Learning Forum

[Amritayan Chatterjee]

Dear Team,

I wanted to simulate bend loss in FDE. But I am getting different loss for different FDE x-span and y-span. Sometimes I am getting negative loss value. How to fix the FDE span for bend loss calculation in FDE module?

Waiting for your answer.

Sincerely,

Amritayan Chatterjee

[Amrita Pati]

Hi Amritayan,

The negative losses could be because the waveguide is too close to the PML boundaries. Since the mode in a bent waveguide radiates out of the bend, I will recommend you to increase the FDE span in that direction. Ensure that the center of the bent waveguide is outside the entire simulation region (simulation + boundary). You can perform covergence testing on the span of the waveguide to determine the required x and y span. The negative losses could also be due to inaccurate material fitting. I would also suggest you to check the Material Explorer and make sure that the imaginary part of the refractive index is not negative, which could introduce gain. If the losses are still negative try changing the PML settings in the Advanced Options Tab of the solver to the following:

• PML kappa: 2
• PML sigma: 30
• PML layers: 24
• PML polynomial: 3

Regards,
Amrita

[Amritayan Chatterjee]

Hi Amrita,

Thanks for your reply. I also found that negative loss we should as it is a dicretization error from lumerical in google answewered by ansys team. Should I bother about the negative loss value or not?I did the convergence testing and accordingly chose the FDE span but loss came negative. I increades the boudary, then the loss became positive. But my laptop is 32gb, so it is unable to process that large FDE for furtur calculation. I have attached the convergence. If I took x=24 and y=12, loss is positive but my laptop is hanging to run. So how do i solve this? how to fix the span for bent? if i fix it from convergence, it will be very small. I have attached the bend loss centre picture also. Please suggest a solution for that. My materail dont have negative refractive index. How do i sure that my FDE span is ok and im getting the correct bend loss for my structure? can you please expalin in detail about the PML settings you mentioned?

Regards,

Amritayan

 

 

[Amrita Pati]

Hi Amritayan,

Sometimes we see small values of negative losses due to discretization error which can be safely ignored. But in bent waveguides where you would naturally expect to see losses, we shouldn't see negative values. This most likely occurs due to the modes of the waveguide interacting with the PML boundaries. So, the PML boundaries should be moved further away from the waveguide on the outer side of the bend since the modes radiate in the outward direction. This is why you see positive values when the FDE span is larger. So, I believe you would need higher computing power in order to be able to perform the analysis.

Generally in waveguides, we can have a fine mesh in the waveguide region and a corser mesh in the rest of the region. But since in a bent waveguide modes extend outside the geometry, a coarse mesh outside might also provide incorrect results. So, I would also suggest you to increase the mesh accuracy for a given FDE span and see how the results vary. Also, in the images you have attached I do not see a negative values. Are they just the magnitude of the loss?

You can find more information about the PML parameters on the page below and the definition in the reference on this page:
PML boundary conditions in FDTD and MODE

Usually expert users use their custom parameters. I suggested some parameters in my earlier response because I was able to get rid of the negative losses in some of my examples using those paramaters  so I thought they might help.

 

Regards,
Amrita

[Amritayan Chatterjee]

Hi Amrita,

I have used 50nm mesh in FDE and mesh override dy= 2nm and dx=10nm found from convergence test. Attached images has positive loss as my FDE my large. I checked with your PML settings, the loss become postive for a bend radius of 500um bt for other values like 470um its still negative. For some values it gives postitive loss not for all. Another query I have for anisotropic material, when FDE solve neff, is it taking all the axis value of the anisotropic material? Like for my case x and y axes has different refrative index, and my waveguide is z propagating.

 

Sincerely,

Amritayan

[Amrita Pati]

Hi Amritayan,

I apologize for the late reply, I was out of office. It can be a bit tricky to accurately calculate the bending losses due to the radiation. So, convergence testing is recommended. As your results have converged for the larger FDE span (which is expected), I wouldn't change the advanced PML parameters.

For the anisotropic material, it would consider the different indices along the different axes.

Regards,
Amrita