Home Forums FDTD Solving waveguide mode

Viewing 5 reply threads
  • Author
    Posts
    • #19384
      Alvin Hui
      Participant

      When using the mode solver associated with OptiFDTD, there are ADI and FD Anlso methods. Can anyone suggest when to use ADI and when to use FD Anlso? My situation is that the waveguide is quite small (width: 450 nm x height: 220 nm). The input light wavelength is 1550 nm. There is always no mode found when using ADI method. When I use FD Anlso method, I found that there is one mode by setting the Eigenvalue Solver to Subspace Iteration. However, I am just doing try and error and not sure whether the mode is correct or wrong. Can anyone suggest what are those properties? Thanks!

    • #19523
      Steve Dods
      Participant

      The FD mode solver and ADI one use different meshes. Are they set the same in your project? The ADI mode solver sometimes misses modes if the complex acceleration parameter is not a suitable value. Can you attach the project that got these results? I might help us to find the reason for the difference.

    • #19533
      Alvin Hui
      Participant

      Thanks Steve! After I reduced the mesh size, the ADI mode solver returned one mode. However, the mode is a bit different from the mode found by the FD method.

      Modes Modal Index Polarization
      Mode 1 2.31022444, -0.00336495 FVect TE (by ADI method)
      Mode 1 2.34768099, 0.0 FVect TE (by FD method)

      The project is attached. Thanks for your help!

    • #20347

      wish one will be suggested? ADI or FD?

    • #20361
      Steve Dods
      Participant

      FD mode solver has been proven to be more accurate. Especially for higher order modes.

    • #20419
      musabtr1
      Participant

      Dear Alvin,

      The FD more accurate for the small geometries less than 2 micrometre, due to concentrating of the mesh in a small scale that’s why the results become more accurate. for your case I suggest to used the FD mode.

      Make sure that you are using the PML to obtain the accurate values.

Viewing 5 reply threads
  • You must be logged in to reply to this topic.