The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
Mar 28th, 2024, 9:32am
Pages: 1 2 3 4 
Send Topic Print
convert noise to jitter (Read 974 times)
youchen
Junior Member
**
Offline



Posts: 22

Re: convert noise to jitter
Reply #30 - Nov 12th, 2008, 10:58am
 
Hi Frank. I notice that the transient analysis gives different timing delays over time for the same driven logic circuit. However, since transient analysis does not consider device noise (no other noise is injected either), why the timing delay is not constant over many cycles? I understand the initial transient effect may cause that, but I am supposing that the delay should approach constant as time goes on. I do not seem to see that.
Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #31 - Nov 12th, 2008, 11:37pm
 
If you are not using Transient Noise, the reason is probably limited numerical accuracy of the simulation. Have you tried setting Accuracy Defaults to conservative in the tran setup form?
Back to top
 
 
View Profile WWW   IP Logged
hiSkill_11
Community Member
***
Offline



Posts: 35

Re: convert noise to jitter
Reply #32 - Apr 24th, 2012, 12:47am
 
Hi, Frank, I am doing sampled pxf for measuring the tf from power supply to jitter( 40MHz crystal oscillator), but I can't setup the pxf analysis form because I don't know what should be typed into the probe field though I can select the output clk net into the voltage field. I don't known exact meaning of the probe signal and it's purpose as well. Could you help me on that?  Also I want to know the meaning and the effect of the maximum samples and sample delay on the result. Thanks in advance.

Frank Wiedmann wrote on Nov 1st, 2008, 2:19pm:
You always have to do a pss analysis first for the periodic small-signal analyses pac, pxf, and pnoise because the pss analysis determines the periodic operating point for these analyses.

If you want to simulate jitter, use sampled pxf with an absolute frequency sweep. As mentioned in http://www.designers-guide.org/Forum/YaBB.pl?num=1187679312/1#1, you convert the result to jitter by dividing it by the slope of the signal, in the same way as for tdnoise. As you correctly noted, the result is a transfer function, so there is no need (and no possibility) to specify an input spectrum.

Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #33 - Apr 24th, 2012, 2:50am
 
Back to top
 
 
View Profile WWW   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #34 - Aug 16th, 2015, 8:19am
 
Hi Frank,

I have read many threads from you about how to simulate the random jitter from device using PSS+Pnoise. Unfortunately, I didn't got consistent results from tdnoise and pnoise jitter.

I would like to describe my question here. The circuit I am simulating now consists of one input 16GHz buffer, divider, 4GHz buffer, CML MUX, 4GHz buffer, and one CML phase interpolator.

I set the noise type to "jitter". In the tdnoise, I could get the integrated output noise N. From the PSS, I could get the dv/dt at both rising edge and the falling edge. By using the equation in chapter 9, I could get the Jee ~200fs.

However, if I use the pnoise jitter directly, the integrated jee I got is around 400fs.

Do you have idea what causes this discrepancy?

You help is highly appreciated.

Regards,
Julia
Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #35 - Aug 17th, 2015, 4:20am
 
If you specified the timepoint for the tdnoise analysis correctly, the reason for the discrepancy can only be an error in your formulas or a misunderstanding about how Jee is calculated. The tdnoise and pmjitter analyses are practically identical, the only difference is how the timepoints for the analysis are specified. When you plot the results from the results browser, you should get identical results from the pnoise_td and the pnoise_pmjitter folders.
Back to top
 
 
View Profile WWW   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #36 - Aug 17th, 2015, 8:31am
 
Hi Frank,

I just know that the jee contains DSB noise, is it possible that the integrated output noise contains only SSB noise?
Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #37 - Aug 17th, 2015, 8:48am
 
The pnoise timedomain and pmjitter analyses simulate sampled noise. I am not sure if talking about DSB and SSB noise is very meaningful in this context.
Back to top
 
 
View Profile WWW   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #38 - Aug 17th, 2015, 9:17am
 
Hi Frank,

I am sorry I dont know how to attach three pics in one thread...

I set the noise type to "jitter", and then plot the integrated output noise shown below

Back to top
 

output_noise.png
View Profile   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #39 - Aug 17th, 2015, 9:27am
 
[/img]

from the pss.png, I got dv/dt=48.35G for rising, 48.77G for falling edge

so the jitter=2.34mV/48.35G=48.4 fs for rising,
and =2.2mV/48.77G=45.1 fs for falling edge.
Back to top
 

PSS.png
View Profile   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #40 - Aug 17th, 2015, 9:30am
 
However, the jee I got is twice of what I got from the calculation.

I am kind of desperate here...really dont know where the problem is...

BTW, I am simulating a 16GHz buffer, and I integrate from 1Hz to 8GHz.

Frank, your help is really really appreciated.

Thanks,
Julia

Back to top
 

jee.png
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #41 - Aug 18th, 2015, 1:13am
 
There are rather few points in your pss results, which might make the calculation of the derivative inaccurate. Try to increase the number of points by increasing the value of the maxacfreq parameter (in the pss form press the Options button and select the Accuracy tab).
Back to top
 
 
View Profile WWW   IP Logged
sweet_julia
New Member
*
Offline



Posts: 6

Re: convert noise to jitter
Reply #42 - Aug 18th, 2015, 10:59am
 
Hi Frank,

I can not thank you more for your kind help. Finally I got the problem solved and got the consistent results!

I like analog IC very much despite I could be frustrated by it sometimes LOL.

Frank, thank you;-)

Best Regards,
Julia
Back to top
 
 
View Profile   IP Logged
foxbeibei
New Member
*
Offline



Posts: 1

Re: convert noise to jitter
Reply #43 - Dec 27th, 2015, 10:10pm
 
Hi Frank,

for the following case, the 1mV is the amplitude of the sinewave or is the peak-to-peak values? i tried to use the sampled pxf method you suggested but found that the jitter values calculated this way is half of the transient aboslute jitter. can you please answer this question?

thanks,
foxbeibei

youchen wrote on Nov 2nd, 2008, 11:00am:
Hi Frank, to make sure I do it correctly, can I formulate it as follows? After 'sampled pxf' analysis, I got two tranfer function plot with respect to the Vdd and Vss. So, suppose I know that the power noise or substrate noise is characterized by a sinwave of 1mV and 1MHz, then I find the resulting noise at the output by mulitiplying 1mv with the gain at 1MHz. Finally, I divide the resulting noise by the slope, which gives me the jitter. Is that right? If it is, I was wondering what if the power or substrate noise is not well approximated as a sinwave, but with a certain spectrum, for example 100k to 1MHz, because in reality the power noise is very complicated? Thanks very very much.

Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: convert noise to jitter
Reply #44 - Jan 18th, 2016, 4:28am
 
If you are looking at peak-to-peak jitter, the 1mV should be the peak-to-peak value of the sinewave.
Back to top
 
 
View Profile WWW   IP Logged
Pages: 1 2 3 4 
Send Topic Print
Copyright 2002-2024 Designer’s Guide Consulting, Inc. Designer’s Guide® is a registered trademark of Designer’s Guide Consulting, Inc. All rights reserved. Send comments or questions to editor@designers-guide.org. Consider submitting a paper or model.