[faberryman]

I found this post on another site and found it interesting:


"Except that the current implementations of HDMI have much higher jitter figures than SPDIF. In the Feb 2009 edition of the Hi-fi News magazine Paul Miller measured the following jitter results for a few A/V amplifiers:


Denon AVR-3803A

---------------

SPDIF: 560psec

HDMI: 3700psec


Onkyo TX-NR906

---------------

SPDIF: 470psec

HDMI: 3860psec


Pioneer SC-LX81

---------------

SPDIF: 37psec

HDMI: 50psec


Yamaha RX-V3900

---------------

SPDIF: 183psec

HDMI: 7660psec


And no, those figures for the Pioneer aren't typos - it just shows what can be done when implemented correctly! Unfortunately the Denon and Onkyo amps were rated to sound the best, but I believe that had more to do with their amplifier sections than the jitter and pre-amplifier/processor sections."


I haven't had a chance to read the full article yet, but some of those jitter measurements are astonishingly high. The Denon unit is surprisingly bad.


Perhaps Bob Stuart is right about HDMI and jitter?

 

[antiacid]

are we talking about picoseconds here? You kind of have to understand wtf you are talking about before putting on your "surprised, shocked and annoyed" face.


Here's a link to wikipedia for those who don't know their units:

http://en.wikipedia.org/wiki/Picosecond


Basically, the highest quoted figure of 7660psec (A picosecond is one millionth of one millionth, or one trillionth of a second (0.000 000 000 001 seconds) amounts to 7.7ns or 0.000077ms (If I counted my zeros properly...)


So the true question is: Does the human body hear/see this difference?


I'll guess no.

 

[amirm]



Quote:
Originally Posted by antiacid /forum/post/16156309


are we talking about picoseconds here? You kind of have to understand wtf you are talking about before putting on your "surprised, shocked and annoyed" face.

I hate to say this but he is quite right in being alarmed. Your post is the classic pitfall of using layman logic to analyze something that is quite complex mathematically in comparison. See below.

Quote:
Here's a link to wikipedia for those who don't know their units:

http://en.wikipedia.org/wiki/Picosecond


Basically, the highest quoted figure of 7660psec (A picosecond is one millionth of one millionth, or one trillionth of a second (0.000 000 000 001 seconds) amounts to 7.7ns or 0.000077ms (If I counted my zeros properly...)


So the true question is: Does the human body hear/see this difference?


I'll guess no.

The right reference would be an authoritative paper on the topic of jitter such as this one presented at AES: http://www.nanophon.com/audio/jitter92.pdf


Using the formula in that paper, and assuming the jitter spectrum is simpler than it really is (i.e. NOT the worst case scenario), we can compute what level of jitter we need to obliterate the low order bit in a 16-bit audio sample:


"For sinusoidal jitter of amplitude J=500ps, a 20 KHz maximum level tone will produce sidebands at -96.1 dB relative to the input tone."


Just expanding, a 16-bit audio sample has 16*6 = 96 dB of dynamic range. Meaning the signal will be 96db higher than noise floor. If the difference between the signal and noise floor decreases, then you have a corresponding reduction in sample resolution (i.e. drops below 16-bits).


According to the math then, a 500 ps peak-to-peak (250ps relative to zero) is the maximum jitter that can be tolerated if we want to preserve the full 16-bit value. As otherwise, the sidebands (distortion) created by the jitter modulating the signal rise aboe -96db relative to a full magnitude source.


As you go higher in sample resolution, the jitter must be even lower.


So assuming you are playing CDs and not higher resolution audio formats, then none of the units tested, sans Pioneer have low enough jitter on their HDMI input to fully resolve the input samples.


And again, keep in mind that the math is only easy to analyze if we assume a simple, sine wave spectrum. Jitter is usually far more complex and can do more harm than even the above math shows (although the reverse is also sometimes true in that some jitter is harder to hear than others).

 

[Chu Gai]

A 20 KHz maximum level tone would be quite the rarity for CD, don't you think Amir? Is your concern/interest more along the lines of not losing or obscuring anything or does it speak to the issue of audibility with respect to musical sources?

 

[faberryman]



Quote:
Originally Posted by antiacid /forum/post/16156309


are we talking about picoseconds here? You kind of have to understand wtf you are talking about before putting on your "surprised, shocked and annoyed" face.


Here's a link to wikipedia for those who don't know their units:

http://en.wikipedia.org/wiki/Picosecond

I know what a picosecond is, thank you very much. I am also familiar with jitter measurement for high-end audio and video transports and the HDMI measurements for the units (except the Pioneer) are astonishingly high. Just compare them to the corresponding S/PDIF jitter measurements. Now which digital interface would you prefer to listen to with your high end system?

 

[John Kotches]



Quote:
Originally Posted by antiacid /forum/post/16156309


are we talking about picoseconds here? You kind of have to understand wtf you are talking about before putting on your "surprised, shocked and annoyed" face.


Here's a link to wikipedia for those who don't know their units:

http://en.wikipedia.org/wiki/Picosecond


Basically, the highest quoted figure of 7660psec (A picosecond is one millionth of one millionth, or one trillionth of a second (0.000 000 000 001 seconds) amounts to 7.7ns or 0.000077ms (If I counted my zeros properly...)


So the true question is: Does the human body hear/see this difference?


I'll guess no.

There are a lot of studies out there, some say several hundred nanoseconds, some say several hundred picoseconds and they're double blinded. The jitter spectra undoubtedly plays into this in the end.


Yes, some of us know powers of 10.

 

[Chu Gai]

Those would be with tones and headphones vs speakers, no John?

 

[Mikenificent1]



Quote:
Originally Posted by antiacid /forum/post/16156309


are we talking about picoseconds here? You kind of have to understand wtf you are talking about

How ironic that you're the one asking this question!

 

[amirm]



Quote:
Originally Posted by Chu Gai /forum/post/16157067


A 20 KHz maximum level tone would be quite the rarity for CD, don't you think Amir?

That is correct. I simply quoted the numbers as stated in the article. Re-computing for 5 KHz, results in 4X the jitter (1000 ps relative to zero).


Of course, I can play the game in reverse and ask whether people here are happy with 16 bits or demand 24-bits

. If so, the numbers get crazy in a hurry. If my math is right, for 5 KHz, you need to have jitter no higher than 4 picoseconds!

Quote:
Is your concern/interest more along the lines of not losing or obscuring anything or does it speak to the issue of audibility with respect to musical sources?

My goal above was to simply show that picoseconds is what we are talking about, no matter how small they might sound.


I have post at length in another thread recently on the audibility issues. So I won't rehash any of that here. Suffice it to say that the spectrum of jitter matters as the ear has non-linear response to levels depending on frequency. And that correlated jitter is worse than uncorrelated. In addition, at very low frequencies, masking effects help as the sidebands are too close to the input signal to be heard.

 

[John Kotches]

Amir:


Wouldn't the results of uncorrelated (random) jitter begin to approach that of another randomized function, aka dither?

 

[amirm]



Quote:
Originally Posted by John Kotches /forum/post/16157528


Amir:


Wouldn't the results of uncorrelated (random) jitter begin to approach that of another randomized function, aka dither?

It would as long as its level doesn't increase beyond the LSB. If it does, you still get the benefit of dither but start to lose resolution.


Note that uncorrelated and random are not the same thing. If jitter for example is at 1KHz percisely, all the time, it is uncorrelated but not random. Its effect though, may be easier ignored than one that is synchronized to the input signal.

 

[blathaller]

Antacid


Please get your facts straight before using expression like "wtf" since it does not look like you know, dare I say, wtf, what you're talking about. I apologize to the people in the forum . Pointing a mistake and there was none, should not be so uncivil.

 

[Chu Gai]

IIRC, Amir, Dunn was of the general opinion that far too much was being made over jitter, however I would expect that were I to pay substantial amounts for a piece of gear, I'd expect commensurate performance regardless of whether it was inaudbile. BTW, if you have that link regarding your participation in the jitter topic, can you post it?

 

[fastl]

On page 71 of the March issue of hifinews, they show SACD over HDMI jitter (Sony SCD-XA5400ES) as being 8000 psec. Mentioned in the review is a Sony feedback control system (H.A.T.S) that is claimed to reduce this to 200 psec. The jitter spectrum shown is primarily low frequency (below 200 Hz), so its audible effect is probably minimal. Same issue shows the Onkyo PR-SC886 as exhibiting 430/3920 psec jitter, SPDIF/HDMI. If Pioneer can get the jitter down to don't care levels, makes you wonder why the others can't.

 

[antiacid]



Quote:
Originally Posted by blathaller /forum/post/16157743


Antacid


Please get your facts straight before using expression like "wtf" since it does not look like you know, dare I say, wtf, what you're talking about. I apologize to the people in the forum . Pointing a mistake and there was none, should not be so uncivil.

1- you're missing an "i"


2- from my point of view, those jitter figures are still mostly irrelevant except in special cases, as amirm tried to explain (which just so happens to be so out there that I would consider them theoretical examples)


3- I wasn't pointing a mistake, I was pointing my finger at what I think is, at this time, a certain type of snake oil / advertising piece towards pioneer units. If the jitter isn't actually an issue for most listening purposes, then what does it matter showing the figures in a publication?


In the same line of thought, let's put on our voodoo hats and demand 24bits+ music because it sounds so much better.


I don't disagree with the theory, I disagree with the interpretation made by wishful thinkers

Now if I sounded rude to begin with, then I apologize because I only meant to show my exasperation towards yet another sales pitch based on the unimportant details instead of the big picture.


Seriously, there's about 12 different features that would make you decide on an AVR before it comes down to the jitter measurement. I might go as far as to dare someone with the equipment handy to make a double-blind test of his gear connected w/ hdmi vs spdif. Last I checked, the method has usually been "make a guess, test it rigorously, analyze the results then revisit the hypothesis". Until that has been thoroughly done, I just can't take this issue seriously

 

[amirm]



Quote:
Originally Posted by antiacid /forum/post/16162719


Seriously, there's about 12 different features that would make you decide on an AVR before it comes down to the jitter measurement. I might go as far as to dare someone with the equipment handy to make a double-blind test of his gear connected w/ hdmi vs spdif. Last I checked, the method has usually been "make a guess, test it rigorously, analyze the results then revisit the hypothesis". Until that has been thoroughly done, I just can't take this issue seriously

This would be a fine position to take anywhere but this forum. Please look up on the heading for this subforum. It says > $20K gear. By your logic, no product should be discussed here

.


People in this forum are looking for perfection for that kind of money. To the extent there is a product with much better spec on jitter, that would likely indicate better attention to detail in other areas just the same even if you didn't think jitter mattered.


On issue of audibility, that was not the point of this thread. Until now, no one had bothered to see if there were objective/engineering differences between S/PDIF and HDMI. We had a long thread recently where people post with strong conviction that since HDMI had buffering, it could not have any jitter, let alone numbers this big. Some of us argued that jitter from high-speed circuits leaks into rest of the receiver but folks didn't want to listen. Now we have real data to show that jitter does indeed exist.


Now, it is fair to discuss whether jitter can be heard or not. But I suggest the place for that is another thread. Or continue the one we discussed where I post papers and such on tests conducted where jitter was heard (although at levels higher than what is discussed here).


I think the above is the reason you reveived the heat that you did

. There must be a peaceful place where we discuss the last, last, tidbit of quality, especially if we can measure it. If it isn't here, not sure where it would be. There are plenty of other places to discuss how many connectors there are on the back of an AVR

.

 

[faberryman]



Quote:
Originally Posted by antiacid /forum/post/16162719


Until that has been thoroughly done, I just can't take this issue seriously

That's the great thing about this pursuit; you don't have to take it seriously if you don't want to. I do take it seriously and would hope that you would let me do so without derision. After all, not having done double-blind tests yourself, you are just guessing that those levels of jitter are not audible. Were you to conduct such tests, perhaps jitter of that magnitude might not be audible on your system, in which case your lack of concern would be justified for you. On the other hand...

 

[Chu Gai]



Quote:
Originally Posted by amirm /forum/post/16162839



There must be a peaceful place where we discuss the last, last, tidbit of quality, especially if we can measure it. If it isn't here, not sure where it would be. There are plenty of other places to discuss how many connectors there are on the back of an AVR

.

That's fair enough Amir. Ought one though not consider other aspects which can be measured such as power supply isolation, which can be seen in the jitter measurements, as well as other factors? Just because one has done a stellar job at reducing jitter, doesn't mean they've paid equal attention elsewhere. For example, the PS Audio Digital Link was measured at AVTech and found to have exceptionally low jitter (

 

[DougWinsor]

Remember that jitter its self is not audible but jitter can effect THD ratings of a DAC.


Wolfson states that their DAC's can handle "under 1 ns is recommended, although the part can take higher jitter without much impact on the THD performance."


AK states "the AK4397 and most AKM DAC's can theoretically meet their SNR and THD+N specs with more than 10ns of jitter."

 

[amirm]



Quote:
Originally Posted by Chu Gai /forum/post/16163470


That's fair enough Amir. Ought one though not consider other aspects which can be measured such as power supply isolation, which can be seen in the jitter measurements, as well as other factors? Just because one has done a stellar job at reducing jitter, doesn't mean they've paid equal attention elsewhere.

I am assuming that Hifi News opened each unit and sampled the jitter at the input of the DAC. In that case, they are measuring an entire chain which includes excellent decoupling from the power supply induced noise/distortion.


So while that merit doesn't travel past the DAC, I would say that it takes a lot of attention to cricuit design to get such a low jitter rating.


Having said this, I am not recommending any specific unit based on this data alone. People should still listen to equipment and look at the rest of the components in the system not measured above.

Quote:
For example, the PS Audio Digital Link was measured at AVTech and found to have exceptionally low jitter (