same. Do we think there are little 1's and zero's floating thru the wires? We are talking electricity not 1 ands zero's.


I don't understand the point. The electricity embodies ones and zeros.

Can't we assume that a CD player in proper working order conforms to the Red Book standard? If it does then we have an idea of the error rate in the digital data (roughly one error bit per second). What's the big deal about clocking those digits? The watch on my wrist cost $40 and it's accurate to about one second per month. How accurate does it need to be? If we have the right digits going at the right speed then don't we have music? Isn't the rest about DAC and filtering (ie not the transport).

I ask these questions because I don't know the answers.
 
Oct 17, 2006 at 10:26 AM Post #10 of 19
Quote:

Originally Posted by DennyL
I don't understand the point. The electricity embodies ones and zeros.

Can't we assume that a CD player in proper working order conforms to the Red Book standard? If it does then we have an idea of the error rate in the digital data (roughly one error bit per second). What's the big deal about clocking those digits? The watch on my wrist cost $40 and it's accurate to about one second per month. How accurate does it need to be? If we have the right digits going at the right speed then don't we have music? Isn't the rest about DAC and filtering (ie not the transport).

I ask these questions because I don't know the answers.



My personal view is that on some things we simply don't need to understand how/why some things work.
For example, I understand boolean logic and am an ex programmer. Can't say I understand the technicalities of a network though, despite me using one everyday. Do I care? Not really.

In the context of a stereo, if it sounds better, it IS better. Doesn't matter about the technology, the cost or the marketing BS that's bandied around.

Ref the quality of digital outputs. I have gone from using a DVD player as a transport, to a dedicated transport unit, sitting on stillpoints, with an uprated power cord and now clock modded. I can assure that these changes were not made just so that I could blow dosh on the stereo. I'd much rather have used the cash to go on holiday or buy some ****, but only IF I didn't think it made a difference. I can assure you that the cash involved would have bought a LOT of ****, or a decent holiday in Thailand.

As mentioned, I don't know why the changes have worked, but would be more than happy to demonstrate them to someone else who calls in.

So for all of those skeptics, lets not talk about just the techie BS, go have a listen, THEN judge.
 
Oct 17, 2006 at 11:18 AM Post #11 of 19
It helps to know a bit more about the CDM (CD Mechanism), and the different types out there, and the kind of faults they have when sent back to a service center to getfixed. The two you'll run into are the linear and radial tracking types. This is similar to the tone arm versions on turntables.
The linear types are good when new. Once they ather dust, the linear tracking motor starts playing up. The grease on the metal retaining bar gets sticky, etc. And jitter then becomes a bigger problem. Technics etc. do a special lubricating oil to clean up and re-oil those metal bars with. But it is an engineering job. The cheaper DVD and CD pleyers have an inferior tracking motor. Also, the CD pick-up is a cheaper one. KSS-2XX series from SONY iis very popular.

The radial types are found in Philips and Marantz units, plus anyoone who made a deck out of the Philips CDM. They suffer less from jitter in cold weather, but in the summer their drive transistors are known to fail. I bolted some heatsinks on a modified deck of mine once. SIgnal tarcking improved.

So IMHO opinion, on a digital output only requirement, get the best CDM you can afford if you are losing sleep about the extent of the jitter.
But also clean from time to time that little platter on which your CD rests in your CD player. Even minuscule dirt on there causes variations in the height of the CD with respect to the laser unit. As the disc rotates, the CD pick-up has to refocus in order to keep the dots on the dics into focus. This also adds to jitter problems, plus a few more.
 
Oct 17, 2006 at 4:00 PM Post #12 of 19
Quote:

Originally Posted by thomaspf
Very interesting and in sharp contrast to his earlier publications, where drew the line three orders of magnitude lower. From his 1992 AES paper in section 3.3 on the Audibiity of sampling jitter.



Quite a difference. Do you have a URL to the full 1999 paper?


Cheers

Thomas



Yes, it is very different and Dunn acknowledges that his model may be overcautious - he also never notes explicitly whether his model is based on empirical evidence or purely mathematical modelling. In any case he regards the paper he cites as important enough to modify his views.

His earlier measurements/models are based on a single tone (twice as easy to detect as music) and the legendary 10ps figure is based on a 24Khz pure tone at 120db above hearing threshold i.e an extraordinary test. The threshold of hearing for a 20 year old adult male with good hearing at even a modest 16khz is approximately 30db and rises steeply from that point up, at 24khz the threshold of hearing is approximately 88db (http://www.jstage.jst.go.jp/article/ast/27/1/12/_pdf) thus add 120db to that and you get 208db for a just audible level of 10ps - call me old fashioned but I dont want to listen to anything at that volume - the theshold of pain (freq dependent) is approx 120db. In any case testing at 120db+ for audio seems overkill by any standards. Now there may be applications where such exceptionally low levels of jitter are required but I doubt that CD audio is one of them.

The Dunn 1999 paper
http://www.nanophon.com/audio/1394_sampling_jitter.pdf
 
Oct 17, 2006 at 10:02 PM Post #13 of 19
Actually, in the case of a CD player, I think the thing to consider is how good it is at accurately retrieving the digital information off of the CD in real time. As anyone who has tried to accurately rip a damaged (visible or not visible) CD can tell you, the process is often time consuming because the CD/DVD player cannot read it well without many, many passes. It's quite possible that a higher-priced player would contain a larger RAM buffer and a faster reading mechanism so that it could read further ahead in order to devote more time to difficult-to-read parts of the CD.

The reality is that many CDs cannot be read without errors in real time. This is the main reason why using a CD player, no matter what the cost, is inherently inferior to using a computer as your source.
smily_headphones1.gif
 
Oct 17, 2006 at 11:06 PM Post #14 of 19
Quote:

Originally Posted by Scrith
The reality is that many CDs cannot be read without errors in real time. This is the main reason why using a CD player, no matter what the cost, is inherently inferior to using a computer as your source.
smily_headphones1.gif



Well I wont argue with that in technical terms, i.e a CD copy of a data disc must be identical at a bit level or it is a coaster but the question remains as to whether the level of read errors from a CDP is audiible in real world terms ??

Actually I have made rips (of my own CDs) that have had unrepairable errors even with EAC at nutjob settings
 
Oct 18, 2006 at 12:12 AM Post #15 of 19
Quote:

Originally Posted by Scrith
Actually, in the case of a CD player, I think the thing to consider is how good it is at accurately retrieving the digital information off of the CD in real time. As anyone who has tried to accurately rip a damaged (visible or not visible) CD can tell you, the process is often time consuming because the CD/DVD player cannot read it well without many, many passes. It's quite possible that a higher-priced player would contain a larger RAM buffer and a faster reading mechanism so that it could read further ahead in order to devote more time to difficult-to-read parts of the CD.

The reality is that many CDs cannot be read without errors in real time. This is the main reason why using a CD player, no matter what the cost, is inherently inferior to using a computer as your source.
smily_headphones1.gif



This couldn't be further from the truth. Jitter has little to do with reading the CD. It is mostly introduced in the clocking and clock transmission to the DAC. Computers soundcards have measured much worse than even mediocre CDP's. This is because of the rippled SM power feeding the soundcard's clock. When have you ever seen a decent powersupply feeding a soundcard or even just its clock? Until then a computer even when used as a transport to a DAC will have more jitter than a CDP. The question at hand : is that jitter difference audible?
 

Users who are viewing this thread

Back
Top