This video is taken from the FDE Learning Track on Ansys Innovation Courses.
Transcript
Here are some tips for setting up the solver region.
When studying bound modes, the solver region should be large enough to contain the entire
mode, including any evanescent tails.
If the solver region is too small, the presence of the boundary conditions may affect the
mode.
At the same time, it is not efficient to make the solver region larger than necessary.
Once the simulation is setup, a simple convergence test is to keep making the simulation region
a bit larger or small until the mode profile and effective index are not affected.
You can also plot the mode profile on a log scale, to ensure the fields are very small
at the boundary of the solver region.
For instance, you can check the fields are several orders of magnitude lower at the boundaries
by changing the color scale.
Here we set the minimum to -9, Some waveguide devices may have a high index
substrate, such as Silicon-On-Insulator waveguides.
In such cases, including the lower high index substrate in the solver region can make finding
modes very difficult.
This challenge occurs because the substrate also supports many modes that have a similar
index to the main waveguide.
In such simulations, we generally recommend that you don’t include the lower substrate
in the solver region.
It is not possible to make the FDE solver mesh arbitrarily small.
For 2D simulations, the limit typically occurs around 1000 by 1000 mesh points, although
the exact limit depends on the simulation.
To get the highest accuracy results, it is best to use mesh override regions to make
the mesh small only where it is most helpful, rather than everywhere throughout the simulation
volume.