Home » Guest Moderator Archives » Dan Lavry » Proper word clock implementation
| Re: Proper word clock implementation [message #53590 is a reply to message #53552 ] |
Wed, 23 March 2005 20:36   |
 danlavry
Messages: 997
Registered: April 2004 |
Gold Member |
|
|
| Logichead wrote on Wed, 23 March 2005 22:44 |
TOWARDS COMMON SPECIFICATIONS FOR DIGITAL AUDIO INTERFACE JITTER
In the for what it's worth department, google the title above and you will find a white paper on jitter. In section 3.8 it appears to say that jitter is induced by lower bandwidth cable. This would mean high bandwidth cable would be less likely to induce jitter. Isn't that the point of the Apogee digital cable? (BTW, they include measurements.)
So which manufacturer is selling the snake oil? Pot, kettle, black?
More importantly, when these threads become personal they lose their validity. It is with regret and with respect that I make the following request:
Dan - you need to stop using this forum to sell your products and yourself, or resign as moderator. It is inappropriate for you to use your position to attack your competitors and former employers. You (and any other manufacturer) have too much to gain here to be unbiased.
The rest of you - use your friggin' ears to examine what the numbers tell you. There was a time when everyone knew that it was impossible to fly.
Best regards....H
|
First, what julian is calling a “cable induced” jitter that is a mechanism that was well explained and discovered by Hawksford. The Hawksford paper is called "Is the AES/EBU standard Flawed" It involves an active digital audio signal that changes with the music such as AES (clock plus data). Not a signal that is repetitive such as in the case of WORD CLOCK.
Apogee is selling a word clock cable and calling it a low jitter cable. You are bringing up an application that has nothing to do with word clock, nor does it have to do with any periodic signal such as sync to AES black.
You are transferring results from a different application, where one cable characteristic (high frequency cutoff) interacts with a transformer lower bandwidth to cause a signal droop that will change a comparator trip point. All of that only happens when the signal is of a specific type.
Your example does not apply here at all. The way a cable, a transformer and the rest of the circuit reacts to an AES signal with data does not have anything to do with word clock.
This is in fact a good way to explain why cables don’t make jitter. There is a big difference between a component characteristic and the way a component behaves in a circuit when it is interacting with other components AND given signals.
That is why the concept of cables and jitter are miles apart.
The characteristics that belong to a cable description are constant for all applications, such as capacitance per foot, resistance and so on.
In the paper you quoted the application is different. It is about transfer of a different signal and it takes interaction of ALL 3 elements to make that jitter:
A. A changing signal
B. Rise time limitation due to bandwidth
C. DC blocking high pass
A given cable that may be a problem for transfer of digital audio to a DA can be great for word clock.
Cables do not make jitter. Resistors don’t make jitter. Capacitors do not either. A combination of parts in a given application can cause jitter, and changing resistor values may alter the jitter, but that does not make a specific resistor value a low jitter resistor. The same exact resistor may be the worst value in another application.
I would not mind an cable specification of bandwidth, capacitance, inductance attenuation per length, frequency… All of those are cable characteristics.
Jitter is not a cable characteristic.
Say that Julian were to use an AC coupling made of a blocking Cap value of 0.1uF. Say a change to 1uF yielded better results. Does it make a 1uF cap a low jitter cap?
A sales guy may claim with a VERY LONG STRETCH that a 1uF cap is low jitter for this circuit. But the improvement would not be there when you go for a different application. That’s is why you do not call a 1uF capacitor a low jitter capacitor. No one will call a 1uF capacitor a low jitter capacitor just because it improved jitter in one specific circuit. So no one should call a cable a low jitter cable just because it helped achieve lower jitter in one circuit.
Don’t forget to add to it the fact that we are talking about a different circuit, a word clock link not an AES data link, which is the link discussed in Hawksford and Julian’s papers.
Dan Lavry
www.lavryenginering.com
|
|
|
| Re: Proper word clock implementation [message #53612 is a reply to message #53244 ] |
Wed, 23 March 2005 22:46 |
Logichead
Messages: 12
Registered: March 2005 |
Member |
|
|
Example 2:
A MARKETTING COMPANY ?X? sells cable that they claim are: ULTRA LOW JITTER CABLE and TEMPERATURE COMPENSATED CABLE. I do not know what temperature compensated cable means but I KNOW that THE CONCEPT OF ?ULTRA LOW JITTER CABLE? is not true. Cables don?t have jitter . A cable is a PASSIVE device. It is just a ?piece of material?. The level of jitter is determined by the electrical signal driving the cable! There may be other factors (variable load, induced noise and more). One thing for sure: but there is no such a thing as ultra low jitter cable.
?Jitter to cable? is like a ?laptop to Moses?
?Jitter to cable? is like a ?Pistachio nut to a whale?
?Ultra Low jitter cable? is like ?super high speed tree?
?Ultra Low jitter cable? is like ?butter to a headphone jack?
Please know that in future that statements about personal opinion and marketing will be deleted unless backed up with technical input.
Dan Lavry
Lavry Engineering, Inc.
[/quote]
There does indeed appear to be "cable induced jitter" as noted in the white paper. You seem to agree, in contrast to your previous statements, seen above. To quote the paper, "this model shows how cable induced jitter (is) strongly dependant on the bandwidth of the link." The example quoted started with this statement:
"If a jitter-free interface signal is transmitted down a cable then the data pattern in the signal will modulate the signal zero-crossing timings. This modulation is such that patterns of zeroes produce more delay to the transition timings than patterns of ones, so as the data varies the timings modulate in sympathy."
And later on reached this conclusion:
"If we consider a short chain of digital audio devices, where each device is locked to the previous one, we have several contributions to the jitter at the end of the chain. Each device will add its own intrinsic jitter, and each interconnecting cable will make some contribution with cable-induced pattern-dependent jitter."
Notice also, in the above statements, you did not limit your statements to word clock cables only. (Yes, the thread is labeled "proper word clock implementation" but the subject has ranged far and wide since the thread's begining.)
It remains inappropriate for you - a manufacturer - to moderate a thread which villifies, either directly or indirectly, other manufacturers. Your bias is showing.
Best regards....H
|
|
|
| Re: Proper word clock implementation [message #53650 is a reply to message #53612 ] |
Thu, 24 March 2005 04:43 |
Nika Aldrich
Messages: 832
Registered: April 2004
Location: In transition |
Gold Member |
|
|
Is this thread really about semantics? Cables don't directly induce jitter - the jitter comes from EMI, RFI, other types of noise. Good cables merely reject that jitter better. In this capacity, bad cables are said to "induce" jitter because it is accepted that some degree of noise is a part of the environment, so cables that allow this noise to affect the signal are said to "induce" that jitter. This seems like pretty basic stuff to me?
As far as the filtering of cables, I assume it has been discussed that cables act as filters and can thus cause modulation of the square wave pulses and can therefore create jitter that was otherwise not present. Lowering the cutoff frequency of the filter can increase the modulation. The less square the waveform the greater likelihood that the cable's inherent filtering properties exacerbate the jitter present from noise sources as well. For these reasons as well, good quality cabling helps reduce jitter manifested through the process.
I think that the filtering properties of a cable do indeed "induce" jitter, but even if we reject this for some reason and say that cables don't "induce" jitter, but they can be said to "reject" jitter, and if you accept that that jitter from noise is an inherent part of the environment one could certainly decide to describe the use of poor cables as "inducing the manifestation of this problem." It seems petty to try to pick apart these semantics.
As for Apogee's "low jitter cable" (or "ULTRA low jitter cable" or whatever) I think the implication is that it has high jitter rejection, has better inherent filtering properties, and thus exudes lower amounts of jitter?
Something seems petty about debating the semantics at play here, especially when we all know what was meant, and when AES whitepapers have been written on the significance of using good quality cabling so as to provide lower jitter to digital devices.
Nika
"Digital Audio Explained" now available on sale.
Click above for sample chapter, table of contents, and more.
|
|
|
| Re: Proper word clock implementation [message #53665 is a reply to message #22983 ] |
Thu, 24 March 2005 07:04 |
bananahill
Messages: 20
Registered: January 2005
Location: Los Angeles |
Member |
|
|
Dan,
I have to agree with Logichead. You are out of line here.
You have every right to express your opinions and findings and argue these points.
The conflict is your role as moderator. From the tone of many of the posts, no one is moderating this thread. You are a very involved participant.
I think you should remove yourself as moderator and continue to post.
If you want to remain moderator, you need to do this in an objective and unbiased manner.
Stephen Krause
|
|
|
| Re: Proper word clock implementation [message #53670 is a reply to message #53650 ] |
Thu, 24 March 2005 07:16 |
Logichead
Messages: 12
Registered: March 2005 |
Member |
|
|
| Nika Aldrich wrote on Thu, 24 March 2005 10:43 |
Is this thread really about semantics? Cables don't directly induce jitter - the jitter comes from EMI, RFI, other types of noise. Good cables merely reject that jitter better. In this capacity, bad cables are said to "induce" jitter because it is accepted that some degree of noise is a part of the environment, so cables that allow this noise to affect the signal are said to "induce" that jitter. This seems like pretty basic stuff to me?
As far as the filtering of cables, I assume it has been discussed that cables act as filters and can thus cause modulation of the square wave pulses and can therefore create jitter that was otherwise not present. Lowering the cutoff frequency of the filter can increase the modulation. The less square the waveform the greater likelihood that the cable's inherent filtering properties exacerbate the jitter present from noise sources as well. For these reasons as well, good quality cabling helps reduce jitter manifested through the process.
I think that the filtering properties of a cable do indeed "induce" jitter, but even if we reject this for some reason and say that cables don't "induce" jitter, but they can be said to "reject" jitter, and if you accept that that jitter from noise is an inherent part of the environment one could certainly decide to describe the use of poor cables as "inducing the manifestation of this problem." It seems petty to try to pick apart these semantics.
As for Apogee's "low jitter cable" (or "ULTRA low jitter cable" or whatever) I think the implication is that it has high jitter rejection, has better inherent filtering properties, and thus exudes lower amounts of jitter?
Something seems petty about debating the semantics at play here, especially when we all know what was meant, and when AES whitepapers have been written on the significance of using good quality cabling so as to provide lower jitter to digital devices.
Nika
|
No doubt there are semantics involved here, as is often the case on a forum. That's not what I find important.
First, if you read that white paper (the pdf is included a few posts up) it appears that low bandwidth cable actually adds jitter to jitter-free digital audio. I am only interpreting what I read, but I'm pretty sure I have it right. This was news to me, as I suspect it is to you.
Next, Dan has been using this forum inappropriately, just read his posts if you are uncertain. He has directly singled out Apogee, his former employer and a current competitor for his most severe criticism. Fine, criticise away, but not as the moderator.
Dan, again - you are damaging your reputation as well as Pro Sound Web's. Imagine your response if it were someone at Apogee moderating this thread and saying these things about you.
IMO Apogee should contact the operators of the forum and complain. If only to save Dan from himself, this entire thread should be deleted.
Best regards....H
|
|
|
| Re: Proper word clock implementation [message #53689 is a reply to message #53670 ] |
Thu, 24 March 2005 08:10 |
Nika Aldrich
Messages: 832
Registered: April 2004
Location: In transition |
Gold Member |
|
|
| Logichead wrote on Thu, 24 March 2005 08:16 |
First, if you read that white paper (the pdf is included a few posts up) it appears that low bandwidth cable actually adds jitter to jitter-free digital audio. I am only interpreting what I read, but I'm pretty sure I have it right. This was news to me, as I suspect it is to you.
|
If you're talking about the ability for the filtering characteristic of cables to provide a modulation effect on the square wave data run through the cable this isn't exactly new stuff - it's pretty basic waveform theory. The question has to do with the practical results of that - what type of bandwidth is necessary in order to "induce" this jitter on a bad cable, especially when dealing with 44.1kS/s pulses? I suppose I could run a test in Matlab and see.
So I concede Dan's point in advance of actually knowing - that even in this pathological case, jitter isn't induced for some reason. Still, there can be little denying that some cables reject jitter better than others.
Nika
"Digital Audio Explained" now available on sale.
Click above for sample chapter, table of contents, and more.
|
|
| |
| Re: Proper word clock implementation [message #53699 is a reply to message #53689 ] |
Thu, 24 March 2005 08:37 |
danlavry
Messages: 997
Registered: April 2004 |
Gold Member |
|
|
I thought more about what I said last night, and I just want to be more specific than I was.
I find the term “cable induced jitter” to be a poor choice of words. Julian did the math well, and he performed his experiments and collected data. I guess he was focused on the math derivation, and on getting the massage across that AES signals are impacted by a cable.
I would say that the jitter in an AES link is impacted by the cable, impacted by the transformer or capacitor. Impacted, not induced. The word induced suggests that the cable does something actively. But the cable does not act. It reacts. There is no capacitor induced jitter either for the same reason – the capacitor is passive. As a passive device, a cable reacts differently to different input signals and the reaction depends on interaction with other components.
The original discoverer of the effect of jitter on a digital audio link was Hawksford. He explained the mechanics very well. The signal has droop due to low frequency coupling. The droop is related to the time duration of each logic state. With data changing on an AES link, the droop time duration changes also which causes a “wobble”. The waveform voltage goes up and down tracking the digital logic states. Given that the signal transitions between states has a finite rise time (it is not vertical) the time where a comparator reference (threshold) is crossed by the wobbly signal also wobbles. That is the jitter we are taking about. Data link jitter.
I find the term cable induced jitter to be off the mark. The cable does not induce jitter. That is not the way I heard Hawksford explanation to be, nor is it my understanding of how the mechanism is. The cable does not do anything by itself. It does not induce anything it is passive. It “just sits there”. Just like other passive parts (resistors, caps, diodes, empty circuit boards, metal chassis and more). Is a metal chassis a low jitter chassis? Supose it that help shield a circuit from interference that cause jitter in one application, and it increases it in another case. The chassis has attributes (length, width, material data…) but you can not say it is an low jitter chassis. It is just a passive piece of metal. It is passive material and it does not “make waves” by itself. Just like a cable.
I assume there was no harm intended by Julian, just a poor choice of a word. But someone is using that for commercial promotion of a product. And we are not taking rocket science here, it is just a piece of passive copper with some insulation. Why does someone home in on the one “off phrase” in an article? Is it lack of depth and technical competence? Readiness to just say anything? A misinterpretation of a poor choice of words.
Does someone that says that cables make jitter and sell a low jitter cable be required to explain how cables make jitter? Just pointing at “so and so said this or that” does not make it so. Where is the mechanism for cables to induce anything? It is not there. If you believe it and sell it, measure it. Guess what- whatever jitter numbers you measure in an AES data link will change with different transformers and even with sample rate. Cables are reactive. They are put into a structure and become a part of an interaction.
A piece of cable on the ground, connected to nothing does not “make signals”, and jitter without a signal makes no sense. A cable has characteristics. There are mechanical (weight length…) and electrical (capacitance, resistance…) so you can say “low capacitance cable” or 10 foot cable. Those characteristics go with the cable wherever it is in all applications.
It is improper to take a passive device into a specific application with specific conditions and assign to it attributes that belong to that specific case. The number 2 can mean 2 billion dollars or 2 cents. Shell we call the number 2 very significant? It is just there passively. It is not inducing money. Like a cable, the number 2 is “just there” until you give it some use, some context. It is passive, like a cable. The word is passive.
And suppose that I am dead wrong (I am not). Then the cable induced jitter explanation is a reason to stay away from external clocking in favor of internal clocking. External clock box always requires cables. If Apogee thought that a cable makes jitter they should encourage the use of internal clock, not an external Big Ben.
Regards
Dan Lavry
www.lavryengineering.com
|
|
|
| Re: Proper word clock implementation [message #53705 is a reply to message #53689 ] |
Thu, 24 March 2005 08:52 |
Logichead
Messages: 12
Registered: March 2005 |
Member |
|
|
| Nika Aldrich wrote on Thu, 24 March 2005 14:10 |
| Logichead wrote on Thu, 24 March 2005 08:16 |
First, if you read that white paper (the pdf is included a few posts up) it appears that low bandwidth cable actually adds jitter to jitter-free digital audio. I am only interpreting what I read, but I'm pretty sure I have it right. This was news to me, as I suspect it is to you.
|
If you're talking about the ability for the filtering characteristic of cables to provide a modulation effect on the square wave data run through the cable this isn't exactly new stuff - it's pretty basic waveform theory. The question has to do with the practical results of that - what type of bandwidth is necessary in order to "induce" this jitter on a bad cable, especially when dealing with 44.1kS/s pulses? I suppose I could run a test in Matlab and see.
So I concede Dan's point in advance of actually knowing - that even in this pathological case, jitter isn't induced for some reason. Still, there can be little denying that some cables reject jitter better than others.
Nika
|
Thanks for your reply Nikka. I also appreciate the reasoned tone of your comments. The practical implications of cable induced jitter appear to be small. There are indeed limits on what the numbers tell us! While you appear to say there is essentially no such thing (in practical terms) as cable induce jitter, the white paper seems to say otherwise. It is an infinitismal point we are dealing with here, no need for histrionics. I can see both conclusions.
As you say, "there can be little denying that some cables reject jitter better than others." That is all Apogee appears to claim, at least on their website.
I don't want to read too much into it, but your silence on Dan's actions as participant and moderator leaves me wondering how you feel about a manufacturer using a forum as Dan has.
Best regards....H
|
|
|
| Re: Proper word clock implementation [message #53707 is a reply to message #53705 ] |
Thu, 24 March 2005 09:01 |
Nika Aldrich
Messages: 832
Registered: April 2004
Location: In transition |
Gold Member |
|
|
Eh gad, that was the longest, most vindictive and most nuanced apology I've seen in a long time.
Nika
"Digital Audio Explained" now available on sale.
Click above for sample chapter, table of contents, and more.
|
|
| | | | | |
| Re: Proper word clock implementation [message #53791 is a reply to message #22983 ] |
Thu, 24 March 2005 17:05 |
 |
Eliott James
Messages: 283
Registered: May 2004
Location: Atlanta |
Active Member |
|
|
Oil and water. "Ear" people and "Tech" people. The two seldom meet in understanding, although both are equally right and wrong on a consistant basis about the same topics.
Eliott James
http://www.EliottJames.com
|
|
|
| Pages (14): [ 13 ] |
  |
Goto Forum:
Current Time: Fri Dec 5 10:22:06 CST 2008
Total time taken to generate the page: 0.02884 seconds |
Members
Search
FAQ
Register
Login
Home
