That's a refreshingly clear, cogent, and frank discussion by Siau. He must drive the marketing types crazy. I wonder whose volume control they are using in the DAC2?
Yes, because the the loudspeakers are directional and looking at the energy time curve they don't get the junk you seem to think they do. Speech is actually quite good at a signal to noise ratio of 0 dB.
The monitor loudspeakers are audio delayed to match the outside loudspeaker coverage. Since we point loudspeakers at the audience and not across the field they really don't contribute much energy.
And what kind of stadiums are you used to?
Jitter does not show up as a time delay issue. In those cases 10 mS is quite stable enough. (Had to believe but air actually moves which is a different issue.) Jitter shows up as a form of distortion. (Also when working it is 12 nS not working can be as low as 20 nS up to about 50 nS.)
But if you want to try your threshold perhaps someone can set up some .wav files.
I think you will find 50 nS of jitter does sound bad.
People like John Siau and me are design engineers, often at VP level, depending on the company. We are NOT MARKETING, that is another type of individual, AND it is difficult to keep marketing completely happy, when we admit to our less than perfect designs.
Many here try to confuse MARKETING with DESIGN. They are completely different functions, usually done by completely different people, and we serious designers just want to do the best job possible. Marketing wants to make the best presentation possible. It is not always the same, and my marketing people are always trying to quiet me, and I, in turn, am always criticizing them if they exaggerate one of the products that I am associated with. It is an uneasy relationship, but doable.
Many here try to confuse MARKETING with DESIGN. They are completely different functions, usually done by completely different people, and we serious designers just want to do the best job possible. Marketing wants to make the best presentation possible. It is not always the same, and my marketing people are always trying to quiet me, and I, in turn, am always criticizing them if they exaggerate one of the products that I am associated with. It is an uneasy relationship, but doable.
Where have I mentioned "junk I seem to think they get"? I've no idea what you are saying.
Ok. So are you saying that with crowd noise at 130 dB, and a PA system screaming at that level, that speech is good enough that they can hear a 12 nSec jitter problem?
Well, to be honest, I'm more used to hearing the PA system from the running track around the outside of the football field. There, you can hear it all, from near speakers to far speakers. Oh, and at the last play at Shea, nosebleed, audio was really really poor.
Ok, so we're getting a little closer. The control room has the windows open, they are hearing 130 dB crowd noise, there are in room monitors delayed so that they time align to the closest speakers (well, time aligned with respect to a single plane in space, I assume it's where the personnel sit).
And they can hear the distortion caused by 12 nSec of jitter. Well, ok so we've reduced the problem from "an entire football stadium" to a room and comparison between a delayed monitor and a biggere external..
I've been following this thread, has anybody claimed that and I missed it?
Whoa, waittaminute. 10 mSec is quite stable enough for what? We were talking about 12 nSec of jitter as audible in an entire stadium. Now 6 orders of magnitude bigger is stable? Am I the only one having difficulty keeping track of what you are talking about?
I've no idea. I'm still asking questions to figure out exactly what you are saying.
John
JN
They don't hear the issue when the crowd is screaming. They notice when setting up before the game. But it is a big place and nowhere near as good a listening environment as a home.
That they picked up the clock was off and the jitter went up as a result and called to report a problem that was solved by turning the clock on, would strongly support the position that the increase in jitter was the issue.
Arrival time is not the issue or even the variation in it. It is what jitter does to the accuracy of signal reconstruction. In a normal stadium the arrival time varies with the wind.
At 192K 24 bit you have 217 nS per bit, (Actually less with overhead) why is it surprising that things get screwed up with jitter?
They don't hear the issue when the crowd is screaming. They notice when setting up before the game. But it is a big place and nowhere near as good a listening environment as a home.
That they picked up the clock was off and the jitter went up as a result and called to report a problem that was solved by turning the clock on, would strongly support the position that the increase in jitter was the issue.
Arrival time is not the issue or even the variation in it. It is what jitter does to the accuracy of signal reconstruction. In a normal stadium the arrival time varies with the wind.
At 192K 24 bit you have 217 nS per bit, (Actually less with overhead) why is it surprising that things get screwed up with jitter?
Yup, and he emphasizes one of the most maddening things about most folks systems: poor gain structure.
I can't believe my eyes what Benchmark say about negative feedback:
"Most amplifiers use a substantial amount of feedback to reduce the distortion at the output of the amplifier. A feedback circuit compares the amplifier output to the input signal and produces a difference signal that is added to the input. This feedback corrects distortion after it starts to occur. If a distortion event happens quickly, the feedback network may be too slow to correct the distortion.
"
"Most amplifiers use a substantial amount of feedback to reduce the distortion at the output of the amplifier. A feedback circuit compares the amplifier output to the input signal and produces a difference signal that is added to the input. This feedback corrects distortion after it starts to occur. If a distortion event happens quickly, the feedback network may be too slow to correct the distortion.
"
Slew induced distortion (SID) springs to mind.
If you don't take care with your feedback loop and band limiting of the input signal, then this is not an untrue statement.
However, he could have made that clearer.
All that's important in my book is that we know exactly (and we do, now in 2016) how to prevent any and all types of distortion relating to bad feedback design
If you don't take care with your feedback loop and band limiting of the input signal, then this is not an untrue statement.
However, he could have made that clearer.
All that's important in my book is that we know exactly (and we do, now in 2016) how to prevent any and all types of distortion relating to bad feedback design
Good, more info.
So, it's quiet, and they can apparently discern when the master clock is off by some kind of comparison between a delayed monitor and the speakers right outside the booth.
If one assumes that absolutely nothing else happens when the transmitters and receivers have all lost their clock.
So really, you have assumed that only jitter is affected by the systemwide loss of a master clock. You've nothing else upon which to base your assumption that they were hearing a 12 nSec jitter, as opposed to a whole other batch of system responses that could result when the master clock is lost.
That math is terribly incorrect and without merit. The concern is with the reconstructed waveform, not the bit rate. You've not shown any evidence to support your claim that they could hear a 12 nSec jitter in that application.
John
Or LIGO. I don't know if I'll get over this, the head of instrumentation engineering at LIGO just called our CTO and asked to talk to the designer of the AD797. Apparently it was a key component (in his words).
Nice one Scott.
One billion or so years ago, when those two suckers slapped into, and then devoured each other, apparently the energy released was equal to the entire energy output of all the suns in the universe during the event (source: Kip Thorne, CALTEC).
Nevertheless, coming back to this epoch, the AD797 makes an outstanding line amplifier and I am using the part exclusively for that job now.
Slew induced distortion (SID) springs to mind.
If you don't take care with your feedback loop and band limiting of the input signal, then this is not an untrue statement.
However, he could have made that clearer.
All that's important in my book is that we know exactly (and we do, now in 2016) how to prevent any and all types of distortion relating to bad feedback design
Read it carefully.
This feedback corrects distortion after it starts to occur.
JN,
The 12 nS is working it is 20-50 nS RMS when not. When the master clock fails the system uses the internal to the mixing board clock. It has more jitter than the master clock.
FYI the data carries more than just the 2 channels of audio. If you have jitter in the data you can get the wrong bit. I think you need to brush up on how digital audio is transfered.
There is no comparison between two loudspeakers. Both come off of the digital system.
Ever hear what happens if you put a CD digital output into most of the digital consoles running at 48K? No surprise you actually get drop outs.
Now what do you get at 1/(2 x 96,000 x 24) ? I get 2.17e-7. With the overhead it is less than that. The unmeasured quantity is what the peak to RMS ratio is.
Just not worth carrying this on.
The 12 nS is working it is 20-50 nS RMS when not. When the master clock fails the system uses the internal to the mixing board clock. It has more jitter than the master clock.
FYI the data carries more than just the 2 channels of audio. If you have jitter in the data you can get the wrong bit. I think you need to brush up on how digital audio is transfered.
There is no comparison between two loudspeakers. Both come off of the digital system.
Ever hear what happens if you put a CD digital output into most of the digital consoles running at 48K? No surprise you actually get drop outs.
Now what do you get at 1/(2 x 96,000 x 24) ? I get 2.17e-7. With the overhead it is less than that. The unmeasured quantity is what the peak to RMS ratio is.
Just not worth carrying this on.
Last edited:
So, you have purposely purchased systems which have no error correction? I thought nowadays it was common practice to put in a CRC or sumptin else so that single bit errors are addressed.
Ah, another data point. Trying to get the relevant information outta you is like pulling teeth.. So now, the "can hear 12 nS jitter in a stadium" has become "a digital system which can't fix a one bit error and is too sensitive to clock jitter".
So, you're saying that once a master clock is lost, one system changes it's data rate?? Or, that the systems are so bad that they cannot handle a signal if it has 50 nSec of jitter??
Why? Because we're getting closer and closer to what the actual problem was while getting farther and farther from ""12 nSec jitter can be heard in a stadium""?
John
Ed I would never question your work. Equipment for live sound has an entirely different set of processing latency issues that the rest of us don't have to deal with. Putting 44.1K into a SRC at 48K in even your cheapest sound card does not have dropouts but of course the latency does not matter.
When I read just about any manufacturers data sheets/white papers I see numbers like 50 - 100ps jitter so what is the issue?
http://www.cirrus.com/en/pubs/white...e_of_spdif_digital_interface_transceivers.pdf
Yes .
Which, there, translates to overload margin (avoid overdriving the DAC). See “inter-sample "overs" “ section.
This kind of distortion is easy to notice ( if I can hear it, everyone can hear it) when feeding an ordinary DAC from a USB source e.g. computer based digital source.
Use the volume control of the USB source (don’t worry, it’s a transparent volume control) and reduce the volume of the source by 3 to 6 dB.
Listen again with the reduced level.
If you leave the volume control there, you only sacrifice 3 to 6 dB from the max SNR attainable.
George
Last edited:
Except... no listening tests. "may have" "may produce" "potentially-audible"
And of course, no mention of the receding-in-the-rear-view-mirror claims that jitter is the most important aspect.
JN
"In a football stadium I use a master clock to synchronize all the audio digital data. I can get system wide jitter to under 12 nS. In the control room folks notice if the system unlocks from the clock. That puts jitter above 20 nS but I believe still under 50 nS."
So quit SYing on me.
There is error correction but with the large amount of data even good correction lets some errors through. Can you imaging what a single MSB error would do to the audio? No I don't recall what the exact scheme used is.
Again jitter is given in RMS error not peak.
We could now talk about what the actual issue was. A stadium is not a friendly listening environment under any conditions. One of the lack of communication issues is I suspect you think of a control room as a nice quiet studio. Think more of an operations center where directions are being given continually. Acoustic treatment...Er not exactly. An operator or two, no more like 20 folks from the producer, video, cheerleader, stats etc.
Did I mention cheerleader?
If they hear something, it is because they have good ears and the problem is real.
Now quit confusing time alignment with jitter. Bit errors are the issue, the standards are quite old and not always completely complied with.
"In a football stadium I use a master clock to synchronize all the audio digital data. I can get system wide jitter to under 12 nS. In the control room folks notice if the system unlocks from the clock. That puts jitter above 20 nS but I believe still under 50 nS."
So quit SYing on me.
There is error correction but with the large amount of data even good correction lets some errors through. Can you imaging what a single MSB error would do to the audio? No I don't recall what the exact scheme used is.
Again jitter is given in RMS error not peak.
We could now talk about what the actual issue was. A stadium is not a friendly listening environment under any conditions. One of the lack of communication issues is I suspect you think of a control room as a nice quiet studio. Think more of an operations center where directions are being given continually. Acoustic treatment...Er not exactly. An operator or two, no more like 20 folks from the producer, video, cheerleader, stats etc.
Did I mention cheerleader?
If they hear something, it is because they have good ears and the problem is real.
Now quit confusing time alignment with jitter. Bit errors are the issue, the standards are quite old and not always completely complied with.
Ed what professional quality frequency standard has 12ns jitter or is it an issue unique to giant venues?
- Status
- Not open for further replies.