Mesh order's effect on regions that don't overlap

needattention

#1


meshorder

Thanks for reading my problem. Recently I was measuring the transmission at the through port of a ring resonator. The ring’s x span is 0, so it doesn’t overlap with the waveguide. But when I changed the mesh order of the purple ring in the figure, and remained other parameters totally the same, it showed a very different spectrum in the through port.

I was very puzzled because I thought mesh order served no role for objects which do not overlap. So where the problem is?


#2

Hi @yuqinie

You can use index monitor to see how the simulated geometry is affected by changing the mesh order. You can also send me your simulation file for a review if you still had any problem.

Thanks


#3

Thanks. Please help me with that problem. The files are duplicates and I just change the mesh order. I don’t know what happened.
Best,

Yuqi

ring_mesh order 2.fsp (528 KB)

ring_mesh order 3.fsp (528 KB)


#4

Dear @yuqinie

Software uses more steps to define materials in the boundaries. For example, as is shown the screenshots below obtained from xz index monitor, ring will be extended through substrate once the ring mesh order is 2:

Mesh order 2:

image

mesh order 3:

image

To solve the problem, you can refine the mesh in the z-direction as part of convergence testing:
https://kb.lumerical.com/en/index.html?layout_analysis_test_convergence_fdtd.html

Also, I noticed that the FDTD z-span is small and PML may truncate the evanescent fields. Please increase the z-span and make sure that there is at least half wavelength distance between PML and edge of your objects:
Setting about FDTD Region and mesh


#5

Dear @bkhanaliloo

Thank you so much for your suggestion. I would change the z span to 4um. I also choose to keep all the material’s mesh order 2 except etch. So this problem won’t happen.

Best,

Yuqi