EYE DIAGRAM

[Joseph Jiang]

HI everyone，I Have simulated a system of 300km and I got the eye diagram as the file attached.I wander why the value of jitter is zero？

[Damian Marek]

Hello.

Can you please attach your project file and the picture as a jpg or png?

Regards

[Joseph Jiang]

hi,damian,i have upload the project file and printed pic.the problem is that i can not get the jitter value from the eye diagram.the eye diagram is so perfect and i think it is unbeliveable.thank you!

[Damian Marek]

Hi,

I am not sure how the encoding of the data at the transmitter is working. What type of format is that? I would recommend not using the 3R Regenerator yet, but first set up the BER Analyzer with the input binary and electrical sequence using forks.

Let me know and in the meantime I will see if I can get the project working.

Regards

[Joseph Jiang]

hi,damian,i have made a simple project to test the function of measursing jitter.I HAVE compare my project with the example file in the same environment,however, i found that their eye diagrams are similar but different results.in my project,it still can not show the statistics results than the example file.i do not know how it can be that?thank you!

[Joseph Jiang]

I HAVE TRIED A LONG TIME TO SOLVE THE PROBLEM AND I REALLY WANT TO YOUR HELP?THANK YOU,HAVE A GOOD DAY!

[Damian Marek]

Hi,

To create the histogram regions right click the component and open the Component Properties. Under the Histogram tab you can specify the different region areas. Attached is a modified version of your project with one region defined.

I don’t see a large difference between the eye diagrams, only the Q factor. But since the Q factors are already so large the difference might just be due to the randomness of noise and the fact that the bit sequence length is short.

Cheers

[Joseph Jiang]

hi damian,thank you for solving my problem.i have another question.i want to calculate the jitter from the eye diagram,however it is different when i select different range region.i think when i select a smaller region ,the value will be more accuracy.is it right?so expect to your reply!

[Joseph Jiang]

damian，i also want to know if the value of H.Range in the region of statisics is the jitter?if not,how can i get the value of jitter?thankyou!

[Damian Marek]

To calculate the jitter place the region as close to the eye crossing as possible (as you have done) and minimize the vertical thickness, so that it is almost a thin line.

A thicker region will include data points a small distance from the exact eye crossing and lower the accuracy. The jitter is then equal to the range of the horizontal histogram.

Regards

[Joseph Jiang]

DAMIAN,thank you for your help.As you said，i should minimize the region at the eye crossing to get the more accurate jitter value from the range of the horizontal histogram

[Joseph Jiang]

damian，is there a method minimize the vertical thickness to get a more accurater jitter value.i use MARKER tool to evaluate the vertical region reducing the vertical thickness.is there any other method to minimize the vertical thickness?thank you!

[Damian Marek]

Since you can change the vertical positions inside the component by fixing y1, y2 you are really only limited by how small of a difference you write out. Realistically though at one point the vertical thickness will not have a large impact on your results.

[Joseph Jiang]

hi damian，i’m sorry for trouble you again.when i get the jitter value from the eye diagram visulizer i found some difference and i have some question.
1、when i simulate a 100km link and i set the bit rate and bit sequence length 100Mbps and 1024bit,then the jitter get
is about about ps level.i want to know that if there is any method or theory to vertify the simulation.
2、Also i found that when i set different samples per bit OR bit rate，the jitter will be different.i wander how they effect the jitter value?
i will upload my project and the pptx file i have conclude.can you help me to understand it? thank you!

[Damian Marek]

No worries!

1) The horizontal range is actually going to demonstrate the fraction of bit period attributed to jitter. The x axis is Bit Period, so this accounts for some of your problems. You will need to multiply these numbers by the bit period. For example, your first slide would actually have a jitter of ~0.03 ps. This also explains the change in jitter as you modified the Bit rate.

2) I think you are seeing changes in the jitter as a function of sample rate because the jitter is very low, so as you increased the sample rate it increased sharply.

Regards

[Joseph Jiang]

hi damian,first thank you for your reply,but i still feel confused about what you said:
1、I understand what’s your meaning in the front half part，however,the jitter value in my first slide should be 0.00294339(H RANGE) multiply 0.01us(because the bit rate is 100Mbps) and it equals ~30ps.you can check it!when i modified the bit rate the H RANGE and bit period should be different,so the jitter should be changed. In my opinion,the jitter is relative to the time length between the samples point.for example,if the bit rate is low like 1Mbps,if we set the samples per bit 32,then the time length should be ~30ns(1us/32) and the jitter should be the same level.if we increase the bit rate to 1Gbps, the time length should be ~30ps(1ns/32)then the jitter should be that level.as a conclusion,if we increase the bit rate and decrease the samples per bit to make the time length between the samples point the same level,the jitter should be also the same level.however,i found that is not true.
2、yes,i found that the jitter increase when the sample rate increase.so is it right that the jitter value getting from the high sample rate will be more accurate.

thank you for your reply twice!

[Damian Marek]

Ok!

I was confused with my ns and ps. The jitter range is indeed 30 ps or 0.03 ns**. Just looking at your project again, I think the increased jitter as a function of bit rate is due to the dispersion in the fiber. The shorter pulses are spread more by dispersion, so the eye crossing region should be larger by that train of thought.

Regards

[Joseph Jiang]

hi damian，as you said,when the bit rate increase,the dispersion should be spread,then the jitter would be lager than those low bit rate.however,from my slide or project ,you can find that it is not the same but inverse function.I really confused about that.as far as i know,i we increase the bit rate ,the jitter should be larger.can you help me understand it ?(In my opinion,the jitter is relative to the time length between the samples point.for example,if the bit rate is low like 1Mbps,if we set the samples per bit 32,then the time length should be ~30ns(1us/32) and the jitter should be the same level.if we increase the bit rate to 1Gbps, the time length should be ~30ps(1ns/32)then the jitter should be that level.as a conclusion,if we increase the bit rate and decrease the samples per bit to make the time length between the samples point the same level,the jitter should be also the same level.however,i found that is not true.)
Also i found that the jitter increase when the sample rate increase.so is it right that the jitter value getting from the high sample rate will be more accurate.

[Damian Marek]

Hello,

I was looking at your project again and I noticed that the Histogram regions were not perfectly placed for 100 Gbps. That could be the issue.

[Joseph Jiang]

HI，DAMIAN，I’m sorry for that i cannot understand you!i think that is not my question because i do not want to simulate a 100gbp system and i just want to know how the bit rate and sample rate effect the jitter in the eye diagram!excuse me for asking you read my question again!thank you very much！

[Damian Marek]

That picture I attached is a screenshot of your project for the 100 Gbps bit rate. It appears as though the histogram is not centered well and is giving you inaccurate results for the higher bit rate, so the conclusions you have drawn from your results are also probably incorrect.

[Joseph Jiang]

hi，damian，i set the bit rate 100Gbps and a get a eye diagram which is in a mess,not the same like yours.i do not know the reason.i have a conclusion that if the histrogram is centered well,then we can think the jitter value is accurate. is it right? also is it that when we set the samples per bit higher,the jitter will be more accurate.thank you!

[Damian Marek]

Yes, I was confused again about the units! Mbps and I think at that time I had changed the sample rate to a value much higher. Just a note about the sample rate and jitter value:

I found that from looking at the high sample rate eye diagrams that a) they slightly change the position of the eye crossing and b) there is a better chance of a data point far outside the norm which can change the range measurement. Instead of the range you could try looking at the mean value and standard deviation. Using these you can find the variance of the data. (Square of the standard deviation)

Regards

[Joseph Jiang]

yes,damian,i found that if i change the sample rate,the h range and mean value and standard deviation will have litter difference.however,if the sample rate is high,the histogram will be centered well,so i think more higher sample rate,more accurate jitter! is it right?
2、i am still confused about the effect between jitter and bit rate.as we know，more higher bit rate,more higher jitter.however,when i test it in optisystem,it is not the same.for example,if i set the bit rate 100Mbps,10M and 1M,the jitter(h range:0.00299 bit period) will be almost the same when they do not multiply the bit period.so when the multiply the bit period,the jitter will have different order of magnitude.for example,1M:0.00299*1us=2.99ns ； 10M：0.00299*0.1US=0.299ns;100M:0.00299*0.01US=0.0299ns.it should not be that?

[Joseph Jiang]

HI,DAMIAN,I am really confused about it and i need your help！thank you！

[Damian Marek]

I think the main problem in this simulation is the lack of jitter! The directly modulated laser source is probably the only source of noise in this simulation and even the RIN is unchecked. The pulses are large enough and low power enough that nonlinearity or dispersion is not an issue at all.

What value of jitter are you expecting?

[Joseph Jiang]

hi,damian,i want to calculate the total jitter including Deterministic and random jitter in the system.i do not think the main problem in my simulation is the lack of jitter because the problem also exist when i use the example MEASURE JITTER ,which add jitter and noise source.May be It is the main noise source in my simulation when i use a 100km undirectional fiber.when i use a ultra-long bidirectional fiber like 1000km and bidirectional EDFA,may be they are the main noise source.thank you for your response!

[Damian Marek]

In the example Measure Jitter the jitter is independent of the bit rate as well. The jitter is added as a fraction of the bit period and the other added noise is quite low. The fact that the H. Range value changes for large sample rates is because the increased number of samples will increase the chance of data points far outside the normal deviation. This is why a simple range is not often used in statistics. If you compare the H. Std. Dev. they are much more similar for different sample rates.

Regards

[Joseph Jiang]

OK,Damian,as far as i conclude,if the noise is high enough,the jitter will be a function of bit rate,i.e.the jitter will be higher in high bit rate than low bit rate according to the theory.am i right? THANK YOU FOR YOUR HELP!You are really a expect in optisystem team!

[Damian Marek]

Generally yes, but as we found out it depends on the source of the jitter!

[Ravil]

Do you mean that you have section of 300 km or that’s the whole length of your transmission path?

[Joseph Jiang]

yes,the length of optical link is 300km and it should have jitter in reality.

[shivakumar rjgr]

hi,
i’m confused to find the jitter value from the eye digram. i.e., where the exact location of histogram so that the correct jitter value can be obtained. i’m attaching the image of eye diagram and please explain the correct position of histogram.should i consider the red one or yellow one.pls help

[shivakumar rjgr]

hi everyone,
all i need ask, is there any formula to calculate timing jitter numerically to validate the result that we got from histogram of an eye diagram.thank you

[Author]

Posts