Bit of a Doh! moment for me. Jcx is correct in his analysis. The trick is to remember that the range of DACs he is lumping under "multi-bit" are all non-return-to-zero. That is the crucial determinant. Multi-bit is actually not the right term - you can have return-to-zero multi-bit DACs - although they are typically not employed in audio.
Now the question becomes one of looking at the internal architecture of these DACs. The question is one of whether the clock event that terminates one sample output period is the same one that starts the next sample period. If it is, jcx's analysis remains valid. If it isn't, we return to the problem at hand. This question becomes hard to answer as we have less and less information about the exact instantiation of any DAC's design. This sounds a bit silly, since surely it is the same event, but internally it may not be, and components of the DAC may be switched at different times, in readyness for the output switch. The settling time, and other second order effects may come into play here. Any asymetry in the transition between sample is a conduit for jitter that isn't subject to jcx's analysis. This is always going to be a hard to charaterise.
For any return-to-zero DAC the problem is clear, and that includes all pulse density modulation dacs. The issue very likely also besets most with switched capacitor filters.
Now the question becomes one of looking at the internal architecture of these DACs. The question is one of whether the clock event that terminates one sample output period is the same one that starts the next sample period. If it is, jcx's analysis remains valid. If it isn't, we return to the problem at hand. This question becomes hard to answer as we have less and less information about the exact instantiation of any DAC's design. This sounds a bit silly, since surely it is the same event, but internally it may not be, and components of the DAC may be switched at different times, in readyness for the output switch. The settling time, and other second order effects may come into play here. Any asymetry in the transition between sample is a conduit for jitter that isn't subject to jcx's analysis. This is always going to be a hard to charaterise.
For any return-to-zero DAC the problem is clear, and that includes all pulse density modulation dacs. The issue very likely also besets most with switched capacitor filters.
If I house the DAC chip inside the same box the transport resides, I can tap into the system's master crystal oscillator (or replace it with a high quality one if necessary) instead of having to use a PLL. Thus avoiding jitter issues (I realize that a free running high quality xtal oscillator isn't completely jitter free, but it will be a lot better than a good PLL).
Or if you must have the DAC in its own box, I'd buffer the transport's master oscillator and feed that on its own coax cable to the DAC circuits. No PLL, just re-buffer it and use it directly. Or a simple divide down using flip-flops. This would have to be a custom box though, not a universal add on.
Or if you must have the DAC in its own box, I'd buffer the transport's master oscillator and feed that on its own coax cable to the DAC circuits. No PLL, just re-buffer it and use it directly. Or a simple divide down using flip-flops. This would have to be a custom box though, not a universal add on.
Well, let's assume that his analysis is right. We all know that the DAC makers have lied to us in the past.
We then have to come up with a plausible explanation why the effects of correlated jitter are much more apparent at higher o/s rates. Or put another way: you can get by with a lousy (e.g., the one inside a '7220) easier with lower o/s.
I'll let someone with a more theoretical background.........jcx....Bruno......Guido T (clock monger! can't be trusted).......to rack their brain to come up with an explanation.
Only one that comes to mind might be the images are pushed further away with higher o/s rates. Implies that lower o/s masks jitter effects with all the crud generated by reconstruction, etc.
Jocko
We then have to come up with a plausible explanation why the effects of correlated jitter are much more apparent at higher o/s rates. Or put another way: you can get by with a lousy (e.g., the one inside a '7220) easier with lower o/s.
I'll let someone with a more theoretical background.........jcx....Bruno......Guido T (clock monger! can't be trusted).......to rack their brain to come up with an explanation.
Only one that comes to mind might be the images are pushed further away with higher o/s rates. Implies that lower o/s masks jitter effects with all the crud generated by reconstruction, etc.
Jocko
Well let's put this simple: oversampling also multiplies jitter. 😀
Not really the dac's fault, it's just the deam digital filter.
Not really the dac's fault, it's just the deam digital filter.
Jocko Homo said:
Hi
As the clock rate lowers, the on-chip noise coupling is reduced as well. This in general enhances sound quality. The removal of the digital filter in turn reduces the amount of jittery signals to the DAC as well.
wa2ise said:
Depends how good the PLL is, but you are right. Avoiding the SPDIF takes away the jitter problems it induces.
Then you are left with a free running oscillator, which can be made realy good. But you don't want to know how many manufacturers aren't able to build or buy decent oscillators. I am talking CD players over 10k
Worst I saw was a one box player taking SPDIF from drive and putting that into a CS8414 to recover the clock (again) 
best
Guido Tent said:
Worst I saw was a one box player taking SPDIF from drive and putting that into a CS8414 to recover the clock (again)
Must have been a product from "the sleeping lion"...
Arne K
Guido Tent said:Worst I saw was a one box player taking SPDIF from drive and putting that into a CS8414 to recover the clock (again)
Nothing shocks me anymore.
Guido Tent said:
Depends how good the PLL is, but you are right. Avoiding the SPDIF takes away the jitter problems it induces.
Then you are left with a free running oscillator, which can be made realy good. But you don't want to know how many manufacturers aren't able to build or buy decent oscillators. I am talking CD players over 10k
best
Hi,
Do you mean this one?
http://www.unisonresearch.com/unico/unicocd.asp
Maybe they struck a deal with Crystal 😎
Kind regards,
Ashley.
Lots of s**t out there
Quote:
"Digital signal processing is carried out first by a CRYSTAL CS8414 chip which provides clock-jitter reduction and correction and control of the audio data."
There must be a school of thought there.
Another fine example: the Audio Analogue Primo CD.
http://www.audioanalogue.com/eng/primocd.php
Samsung CD-ROM.
The internal dac of the CDROM is used, analog output is buffered.
SPDIF direct from the CDROM's digital out.
ash_dac said:
Quote:
"Digital signal processing is carried out first by a CRYSTAL CS8414 chip which provides clock-jitter reduction and correction and control of the audio data."
There must be a school of thought there.
Another fine example: the Audio Analogue Primo CD.
http://www.audioanalogue.com/eng/primocd.php
Samsung CD-ROM.
The internal dac of the CDROM is used, analog output is buffered.
SPDIF direct from the CDROM's digital out.
carlosfm said:
Quote:
"Digital signal processing is carried out first by a CRYSTAL CS8414 chip which provides clock-jitter reduction and correction and control of the audio data."
There must be a school of thought there.
Another fine example: the Audio Analogue Primo CD.
http://www.audioanalogue.com/eng/primocd.php
Samsung CD-ROM.
The internal dac of the CDROM is used, analog output is buffered.
SPDIF direct from the CDROM's digital out.
Hi,
Hey if you can't get a CS8414, go and use a CS8412 instead if your feeling lazy :-
http://www.soundresearchlabs.com/product.asp-itemid-102-catid-33
http://www.adcom.com/specifications.htm#/main/data/specifications/itemgcd_750.xml
Kind regards,
Ashley.
ash_dac said:
Funny business, audio. Many manufacturers claiming the best performance of the world and designing the lousiest equipment one can think of.......
arggggggh
Guido Tent said:
Indeed...
I don't know if audio performance is of much importance for these guys, because the marketing dep. deals with the rest.
It must be really chronical stupidity to use extra parts and circuitry and then mess up the whole thing.
Where's the logic behind this? Who designs these nonsense?
But I've seen too many things, and nothing shocks me anymore.
ash_dac said:
Here is a review of the Unico (one with the cs8412):-
http://www.audioenz.co.nz/2004/unico_cd.shtml
"The chassis and lid is made from heavy duty steel giving the player rigidity and freedom from a rattling lid."
A strange review that made me chuckle.
😀
Kind regards,
Ashley.
ash_dac said:
One has to read 'between the lines'.
For me this tells everything:
"Any sonic compromises seem to be cannily chosen as the music flowed with an ease and naturalness that made every disc listenable, even poorly recorded stuff like The Corrs".
No need to say anything more.
You don't get it, do you??
How can you possibly find fault when the "design has provided the right proportion between the front faceplate’s size and the chassis’ depth, which allows for different placements without affecting the famous quality of the audio analogue Sound"?
C'mon, what else could you ask for??????
Somehow, I don't get it either. Unless it is the length of the cable coming out of the back of the CD-ROM that is somehow magical.
But I suspect not.
How do these guys manage to find people who are dumb enough to buy it, and I have trouble giving some of my stuff away? I need better BS to spread.
Jocko
How can you possibly find fault when the "design has provided the right proportion between the front faceplate’s size and the chassis’ depth, which allows for different placements without affecting the famous quality of the audio analogue Sound"?
C'mon, what else could you ask for??????
Somehow, I don't get it either. Unless it is the length of the cable coming out of the back of the CD-ROM that is somehow magical.
But I suspect not.
How do these guys manage to find people who are dumb enough to buy it, and I have trouble giving some of my stuff away? I need better BS to spread.
Jocko
carlosfm said:
It must be really chronical stupidity to use extra parts and circuitry and then mess up the whole thing.
Where's the logic behind this? Who designs these nonsense?
But I've seen too many things, and nothing shocks me anymore.
I've done lots of system level engineering (which is essentially the sensable arrangement of various function blocks inside a product, be it a chip, set top box or here a CD player). You'd think a lot of it would be obvious, but it's suprising how many people will do goofy things like the above.
Ideally the industry standard for CD transports and separate DAC boxes should have had the master oscillator in the DAc box (so the DAC gets a jitter free clock) and the transport is slaved to that master oscillator. The transport, being all digital "pixels" won't much care if the clock has some jitter, as the DAC would have a buffer to clean it up.
- Status
- Not open for further replies.