- This topic has 20 replies, 6 voices, and was last updated 8 years, 7 months ago by MHD NA.
-
AuthorPosts
-
-
February 28, 2016 at 4:23 am #30421Tung Tran VanParticipant
After optimizing pump power and frequencies of Raman Amp for gain flatness, output OSNRs are very low. After several EDFA amplifiers, system performance is so bad.
How can i edit optimization to improve the DWDM system performance?
I attached my project (layout2)
Thanks -
February 29, 2016 at 12:19 am #30435alistuParticipant
Hi Tung,
A simple change you can make in order to get better results in your system would be to increase the input laser power level if it is possible for you in the WDM transmitter component. I increased the power to 7dB and I got Q-factor of more than 6. Please try it and see if it is helpful.
Regards
-
February 29, 2016 at 1:26 am #30446FAYIQA NAQSHBANDISpectator
Hello Tung,
I simulated your system design at i was able to get Quality factor up around 6. As Alistu mentioned above , you can increase laser input power.
I also tried to lessen number of ports at the Demux side. Why are you floating so many ports when they are unused. Make the number of ports as required.
Please check the osd file i have attached Herein. Thanks! -
February 29, 2016 at 12:44 pm #30511Tung Tran VanParticipant
Thanks all.
@Fayiqa Naqshbandi: :(. I’m using OptiSys v7, so I can’t open your osd file.
But when increasing the input laser power, Optisys will caculate too slowly because of nonlinear effects. 🙁-
February 29, 2016 at 1:16 pm #30518alistuParticipant
You are right Tung. The calculations take more time when the input power is increased due to nonlinear effects. However, when I increased the power to 7dB, the calculations were do in less than 40 seconds. But if the power exceeds some high value, then the calculations may take several hours to finish.
-
-
February 29, 2016 at 1:17 pm #30519Aabid BabaParticipant
Hello Tung,
I suggest you to increase the laser power even though it takes some time. If the simulation time stays below 1 minute then its fine.
I tell you there are simulations that take more than 25 minutes but its fine as long as its optisystem 7.regards
-
February 29, 2016 at 1:30 pm #30524alistuParticipant
Hi Aabid,
The simulation duration is very relative. I have designed a system whose back-to-back calculations take about three minutes. To be more specific, it very much depends on the number of components used, which increase with the increase in the number of users in an access system for example. So I believe the only thing that should be considered is the results being corrupted or not.
Regards
-
-
February 29, 2016 at 1:39 pm #30526Aabid BabaParticipant
Hello Alistu,
Well may be but i have seen other softwares like CST , CAD feiko take days for simulation for just one filter and if you compare optisystem with those it is lightening fast. So as if you ask me i would be happy if my simulation takes even 20 minutes.
And as far as optisystem is concerned,, sweeping many parameters at same time takes more time rather than number of components.Regards
-
February 29, 2016 at 1:52 pm #30530alistuParticipant
Hi Aabid,
It is not a matter of OptiSystem being fast or slow, but the point is “fast” and “slow” are relative terms. Iteration also makes the system calculation duration change and I agree, so I think we shouldn’t set any time for the duration of acceptable or unacceptable results, It would not be correct to say that if simulations take less than a minute then it is OK.
Regards
-
-
February 29, 2016 at 1:47 pm #30528aasif bashir darParticipant
hi all,
i agree with aabid baba
our co – workers working onthe software like cad feko, cst ,etc; click on the simulation button before sleeping. and at the moring they find simulation 10 percent complete.itz kind of joke.
i appreciate that simulation using optiwve is much quick.
with regards
-
February 29, 2016 at 2:30 pm #30536alistuParticipant
In our discussion, there is no talk of other simulation softwares. Anyway, different softwares are used for different purposes. OptiSystem is used to implement systems, while some other softwares may just be used to calculate and solve electromagnetic equations. So it would not be OK to compare all of them with each other in the same place.
-
-
February 29, 2016 at 2:38 pm #30540Aabid BabaParticipant
Hi Alistu,
I am not saying that if simulation time is below one minute then the throughput will be fine. It was just a reply to Tung where he mentioned that increasing power takes him long time. I ran his simulation and it took me 47 seconds to simulate it and as you and fayiqa had mentioned Quality factor was almost 7 or so.
That was the reason i told him not to worry.
Regards-
March 1, 2016 at 6:18 pm #30778alistuParticipant
Yes, and I agree with you that the simulation time is OK in this implementation. However, my contention is a specific calculation time should not be set in general to evaluate performance of different systems so as to determine whether the nonlinear effects have reached the destructive threshold or not.
Regards
-
-
March 1, 2016 at 1:16 pm #30764Tung Tran VanParticipant
Hi all.
Because my project is so complicated, I have deleted some components in it to look better. At some nodes with Raman Amp (node3, node 4), the received signal quality is so bad, Q factor is low. Plz, give me a suggestion to improve system performance.
I attach my project with wavelength assignment schemes.
Thank you so much.-
March 1, 2016 at 6:15 pm #30777alistuParticipant
Hi Tung,
I believe using dispersion compensating fiber would help you get better results. I have noticed now you are reading data from the file, and if the same data file is used, no compensation is done. The use of dispersion compensating fiber is very common for systems with rather long distances.
Regards
-
March 1, 2016 at 9:25 pm #30787Tung Tran VanParticipant
Thanks alistu
I used DCFs in every nodes, they are between PAs & OAs. I set dispersion.dat for SMFs (G652) & DispersionDCFNew.dat for DCFs (named DCMs). So I think dispersion is compensated. Moreover, the bad received signal only at the nodes with Raman Amps.
-
-
-
March 18, 2016 at 1:37 pm #33147Tung Tran VanParticipant
Hi all,
in theory, an optical Raman amplifiers (ORA) increases the OSNR of input signals, but my designed ORA reduces OSNR wdm channels greatly.
what should I do to improve my ORA?
Thanks all-
March 18, 2016 at 1:56 pm #33156FAYIQA NAQSHBANDISpectator
HELLO Tung Tran Van…
well i think it has to do with the signal iterations…i suggest u to increase the signal iterations from 1 to two or three in the global layout window and then when you again check for the OSNR , increase the signal index and see where your OSNR is better..For changing this signal index , i suggest you to follow the link below.. I hope it is helpful to u..
optiwave.com/resources/applications-resources/lesson-5-bidirectional-simulation-working-with-multiple-iterations/
Thanks and regards-
March 20, 2016 at 12:16 pm #33332Tung Tran VanParticipant
Thanks FAYIQA NAQSHBANDI
as I know, a iteration parameter is often used in the bidirectional or ring network design . If my system is the point – point transmission , it can improve ouput OSNR of the ORA ?
-
-
-
March 20, 2016 at 1:24 pm #33341MHD NAParticipant
please,increase the power to 7dB and I got Q-factor of more than 6.
-
March 20, 2016 at 1:27 pm #33342MHD NAParticipant
i think it has to do with the signal iterations…i suggest u to increase the signal iterations from 1 to two or three in the global layout window and then when you again check for the OSNR , increase the signal index and see where your OSNR is better..For changing this signal index.
-
-
AuthorPosts
- You must be logged in to reply to this topic.