PML boundaries on MODE EME not giving proper S Parameters

    • jacob.campbell

      I am running Lumerical 2021 R1.4. 

      A confusing problem happened: when using EME to design a taper, I get pretty normal results when I use Y symmetric or antisymmetric boundary conditions, but when I run the same simulation with PML boundary conditions, the S parameters go to very small values (<1e-15). This is a large disparity dispite the nearly identical field profiles.

      When the EME taper examples are run with the three boundary conditions mentioned here, we get the following (graphs are for S21 vs group span during span sweep):

      This problem is present in our custom taper designs as well, and it makes us doubt our simulation fidelity. To my knowledge, all-PML should match at either anti symmetric or symmetric in a geometriy that is symmetric about that axis.

      Changing energy conservation from passive to conserve energy did not help when I tried that.

    • Guilin Sun
      Ansys Employee

      I guess one of the main issue is the output mode:

      as you can see, when you use PML, the field profile is similar to sym-ymin. In this case you will need to choose the output mode with y-symmetry property.

      As  you can see, the mode profiles agree. So it is the mode issue. If you are not sure the polarization of the mode, you can choose several output modes. Or you can use "user select" to choose proper output mode.

      Please note that, when no symmetry BC is used, the effective number of modes is roughly decreased to half of the case with symmetry BCs, since in general odd/even modes cannot be coupled to even/odd modes. When symmtry BC is used, the other half of modes without this symmetry will not appear. I assume roughly odd and even modes are half-half by numbers.


Viewing 1 reply thread
  • You must be logged in to reply to this topic.