Learning Forum

[Sara Congia]

Hello everyone,

I am trying to simulate directional couplers (FDTD region 100x50µm) using the FDTD software, and everything works when I keep my wavelength around 1550nm. For my application I have to design structures (in silicon nitride) close to visible wavelength. Let's say lambda=800nm.

My simulations always diverge, therefore I tryed to follow the instructions written here:
https://optics.ansys.com/hc/en-us/articles/11277217507603-Troubleshooting-diverging-simulations-in-FDTD

By following the webpage, seems that I have a PML type of divergence, that I have tryed to fix by changing the settings in the PML (k, alpha, stabiliez ...). But the simulation still diverges. In particolar, by using a monitor on top of the structure, the field grows at the output port (see the figure).

Do someone have any other suggestion?

Thank you,
Sara

[Amrita Pati]

Hi Sara,

As first step we should identify which type of divergence it is. What happens when you change all the PML boundaries to metal? Does the simulation still diverge? 

Regards,
Amrita

[Sara Congia]

 

Hello, 
Thank you for your answer.

Yes I followed all the instructions written in the webpage: 
https://optics.ansys.com/hc/en-us/articles/11277217507603-Troubleshooting-diverging-simulations-in-FDTD

It is a PLM-type divergence. So I put PML to Stabilized, then I changed alpha by putting to Custom… But nothing worked

Regards,
Sara

 

[Amrita Pati]

Hi Sara,

Like the article might have mentioned it could also be due to the material fitting. Sometimes the instability occurs where the dispersive material intersects with the PML. Have you tried terminating the object just before PML (with the extend structure through PML option disabled in the Boundary Conditions Tab of the FDTD object) and see if the divergnce still occurs? 

Regards,
Amrita

[Sara Congia]

 

Hi Amrita, 

I tried with different materials, and the simulation alway diverge. I have just tried what you suggested, to stop the object WG just before the PML. Of course, at that point the light is reflected back, therefore the simulation diverges due to the reflections, this time. Should I forse the simulation maximum time to espire before the light reachs the PML boandaries? 

 

 

Thank you,
Sara

 

[Amrita Pati]

Hi Sara,

Forcing the simulaion to completion before it reaches the auto shutoff level might result in inaccuracy. For calculating the Fourier transform the fields must decay sufficiently enough and an auto shutoff of 1e-05 works for most cases. I would recommend you to check how the auto shutoff level changes over the duration. If the simulation diverges just before reaching the auto shutoff, for ex. it decays to 5e-05 and then starts growing exponentially, you may try to end the simulation earlier. But if it starts diverging pretty soon then there is no point doing this.

Do you have a script that you are using to produce the entire simulation setup? In that case you can send the script here and I can run some tests as well. Thank you!

Regards,
Amrita