Hi, Anyone know how to measure jitter by using digitizer and DSP algorithms. Any doumentation, books which I can reference ? Thanks. Regards KC kccheng@LinuxDAQ-Labs.org
jitter measurement methods
Started by ●August 9, 2005
Reply by ●August 9, 20052005-08-09
KC wrote:> Hi, > > Anyone know how to measure jitter by using digitizer > and DSP algorithms. Any doumentation, books which > I can reference ? Thanks.Jitter of what? Why not an oscilloscope? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●August 9, 20052005-08-09
"KC" <kcc1967@gmail.com> wrote in message news:1123614208.664839.161730@g49g2000cwa.googlegroups.com...> Hi, > > Anyone know how to measure jitter by using digitizer > and DSP algorithms. Any doumentation, books which > I can reference ? Thanks. >Hello KC, One item to study is Allan variance" Here is a description by David Allan: http://www.allanstime.com/AllanVariance/ Clay
Reply by ●August 9, 20052005-08-09
"Jerry Avins" <jya@ieee.org> wrote in message news:uridnZ2dnZ1F3CS1nZ2dna6cZN-dnZ2dRVn-z52dnZ0@rcn.net...> KC wrote: >> Hi, >> >> Anyone know how to measure jitter by using digitizer >> and DSP algorithms. Any doumentation, books which >> I can reference ? Thanks. > > Jitter of what? Why not an oscilloscope? > > JerryHello Jerry, I recall using a 'scope to roughly measure phase noise on a clock. You can set the trigger to normal and set the horizontal span to include 10 to 50 cycles. Then just look the width of the trace relative to the period. It usually gets fatter as you move away from the trigger point. And by the time you get over to the right hand side, it may have maximized out. This can be roughly converted to a phase noise std. dev. spec. One can certainly bound the phase deviation this way. I later wrote some code to take the received sinusoid and find the best fit sinusoid and calculate the stats on the phase errors (ideal minus measured). Clay
Reply by ●August 9, 20052005-08-09
Hi Jerry,>> Anyone know how to measure jitter by using digitizer >> and DSP algorithms. Any doumentation, books which >> I can reference ? Thanks. > >Jitter of what? Why not an oscilloscope?For example, jitter of clock signal. And why not just use scope ... because I want to know how instead of just get the result. Thanks KC
Reply by ●August 9, 20052005-08-09
"Clay S. Turner" <Physics@Bellsouth.net> wrote in message news:_48Ke.6888$op.3275@bignews4.bellsouth.net...> > "KC" <kcc1967@gmail.com> wrote in message > news:1123614208.664839.161730@g49g2000cwa.googlegroups.com... >> Hi, >> >> Anyone know how to measure jitter by using digitizer >> and DSP algorithms. Any doumentation, books which >> I can reference ? Thanks. >> > > Hello KC, > > One item to study is Allan variance" > > Here is a description by David Allan: > > http://www.allanstime.com/AllanVariance/ > >Whatever you do you need either a good clean stable comparison clock or a stable delay. Best of Luck - Mike
Reply by ●August 9, 20052005-08-09
in article _48Ke.6888$op.3275@bignews4.bellsouth.net, Clay S. Turner at Physics@Bellsouth.net wrote on 08/09/2005 16:12:> "KC" <kcc1967@gmail.com> wrote in message > news:1123614208.664839.161730@g49g2000cwa.googlegroups.com... >> Hi, >> >> Anyone know how to measure jitter by using digitizer >> and DSP algorithms. Any doumentation, books which >> I can reference ? Thanks. >> > > Hello KC, > > One item to study is Allan variance" > > Here is a description by David Allan: > > http://www.allanstime.com/AllanVariance/ya know, i remember seeing an AES convention paper presented (maybe Bob Adams, but i can't say fur sure) where the nature of the jitter can be mapped to an additive noise model of some sort. i thought that if somehow you could avoid both conditions of the jitter being so small that you can't measure it against the other normal quantization noise of the A/D and D/A, and that the jitter ain't so large that it changes the p.d.f. of the so-called "sinusoidal p.d.f." where uniform p.d.f. is runned through a sinusoidal function, you might detect in the measured p.d.f. (the estimate of frequency of occurrence of different noise amplitudes), you might detect among the expected p.d.f. of the output error a sum of that sorta random variable and a sinusoidal random variable. sinusoidal r.v. p.d.f.: p(alpha) = (1/pi)/sqrt(A^2 - alpha^2) drive the system with a good (in terms of its own clock) sinusoid af amplitude, A, and make some kinda estimate of the component with that p.d.f. and maybe one can argue that its source is jitter. this is just an incomplete recollection where i am filling in holes from memory (and trying to make sense of it). no warranty implied. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
Reply by ●August 10, 20052005-08-10
KC wrote:> Hi Jerry, > > >>>Anyone know how to measure jitter by using digitizer >>>and DSP algorithms. Any doumentation, books which >>>I can reference ? Thanks. >> >>Jitter of what? Why not an oscilloscope? > > > For example, jitter of clock signal. And why not just use > scope ... because I want to know how instead of just get > the result.If you can just get the result with an oscilloscope, then you know how to get the result. What's the question? Another way: use a crystal oscillator to generate a stable sinusoid at the clock frequency, then use the clock to trigger an ADC that catches the sinewave near the zero crossing. Jitter will appear as an amplitude variation in the result. This is a DSP analog of phase contrast microscopy. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●August 10, 20052005-08-10
Hi,>> For example, jitter of clock signal. And why not just use >> scope ... because I want to know how instead of just get >> the result. > >If you can just get the result with an oscilloscope, then you know how >to get the result. What's the question?I did more search on Google ... so let me re-ask again my question. I found something like measure jitter using BERT scan, and I want to find what's BERT scan is ... and how to do it if I only have waveform capture from digitizer. I also found some methods such as ET (Equivalent Time), RT (Real-time) and TIA (Time interval analysis) methods ... any doc or books describes these methods and algorithms ? Thanks KC>Another way: use a crystal oscillator to generate a stable sinusoid at >the clock frequency, then use the clock to trigger an ADC that catches >the sinewave near the zero crossing. Jitter will appear as an amplitude >variation in the result. This is a DSP analog of phase contrast microscopy. > >Jerry
Reply by ●August 10, 20052005-08-10
KC wrote:> Hi, > > >>>For example, jitter of clock signal. And why not just use >>>scope ... because I want to know how instead of just get >>>the result. >> >>If you can just get the result with an oscilloscope, then you know how >>to get the result. What's the question? > > > I did more search on Google ... so let me re-ask again my question. > I found something like measure jitter using BERT scan, and I want > to find what's BERT scan is ... and how to do it if I only have > waveform capture from digitizer. > > I also found some methods such as ET (Equivalent Time), RT (Real-time) and > TIA (Time interval analysis) methods ... any doc or books describes these > methods and algorithms ?http://grouper.ieee.org/groups/802/3/ae/comments/d3.2/asanders_1_0901.pdf Any measurement derived from digitized data will be the sum of the jitters in the signal and the digitized clock. Unless the digitizing clock is much faster than the signal being measured, providing at least one sample on most transitions, even a rough estimate requires luck. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
