Photonics

Photonics

Topics related to Lumerical and more.

Comparison of EME and FDTD’s trnasmission

    • YOON Seokhyeon
      Subscriber

      Hello everyone.

      I'm studying taper structures made of Silicon nitride.

      First, I conducted a simulation using the EME solver’s propagation sweep at a wavelength of 1310 nm to find out the propagation loss of a SiN taper.

      The result was as follows: I got a length of 5 um (propagation loss = 0.945). 

      Then, to compare it with the EME solver’s propagation loss, I simulated the same structure(legnth = 5 um) using FDTD. 

      At the wavelength of 1310 nm, I got a propagation loss of 0.999. I am curious why the data from the EME solver and FDTD are different. 

      I believe the two simulations should produce the same data. If anyone has experience with this or knows about it, please respond.

      I can provide additional structural parameters if asked.

       

      Thanks,

      YOON.

    • Guilin Sun
      Ansys Employee

      It is hard to define what amount of difference can be considered as the "same" result. When we use two different solvers, we know their sources of errors are different. In EME, it might be the number of groups, cells, and meshes; in FDTD, it can be mesh accuracy, PML and others. In terms of FDTD, the finer mesh can lead to better result, provided that the PML is thick enough to precent any reflection. 

      In your FDTD file, I can see that the structure ended in front of PML, which is not commonly used. Usually we recommend that the structure is extended outside of PML. In addition, the default mesh accuracy 2 (and the 8 layer of PML) is to simulate the device with relatively fast yet reasonable result. It does not mean it is highly accurate. 

      Please get more information regarding to converging tests:

      https://optics.ansys.com/hc/en-us/articles/360034915833-Convergence-testing-process-for-FDTD-simulations

      https://optics.ansys.com/hc/en-us/articles/360037172994-Convergence-testing-process-for-EME-simulations

      Personally, I would not compare them. Instead, I will just use one solver and do convergencing test to make sure the result approaches its theoretical value. Comparing the result accuracy from two solvers needs significant efforts.

       

    • YOON Seokhyeon
      Subscriber

      Thank you for your reply.

      I'll try it the way you told me.

       

Viewing 2 reply threads
  • The topic ‘Comparison of EME and FDTD’s trnasmission’ is closed to new replies.