the simulation about PSR of the published paper

dear all
I have bulit a 3D FDTD model file about polarization spilitter rotator based on a published paper.The I want to get the similar result of the electric field distribution just as the paper shown.

But what I get from my simulation is different from it,which as shown below.

furthermore,I found that the mode source of TM in my file looks strange.
Is the TM mode source I set right?Hope to get your feedback.Thanks a lot.
1.pdf (1.1 MB)
1.fsp (370.5 KB)

Hi @jbwei,

There’s a couple of possible issues with your simulation. The first, is the mesh, that seems a bit coarse, but that is ok for a starting point.

The second issue is the regarding the mode calculation. The mode source seems a bit small, so I would increase its size vertically and horizontally, to make should its boundary won’t affect the mode calculation. You can start with half a wavelength and check if the fields have decayed enough when they reach the limits.
This may implies that you increase the size of the FDTD region, and there goes the main challenge: your structure is very big, so you may need quite some resources (memory and simulation time) to run the calculation.
With such propagation length, you could use the Eigen Mode Expansion (EME) solver from MODE Solutions.

I hope this helps!

Best regards,


dear gbaethge
Thanks for your reply. I will have a try just as you mention,including increase the size of the FDTD region and the size of the mode source. It will take me some time to run the simulation. If I have any question,I will post it out.Thanks again for your help.

dear all
I have increased the size of the FDTD region and the size of the mode source.The result is shown below.

the result looks like better than the result got before but it still has a bit problem. Do you have any other advice. Thanks in advance.3.fsp (611.4 KB)

dear all
Can anyone help me solve the question above.Thanks a lot

Hi @jbwei,

Sorry for the delay! The one point I would check now is the mesh size. You need to have a mesh fine enough to correctly resolve the structure, including the gap between the 2 waveguides.
I would use a mesh override object to specify the mesh size there, and check how the mesh size affects the results.