oudplayer wrote on Wed, 28 April 2004 13:00 |
I'm not sure if this gets you to a "spec" or not, but an increasing number of digital products are using what Lynx calls "steady clock" (I forget the other jargon terms in use), a technology that is able to correct drifting incoming clock signals. Perhaps there could be some measure of the ability to correct incoming clock sources? |
Ethan Winer wrote on Wed, 28 April 2004 10:43 |
Why must jitter be the explanation, as opposed to lower distortion? Or any of a number of more likely culprits. All of this stuff is easily measured. And a double blind test is at least as useful to determine for once and for all what really matters and what doesn't. |
sdevino wrote on Wed, 28 April 2004 07:35 |
So folks, I am asking for your inputs to help identify concerns of the user in terms of jitter. What information would you like the equiment vendors to supply? And why? |
Ethan Winer wrote on Thu, 29 April 2004 12:32 |
OP, If you've ever tweaked an electric guitar track to perfection only to later discover you had your fingers on the snare EQ, you'll agree that human hearing and perception are flawed at best. |
sdevino wrote on Wed, 28 April 2004 12:35 |
So folks, I am asking for your inputs to help identify concerns of the user in terms of jitter. What information would you like the equiment vendors to supply? And why? |
Ethan Winer wrote on Wed, 28 April 2004 16:51 |
Steve, As far as I'm concerned jitter is a complete non-issue, popularized by gear makers who want an excuse to get you to buy yet more gear. As an audio pro, and a listener, jitter is the very last thing I'm concerned about. The specs I see typically put jitter 110 dB or more below the music. Who gives a flying you-know-what about anything that far below the program material? Especially since it's also masked by the program. |
Quote: |
It amazes me when people obsess over minutiae like this, while blissfully ignoring numerous 30 dB nulls caused by standing waves in their control room. |
Ethan Winer wrote on Wed, 28 April 2004 18:43 |
Again I ask you to think about the relevance of anything that's 110 dB or more below the music. How could that possibly be audible? Why must jitter be the explanation, as opposed to lower distortion? Or any of a number of more likely culprits. |
Johnny B wrote on Thu, 29 April 2004 05:59 |
My own wild ass guess is that we will get better sound when the speeds are drasticaly bumped up and the bit/word is increased to say 32 or 64 bits. I could be very wrong, look at Sony and SACD's specs. 1 bit at a million miles a second. LOL. Maybe I got it half right cuz with Sony, the speed is there. |
Quote: |
One thing is sure, there is room for improvement, otherwise there would be no debate going on of "analog vs. digital." |
Quote: |
Science got us to the moon, but it has not yet cured the common cold nor ended the problems in the A-to-D and D-to-A debate. |
Johnny B wrote on Fri, 30 April 2004 01:46 |
Nika, I know you are a very smart man and have given this area a lot of thought, but there really are a lot of respected people in the industry who believe that analog still sounds better. |
Johnny B wrote on Thu, 29 April 2004 19:46 |
When the staunch adherents to analog agree there is no debate as to the sound quality, that's when the debate will end. |
sdevino wrote on Fri, 30 April 2004 14:11 |
Certainly high frequency jitter that is outside the loop bandwidth of the PLL will not matter. |
Nika Aldrich wrote on Fri, 30 April 2004 10:55 | ||
This is a myth. Jitter, just like audio, aliases so that variations higher than the sample rate still fold back into the legal range. Jitter at 45.1KHz will manifest itself as 1KHz jitter in a 44.1KS/s audio system. For this reason, jitter specs that only show the jitter frequencies in the 1fs range are unhelpful. Isn't a big issue here that it will be relatively easy with just about any given spec to have a clock source that does very well on spec but has drastically poorer performance in certain real-world conditions? Give me a spect that this wouldn't be the case for? With this realization, specs are nearly useless, or no? Nika |
sdevino wrote on Fri, 30 April 2004 20:11 |
Nika you missunderstood me. I am saying that HF jitter on an external clock will not make it to the ADC or DAC clock input in some applications. For instance many converter clocks are regenerated using PLL's or DDS. In these cases the external clock is used as an error or locking reference. A stable clock design would LPF the external clock such that the HF jitter is not even detected. |
Quote: |
You are correct in that HF jitter on the clock interacts with the audio but I am not so sure what exact form it would be. I am sure that it is probably much more complex than simple aliasing. |
sfdennis wrote on Mon, 03 May 2004 16:44 |
Luckily, there is a fairly direct relationship between jitter components of a clock signal and the resulting THD+N produced by an ideal A/D converter at a given sampling rate and resolution. I don?t know if this is the place to get into it, but I?ll summarize for now and write more if somebody asks. You can model jitter as the sum of two kinds of components: noise and periodic errors?that?s what a jitter spectrum tells us. If you decompose the jitter spectrum into these components, then you can automatically produce a ?worst-case? THD+N graph over the audio band. That THD+N graph would tell you how much distortion and noise a given clock device would give you at each frequency. Using such a graph, it would be fairly easy to settle Ethan and Nika?s argument. There are a number of wrinkles to iron out, and I don?t have all the details worked out. But I do think that THD+N graphs of jitter against an otherwise ideal A/D converter would be a lot more useful to recording engineers than today?s jitter specs. -Dennis |
sfdennis wrote on Mon, 03 May 2004 17:21 |
I was acutally suggesting a full audio spectrum plot of THD+N in the and not at a single nominal frequency. Though I expect that some folks will be interested (for silly reasons) in seeing THD+N all the way up to Nyquist and beyond. In any case, with a full audio spectrum THD+N, you'd see the sidebands if they were important and you wouldn't see them if they weren't. |
sfdennis wrote on Mon, 03 May 2004 18:22 |
Nika, Well, you?ve identified my major concern with the suggestion. The way I thought about it is was you would take the jitter spectrum as you suggested early in the thread, and feed it into a program that would spit out the THD+N graph. By using a program as a ?virtual converter?, you?ll be able to get around the idiosyncrasies of any real converter that might be used in a given test. After all, real converters have their own flaws and you wouldn?t want a clock?s jitter spec to be influenced by a particular choice of converter. |
Quote: |
As usual, the devil is in the details. |
Quote: |
First you?d have to take the output of a real jitter spectrum measurement such as is available from the AP setup. That?s the easy part. Then a program would have to model the spectrum as a linear combination of noise and jitter frequency components. |
Quote: |
Steve?s committee will have to agree on basis functions for modeling jitter. |
Ethan Winer wrote on Mon, 03 May 2004 19:54 |
Dennis, > Using such a graph, it would be fairly easy to settle Ethan and Nika?s argument. < Yes, which is why it amazes me that people are still arguing about this. So somebody please give me a worst case number expressed in dB below the program. Then apply Fletcher-Munson to make sure 3 KHz gets the weight it deserves, while minimizing the contribution of low frequencies. --Ethan |
Quote: |
Well really there is no noise component (I mentioned phase noise earlier, but that is so not the problem and so far below the noisefloor of discussion that we should drop the noise spec). |
Quote: |
It doesn't need to be modeled, really. It's a pretty straight ahead formula. |
Quote: |
The two issues that we run into are the range of the jitter spectrum we are going to require in the spec, and the deviance that any downline box can have on the results. Again, much of the problem is the PLL, not the clock! |
sfdennis wrote on Mon, 03 May 2004 20:51 |
Well, jitter has noise in that you can?t tell in advance what the time-error will be at a given instant |
Nika Aldrich wrote on Mon, 03 May 2004 13:21 |
Look at the jitter spectrum of any device in our industry and you'll find that the shape of the jitter spectrum is very static. |
Quote: |
But really, perhaps we should just start paying more attention to the PLLs than the clocks. It's the PLLs that really make the difference. |
Dennis Tabuena wrote on Tue, 04 May 2004 01:37 |
Not sure, Nika, I know what you mean by ?very static?. I should say that I haven?t seen many jitter spectra, but the very best among the devices I?ve seen have a fairly dense 1/f (6-20db/decade) rolloff away from their clock frequency. That would represent the random noise. By ?dense? I mean that there is a lot of color under the curve. Many jitter spectra have one and sometimes two spurs in the tails. These would represent the periodic jitter components. None of the real jitter spectra I have seen consist of a few simple lines. Is this what you mean by ?very static?? |
Zoesch wrote on Tue, 04 May 2004 02:24 |
Anyway, your appreciation of jitter is correct, the reality is that jitter is random within a specific frequency band (Or else clocks would be useless), it's the width of that frequency band and the periodicity that concerns us. Jitter in itself can only be measured as how many errored pulses there are within a specific time period, but if you were to compare two different samples of the clock signal spanning the same lenght in time, you'll realize that the errors don't happen exactly at the same pulses. |
Zoesch wrote on Mon, 03 May 2004 20:24 |
That's why measuring jitter in ppm to me is useless, were the manufacturers to provide the spectral component of that jitter and it would make a lot more sense. |
Nika Aldrich wrote on Tue, 04 May 2004 07:29 |
Are you looking at the crystal itself, or at the output of the box, on the BNC connector? |
Quote: |
Also, dense jitter spectra that has a steep roll-off/octave will not manifest itself as noise on a signal. It will manifest itself as distortion. |
Quote: |
Feed a 1KHz signal through a converter with this signal and listen/look at the results. Then do a 5KHz signal. It would very clearly not manifest itself as noise. Whereas a clock signal itself MAY be noisy, the way it manifests itself may not be. |
Zoesch wrote on Wed, 05 May 2004 01:46 |
so it's not as if the tools to measure jitter don't exist, manufacturers choose not to use them. |
Quote: |
As to how to report these results... why are we reinventing the wheel? ITU-T and ANSI have very descriptive standards for clocking, wander & jitter performance and synchronization. Failing that, the spectral power density of jitter should be sufficient for 99% of people. |
Quote: |
Quite frankly I'm not disappointed at the fact that the industry, as of yet, doesn't have a jitter spec because adopting a spec suddenly adopts an opportunity to abuse it. |
Dennis Tabuena wrote on Wed, 05 May 2004 07:05 |
Clocks and PLLs are either important or they're not. |
Nika Aldrich wrote on Wed, 05 May 2004 07:15 |
...but specs on one are relatively useless without specs on the other. The clocking of a system is exactly that - it is a comprehensive system, and cherrypicking one device from the system to measure is unto itself relatively useless information. |
Dennis Tabuena wrote on Wed, 05 May 2004 16:20 |
If you really believe that, then when do you stop? |
Quote: |
It is useful to spec both. Spec the stability of the clocks. Spec the jitter attenuation of the PLLs in their locking state as well as their lock-time. Also spec their stability in their flywheeling state (in this state, they're their own clock). |