Photonics

Photonics

Why dose bending loss get nothing?

    • KEITA
      Subscriber

      Core is n=1.7, the size is 400nm(hight) × 1.5 μm(width)

      Clad is n=1.45(SiO2), the size is 6.4 μm(hight) × 10 μm(width)

      When the value of bending radius increses, in less than 50 μm, no physical modes were found.

      In more than 50 μm(50.1 μm or more), the value of angular loss can be caluculated.

      However, the value of angular loss for the first time is too low I think.

      When the value of bending radius is 50.1 μm, the angular loss is 0.73 dB/rad.

      However, when the value of bending radius is 50 μm or less, angular loss cannot be caluculated suddenly.

      This is correct?

       

      I have another question. The core is difference from previous one.

      Core is n=1.6, the size is 400nm(hight) × 3.0 μm(width)

      Clad is n=1.45(SiO2), the size is 6.4 μm(hight) × 10 μm(width)

       

      When the value of bending radius increses, the value of angular loss increses!

      When bending radius are 400, 450, 500 μm, angular loss are 1.25, 1.76, 2.25 dB/rad.

      This is unbelievable I think! I expect this waveguide is too difficult in terms of bending.

      What do you think about these?

       

      Thank you for reading.

    • Guilin Sun
      Ansys Employee

      For your first question, you will need to increase the number of the trial modes. By default it is 20. When PML BCs are used, it will create a lot of PML modes, which needs to be get rid off. So more number of modes are required.

      It seems it can find modes for larger bending radius, so you will need more number of trial modes in order to find the modes with smaller bending radius, as the mode confinement is weak.

      As for the second question, you may need to increase the simulation span, or increase that side where the mode shifts. If you have further questions regarding this, please write another post.

       

       

       

       

       

       

       

       

       

       

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