Mode Convergence Problem


#1

Hello,

I am trying to get the transmission through a waveguide for the fundamental TE and TM modes. For one of the modes if I sweep the number of modes vis the mode convergence sweep there is a point where the transmission drops for including a particular number of modes and then picks back up with more modes.

Can someone explain the cause of this? It appears the results are more stable for using a smaller number of modes?

Here is the mode convergence sweep (This is the S42 value, and I launch the fundamental TE and TM modes into port one as the sources)

This is the simulation file if helpful
2018_11_03_1550_Bridge_Coupling_BT_4000_WT_800_Swp_BW_WW.lms (355.0 KB)

Thanks
John


#2

Hi @john.jost

I reduced the yz span of simulation file and obtained the plot below for abs(S42)^2:

image

The area shown by red elliptical is the safe operating place for the number of modes. Sometime by increasing the number of modes you may see unwanted peaks, for example between 45-50 modes, which can be ignored.

You also need to perform convergence testing, such as using finer mesh, to make sure that results convergence:
https://kb.lumerical.com/en/index.html?layout_analysis_test_convergence_eme.html

Hope this was helpful.


#3

Hi,

Thank you very much for taking the time to run my simulation yourself. That is very nice of you.

Yes I agree with you that region looks stable for the number of modes. However, I have found that if I change some of the dimensions of the waveguide or the region that for some parameters it is fine and then other it is not.

Thus for each configuration it appears I need to find the right numbers of modes. This makes sweeping in script over the one of the relevant parameters (like waveguide width) hard.

I had noticed that if I switch the energy conservation option from passive to energy conserving that I get less random peaks and dips in my results and things are bit more stable with respect to the number of modes I chose.

Do you have any intuition if changing the energy conservation settings should have a significant effect?

Thanks for all of your great help. It is appreciated.
John


#4

Hi @john.jost

I think for the mode convergece testing, you need to make sure that simulation is set properly. This includes simulation span, mesh override regions, and boundary conditions. I expect the mode convergence should work in most of the cases. If you still had doubts in the results, you can check them in FDTD.

Regarding energy conservation, this option assumes that there is no loss in the system like material loss and scattering and the energy is normalzied and distributed among the available modes in each cell. So, this will not be a good option if there is loss in the system or when the number of modes is low. We should be able to reach convergence with passive option, and I will be glad to take a look at your simulation file if you still have any problem.