IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
Anti-jitter RAM buffer in DACs, Which have one ?
Pio2001
post Aug 3 2003, 10:00
Post #1


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



In, http://www.tc.umn.edu/~erick205/Papers/pap...html#principles we can read about RAM buffers :

QUOTE
This process is referred to as timebase correction and as stated before, any quality piece of equipment will implement it.


That has been the ritual answer to jitter concerns since I'm on the Internet.

But... if "any quality piece of equipment will implement it", then some non-quality pieces of equipment don't implement it. My question is : which ones ?

The statement "quality piece of equipment" is completely subjective and doesn't give any information. For my grandfather, a "quality" CD reader will be a portable radio-cassette-CD-speakers combo for 100 $ in the supermarket. The "non-quality" one will be the Fisher-Price toy for children from 3 to 7.
For an audiophile friend, the "quality" CD player will be the 25,000 $ Wadia CD Transport+DAC, while any player under 1,500 $ is just worthless crap.

So, do a 200 $ CD Player always have a RAM buffer ? Or is it not considered a quality piece of equipment ? What about a 300 $ CD Player from 1990 ?
Go to the top of the page
+Quote Post
Patsoe
post Aug 3 2003, 19:58
Post #2





Group: Members (Donating)
Posts: 591
Joined: 11-February 03
From: UK
Member No.: 4952



Since it's the cheapest thing, I can't imagine any CD-player not having a simple FIFO-buffer IC.
Since it's not random accessible, I guess it's not properly called a RAM buffer. But it serves the same purpose, right?
Go to the top of the page
+Quote Post
F1Sushi
post Aug 3 2003, 21:13
Post #3





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



All CD players contain buffer memory between the EFM code demodulator (basically just a lookup table) and the CIRC decoder. It's this very buffer that is monitored to control the disc's rotation speed and ensure a constant linear velocity.

This linear velocity is constantly being tweaked to ensure that the buffer never overflows or is ever starved for data. Don't confuse CLV and the disc's constantly decreasing rotational speed - the rotational speed winds down to maintain CLV (with appropriate buffer-driven adjustements).

As far as jitter at the DAC stage is concerned, there isn't any, save for the jitter in the master clock. No amount of buffering will help that. Each stage from the CIRC decoder to the DAC filter is slaved and synchronized to the chipset's master clock (which has an arithmetic relationship with Fs).

When Philips (Signetics) was pumping out 1X and 4X 16-bit CD chipsets in the mid-80s, their databooks had some fairly good application info with plenty of block diagrams. I'll see if I can't dig one up at work after the weekend...

Edit: added paragraph on Fs jitter

This post has been edited by F1Sushi: Aug 3 2003, 21:45
Go to the top of the page
+Quote Post
Halcyon
post Aug 4 2003, 09:36
Post #4





Group: Members
Posts: 244
Joined: 6-November 01
Member No.: 416



QUOTE
As far as jitter at the DAC stage is concerned, there isn't any, save for the jitter in the master clock. No amount of buffering will help that. Each stage from the CIRC decoder to the DAC filter is slaved and synchronized to the chipset's master clock (which has an arithmetic relationship with Fs).


Please correct me if I'm wrong. I'm straying a little off-topic here perhaps, but eventually hopefully covering what Pio2001 asks also.

With external bitstream feed, the master clock (from the perspective of the DA circuit) can be 1) the incoming bitstream's (when using SPDIF) original master clock generator OR 2) the master clock that is near the DAC itself (creating a new reference clock for the DAC, rather than using the clock from the incoming stream).

In many (if not most) cheap receivers, the DAC is slaved to the incoming (SPDIF) bitstream clock. The jitter contained in that signal will not be rejected and will most likely become part of the DA-converted signal's noisefloor (if random and not data induced).

In some receivers (this can be generalised to some external DACs and CD players) the input bitstream to DAC is buffered / jitter attenuated (there is a trade-off between attenuation frequency and power) and signal is fed to the DAC using a new hopefully more stable reference clock with diminished jitter levels.

An example of this aproach is the Chord DAC64 external DAC:

http://www.stereophile.com/printarchives.cgi?624:6

I don't want to debate the goodness/badness of that unit. Just pointing it out as an example of a dedicated RAM buffer for jitter attenuation.

Now, back to Pio2001's original question:

Yes, even simple/cheap cd-players contain a simple FIFO buffer.

Does this mean that their output does not contain jitter or that their internal DA conversion is immune to jitter?

No it does not. Jitter in DA stage is often linked to poor clock (deviation can be measured in hundreds of nanoseconds rather than hundreds of picoseconds), inadequate power filtering (e.g. power available to different parts modulates with the input data) and some other issues that can be avoided with a little bit more careful (and usually a little bit more expensive) electronics.

Does this mean such jitter in cheaper cd players is audible (now just talking about jitter, not I/V conversion, op amps and the analog stage after DAC in general)?

Somewhat debated area. Julian Dunn (in his Jitter Theory App Notes for Audiotech) has discussed the audibility of jitter at various levels using various signals.

Does it mean that the cheap cd-player can be used as a reference quality cd transport (assuming it does the disc reading / error correction properly) when combined with a superior jitter attenuation DAC?

To my knowledge, yes.

regards,
Halcyon
Go to the top of the page
+Quote Post
2Bdecided
post Aug 4 2003, 11:27
Post #5


ReplayGain developer


Group: Developer
Posts: 5148
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



Pio,

I agree with the importance of your question: like most answers which suggest that digital audio "is basically perfect in any decent equipment", people who give these answers don't really understand the truth of it.


Let's assume the incoming audio data has loads of jitter on it, or the clock can't be recovered with a decent degree of stability. Now, the ideal answer is to dump the clock and re-clock it. But there's a problem: the new clock can't be completely independent of the original one. If it's too fast, then eventually you'll be wanting to output the next sample before it's even arrived. If it's too slow, then eventually the buffer memory will overflow.

So, it's clear that, in whatever system we have, even the re-buffered and re-clocked digital audio data MUST be synchronised, in some way and at some level, with the original, jittery data.


This means that the new clock will always be a filtered version of the original one. In a simple FIFO buffer, only the highest jitter frequencies are smoothed out. In a PLL system with a small amount of memory, lower jitter frequencies can be removed. Incidentally, the greater the jitter rejection, the longer the PLL takes to lock to the incomming signal. I've used DACs that take 2-5 seconds to lock.

A system with a large memory buffer will reject all but the lowest frequency components of any jitter (say, a few Hz at most, letting nothing higher through onto the output).


I don't know how many modern DACs use these technologies. All the sound cards I've used seem to lock almost instantly, but they're not separate DACs. The old Sony DAC I had at uni locked instantly. The nice Meridian DACs we had took maybe a second or too. An SME DAC we had took about 5 seconds.

It's an easy test though: turn the DAC and CD player on - start a CD playing, and then connect the SPDIF to the DAC - it'll give you a very rough idea of the technology inside the DAC. Though I wouldn't be 100% certain of this method! There may be methods to make a system appear to lock instantly, even though the lock is poor at first.

In one box CD players, you shouldn't need to de-jitter the signal - you should be running from a clean clock to start with, and pulling the data off disc as needed.

Cheers,
David.
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 12:53
Post #6





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Pio2001 @ Aug 3 2003, 05:00 AM)
So, do a 200 $ CD Player always have a RAM buffer ? Or is it not considered a quality piece of equipment ? What about a 300 $ CD Player from 1990 ?

Weren't we discussing jitter in a single-box $200 or $300 player from perhaps as recently as now or perhaps 10-15 years ago?

unsure.gif
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 13:44
Post #7





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Halcyon @ Aug 4 2003, 04:36 AM)
Does this mean that their output does not contain jitter or that their internal DA conversion is immune to jitter?

No it does not. Jitter in DA stage is often linked to poor clock (deviation can be measured in hundreds of nanoseconds rather than hundreds of picoseconds), inadequate power filtering (e.g. power available to different parts modulates with the input data) and some other issues that can be avoided with a little bit more careful (and usually a little bit more expensive) electronics.

Does this mean such jitter in cheaper cd players is audible (now just talking about jitter, not I/V conversion, op amps and the analog stage after DAC in general)?

Somewhat debated area. Julian Dunn (in his Jitter Theory App Notes for Audiotech) has discussed the audibility of jitter at various levels using various signals.

I think a quick bit bit of clarification on clock jitter and DAC performance might be in order. The switched cap sigma delta architecture does a nice job of providing quite a bit of immunity to master clock jitter, so it's not uncommon to see DACs over the past few years boasting 100-500psec jitter figures, assuming a decent stable clock. Some current PLLDACs boast sub-100psec jitter figures with a clock async to the data stream.

Can people hear that? If they CAN, then they would probably benefit from the likes of a product like the Auric Illuminator...

Edit: a bit of clarification...

This post has been edited by F1Sushi: Aug 4 2003, 16:03
Go to the top of the page
+Quote Post
Patsoe
post Aug 4 2003, 14:05
Post #8





Group: Members (Donating)
Posts: 591
Joined: 11-February 03
From: UK
Member No.: 4952



Those are interesting side notes, F1Sushi.

Could you provide some links? By intuition, I'd say people shouldn't hear effects of less-than-nanoseconds-jitter, thus agreeing with you. But has that ever been verified by listening tests?
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 14:25
Post #9





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Patsoe @ Aug 4 2003, 09:05 AM)
Those are interesting side notes, F1Sushi.

Could you provide some links? By intuition, I'd say people shouldn't hear effects of less-than-nanoseconds-jitter, thus agreeing with you. But has that ever been verified by listening tests?

Right off the bat, one of Analog Devices' current offerings, the AD1959, comes to mind...

http://www.analog.com/Analog_Root/productP...,AD1959,00.html

I could throw in some more, but I think it would be a great exercise for folks to scour some current sigma-delta DAC datasheets and app-notes from the likes of Analog Device, TI/Burr-Brown, Crystal, National, etc, and get a feel for the current state of the art.

I spend half my work day doing this sort of thing, so on a holiday long weekend, forgive me if I don't jump to the occasion. laugh.gif

As for supporting such a statement with a listening test, I'll humbly lay back in the weeds and let someone else pioneer such an endeavor...
Go to the top of the page
+Quote Post
Patsoe
post Aug 4 2003, 14:44
Post #10





Group: Members (Donating)
Posts: 591
Joined: 11-February 03
From: UK
Member No.: 4952



QUOTE (F1Sushi @ Aug 4 2003, 02:25 PM)
http://www.analog.com/Analog_Root/productP...,AD1959,00.html

I could throw in some more, but I think it would be a great exercise for folks to scour some current sigma-delta DAC datasheets and app-notes from the likes of Analog Device, TI/Burr-Brown, Crystal, National, etc, and get a feel for the current state of the art.
OK, I'll see. Thanks.
QUOTE
I spend half my work day doing this sort of thing, so on a holiday long weekend, forgive me if I don't jump to the occasion.  laugh.gif
You're forgiven, not forgotten... tongue.gif What kind of work are you doing (I tried your profile but it's empty like most)?
QUOTE
As for supporting such a statement with a listening test, I'll humbly lay back in the weeds and let someone else pioneer such an endeavor...
Sorry, but if you don't have any evident proof to say that it's inaudible, I really think you shouldn't make such statements. I mean, I too think that it's inaudible but as long as no persuasive reasoning can back my expectations, I'd rather follow Halcyon in saying it's a 'somewhat debated area.
Go to the top of the page
+Quote Post
Pio2001
post Aug 4 2003, 14:54
Post #11


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Thanks, all of you for the answers.
There was a paper about jitter linked from here. I'll look for it when I'm home, but I remember that it was impossible to prove that the noise created by jitter in the ns range was below the threshold of audition.
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 15:01
Post #12





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Patsoe @ Aug 4 2003, 09:44 AM)
QUOTE
As for supporting such a statement with a listening test, I'll humbly lay back in the weeds and let someone else pioneer such an endeavor...
Sorry, but if you don't have any evident proof to say that it's inaudible, I really think you shouldn't make such statements. I mean, I too think that it's inaudible but as long as no persuasive reasoning can back my expectations, I'd rather follow Halcyon in saying it's a 'somewhat debated area.

I agree wholeheartedly with your last statement, but bear in mind that I didn't actually say that it was inaudible - I instead made a joke in an effort to try and deflect any direct criticism regarding audibility...

sad.gif

I think I'm off on a technicality...

laugh.gif

BTW, to answer your question - I'm a hardware engineer with a background in audio embedded systems (DSPs, codecs), networking interfaces, data acquisition hardware for time-domain sonar, and some motor control designs for robotics (PWM controllers & encoder feedback PID speed adjustment stuff).
Go to the top of the page
+Quote Post
Pio2001
post Aug 4 2003, 21:13
Post #13


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Sorry, I completely lost the link. It was an old scientific paper about jitter, a PDF version of an AES article, I think.
There were plots of the maximum noise level intrduced by jitter, in the worst case, together with the human threshold of hearing.
I just remember that the jitter had to be insanely low in order to guarantee absolutely no audible noise in the worst possible conditions and at very loud listening levels (way less than 1 ns).
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 21:21
Post #14





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Pio2001 @ Aug 4 2003, 04:13 PM)
Sorry, I completely lost the link. It was an old scientific paper about jitter, a PDF version of an AES article, I think.
There were plots of the maximum noise level intrduced by jitter, in the worst case, together with the human threshold of hearing.
I just remember that the jitter had to be insanely low in order to guarantee absolutely no audible noise in the worst possible conditions and at very loud listening levels (way less than 1 ns).

Could this be it?

http://www.aesitalia.org/Materiale/archivi...heory%20(2).pdf
Go to the top of the page
+Quote Post
Pio2001
post Aug 4 2003, 22:02
Post #15


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



No sorry, there was no color, it was all about jitter with more discussion about the possible audibility.
Go to the top of the page
+Quote Post
F1Sushi
post Aug 4 2003, 22:13
Post #16





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Pio2001 @ Aug 4 2003, 05:02 PM)
No sorry, there was no color, it was all about jitter with more discussion about the possible audibility.

There's a blurb on page 6 titled "Audibility Considerations" that references another paper and summarizes some practical research...
Go to the top of the page
+Quote Post
Halcyon
post Aug 5 2003, 08:52
Post #17





Group: Members
Posts: 244
Joined: 6-November 01
Member No.: 416



F1Sushi,

thanks for the clarification!

More musings on audibility of jitter from me personally:

As with anything in hearing, the amount of jitter that remains audible, depends on the jitter itself.

If results in random noise, it's much less perceptible than data induced jitter that modulates with the input signal and causes signal dependent distortion rather than (pseudo)random noise.

In both cases, the average levels of jitter can remain roughly the same, but other type of jitter (data induced input signal modulated jitter) remains more audible to the human hearing.

So, as with everything in engineering, jitter avoidance becomes a matter of engineering trade-offs. Work on the part that is considered/assumed to be the most audible and reduce that. Efforts elsewhere may not be worth the engineering efforts / costs involved.

best regards,
Halcyon
Go to the top of the page
+Quote Post
F1Sushi
post Aug 6 2003, 03:23
Post #18





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (F1Sushi @ Aug 4 2003, 04:21 PM)
QUOTE (Pio2001 @ Aug 4 2003, 04:13 PM)
Sorry, I completely lost the link. It was an old scientific paper about jitter, a PDF version of an AES article, I think.
There were plots of the maximum noise level intrduced by jitter, in the worst case, together with the human threshold of hearing.
I just remember that the jitter had to be insanely low in order to guarantee absolutely no audible noise in the worst possible conditions and at very loud listening levels (way less than 1 ns).

Could this be it?

http://www.aesitalia.org/Materiale/archivi...heory%20(2).pdf

Could this be it?

http://www.scalatech.co.uk/papers/aes93.pdf

Excellent paper and great bathroom reading, even if it wasn't it...
Go to the top of the page
+Quote Post
F1Sushi
post Aug 6 2003, 04:10
Post #19





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Patsoe @ Aug 4 2003, 09:05 AM)
Those are interesting side notes, F1Sushi.

Could you provide some links? By intuition, I'd say people shouldn't hear effects of less-than-nanoseconds-jitter, thus agreeing with you. But has that ever been verified by listening tests?

Thought I'd follow up by offering some specs for a typical SPDIF recovery chip, in this case from Crystal. A current offering in SPDIF clock recovery and bitstream interface conversion for run-of-the-mill 3-wire DAC interfaces. Jitter spec on the recovered clock is 200psec typical.

http://www.cirrus.com/en/products/pro/detail/P54.html
Go to the top of the page
+Quote Post
Pio2001
post Aug 6 2003, 12:49
Post #20


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



QUOTE (F1Sushi @ Aug 6 2003, 05:23 AM)

Yes, it might well have been this. Anyway, the informations given are the same. Thank you for finding it.

"Worst case 100% DAC jitter error resulting from a 22 kHz audio signal and a 18.5 kHz jitter. Only 20 ps peak jitter is required for audibility." (page 29)

I doubt that someone can play a 22 kHz sine into his speakers at 112 db and not fry them, and even if I use a 10,000 W soundsystem, I seldom listen to full scale 22 kHz sines.
The problem is that numbers don't guarantee that jitter is inaudible in all cases.
Go to the top of the page
+Quote Post
TrNSZ
post Aug 6 2003, 13:24
Post #21





Group: Developer
Posts: 717
Joined: 25-September 01
Member No.: 20



There is a device available from most high-end audio shops to correct jitter created by clocking differences between devices. I think it's called a Monarch DIP (Digital Processor).
Go to the top of the page
+Quote Post
F1Sushi
post Aug 6 2003, 14:17
Post #22





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (TrNSZ @ Aug 6 2003, 08:24 AM)
There is a device available from most high-end audio shops to correct jitter created by clocking differences between devices.  I think it's called a Monarch DIP (Digital Processor).

Couldn't find any references to it on the web, except for an auction. How does it work? Assuming the input is SPDIF, and the output is likewise, I'm curious as to how it doesn't just create a new jittery SPDIF stream. If it's a different beast, my apologies in advance...
Go to the top of the page
+Quote Post
F1Sushi
post Aug 6 2003, 14:36
Post #23





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



Found it - actually it's from Monarchy Audio, and it's output is SPDIF.

http://www.monarchyaudio.com/DIP.htm

This looks like nothing more than a decoder/encoder, which would seem to me at best like a placebo. The reviews weren' t too revealing, either, except to mention that the unit recovers the clock info from the incoming stream and encodes back to SPDIF...

"Everything improved---clarity, mid-range and treble smoothness, spatial resolution, focus, bass extension and tightness. Transients were cleaner, quicker. There was a more natural decay of instruments in time and space---and not by a small margin. All of the improvements were so obvious I didn't have to strain to hear them."

This reminds me of similar testimonials for the Auric Illuminator. Frankly, I just hope I'm missing something here...

Edit: Corrected a bad assumption.

This post has been edited by F1Sushi: Aug 6 2003, 14:46
Go to the top of the page
+Quote Post
Patsoe
post Aug 6 2003, 15:22
Post #24





Group: Members (Donating)
Posts: 591
Joined: 11-February 03
From: UK
Member No.: 4952



Actually, from the description I'd suspect it does sample rate conversion. That is, from some sample rate to the same sample rate, but through the method of sample rate conversion.
Go to the top of the page
+Quote Post
F1Sushi
post Aug 6 2003, 15:34
Post #25





Group: Members
Posts: 158
Joined: 30-August 02
Member No.: 3236



QUOTE (Patsoe @ Aug 6 2003, 10:22 AM)
Actually, from the description I'd suspect it does sample rate conversion. That is, from some sample rate to the same sample rate, but through the method of sample rate conversion.

Doesn't look like it, according to some of their info...

"The digital signals are decoded and processed, and any jitter is suppressed through the DIP. The three major functions of the DIP are:

1) Signal Conditioning
2) Signal Boosting
3) Signal Isolation (from In to Out)"

and

"The DIP 24/96 makes most D/A converter compatible with most DVD players. It does NOT change the sampling rates."

Seems to me that the final performance of a DAC would depend on the quality of the clock recovery in it's associated decoder PLL.
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 1st October 2014 - 22:18