[Glassman]

a big difference between DVD players and CD/SACD only players is that the latter operate just on single clock frequency, while the former needs to run three of them - 27MHz videoclock together with 44.1k/48k audioclocks.. usually the player's main chip runs on 27MHz crystal oscillator but also generates the audioclocks needed.. and because they cannot be easily divided down from 27MHz, techniques like PLL or DLL have to be used.. this is done inside that large scale integrated circuit with thousands of transistors switching all around on the same substrate.. following are measurements taken on one of the popular DVD chipsets by Zoran..



256x44.1 MCLK jitter spectrum:

256x48 MCLK jitter spectrum:

the whole 2k frequency span around the center frequency is heavily poluted, the spikes are just about 50dB down..




256x44.1 MCLK edge jitter:

256x48 MCLK edge jitter:

here you can see the DLL in action, there is not a single clock, but multiple clocks whose mean frequency is the required one.. the edge jitter is about 2.5ns as you can see..



these are the properities of the player's chipset itself - no clock upgrade kits can help here even though you will find modders installing them all around and poor people paying hundreds of dollars for them!
it is important to say though, that regardless of the jitter, there are no data related problems whatsoever - the whole system is still bit perfect, but the data really are not the whole story..
there is a way though how to deal with all these nastiness, a way to get rid of this buggy clock generation scheme and make absolutely brilliant transport out of even the cheapiest piece of universal player..

to be continued..

~G

 

[Ferbose]

Glassman, I have yet encountered a head-fier who knows about D/A more than you do. Please continue to educate me about such matters.

 

[cosmopragma]

I've subscribed to this thread.
Curious.

 

[Wodgy]

Great stuff. Thanks for sharing these measurements.

 

[drarthurwells]

Glassman:... it is important to say though, that regardless of the jitter, there are no data related problems whatsoever - the whole system is still bit perfect, but the data really are not the whole story..


Art: Does this mean that: Although the data is perserved, the phase or timing of the data presentation is adultered by jitter with implications for sound quality?

 

[Glassman]

the signal at the output of the DAC is modulated by all of those frequencies you can see and even by those you can't see on the spectrum plots because they lay outside the 2kHz span.. those are even higher in fact.. the modulation frequencies can be determined by the actual switching rate of the D/A converter's sigma delta modulator.. here we have 256x masterclock, but the modulator will run most likely at 64x so in order to get the frequencies you will see modulated on the output, you need to divide those on the graph four times.. there is a whole lot of these spuries that gets modulated into audible range and as I said they are even higher than what you can see here, reaching about -40dB.. you won't see much harm when you do THD analysis of the output, but it completely destroys imaging..

btw. thanks to all for your nice comments!

 

[JaZZ]

Glassman...

...thank you for posting these data! Do you have any idea what jitter level or pattern is audible/particularly harmful to sound quality? I'm also asking because my UDP-1 measures fairly bad in the jitter domain (apart from the S/N ratio which I don't care about because it's inaudible anyway) and when used as a transport nevertheless sounds «better» than the DVD 963 SA -- with its exemplary jitter behavior -- on a Bel Canto DAC2.

[size=xx-small]McCormack UDP-1: jitter measurement with CD[/size]

Do you think this is an example for really bad jitter behavior? The jitter measurements with the high-rez formats can be seen here:
http://stereophile.com/digitalsource...ck/index4.html

 

[1UP]

Which Zoran chipset was this? Does it matter / is your point generic to all DVD players - DVD players make suboptimal transports?

 

[clarke68]

Quote:

Originally Posted by JaZZ Do you think this is an example for really bad jitter behavior?

No. You've got an odd (and huge) spike down there around -3250Hz, but besides that the biggest spike is below -110dB...I'd say it's very unlikely you could hear that.


Quote:

Originally Posted by Glassman there is a way though how to deal with all these nastiness, a way to get rid of this buggy clock generation scheme and make absolutely brilliant transport out of even the cheapiest piece of universal player.. to be continued..

C'mon Glassman, the suspense is killing me...out with it!!!