Problem with phase accuracy in waveguide polarisers

[PaulUK]

Does anyone know why it requires an extremely fine mesh to get sensible results when simulating various waveguide polarisers?
I'll post some examples when I get chance bit for square cross section corrugated polarisers I require about 80 to 100 cells per wavelength. For circular cross section waveguide I require over 300 cells per wavelength to get sensible results.

Two orthogonal port modes are being simulated and the S-parameters used to calculate the Axial ratio and differential phase.

This was going to be the basis for a polariser tutorial but the way things are going it wouldn't be practical with such huge meshes. Hope there is a solution to this issue. Thanks.

[Hale_812]

the answer is in your question. circular waveguide requires non-orthogonal meshing

[thorsten]

This sound indeed a bit surprising, but maybe you can supply your script and we can have a look?

In which direction to you need this dense mesh. In all or only in direction of the waveguide or across the cross section?
Do you think that the stair-casing of the waveguide boundaries are the problem?
What value exactly are you looking at? Maybe there is a problem using the correct interpolations?

regards
Thorsten

[PaulUK]

I shall post the examples that I have done and show their results either later or within a few days. I've just moved to Germany from the UK so things are a bit of a mess at the moment. As soon as I get my computer setup I will post what I have.

I previously did these simulations with CST and HFSS and the results converged well. So I used them as a reference or check that should be the aim of the openEMS simulations of the same structures. For the polariser tutorial I'd like to use openEMS off course.

Thanks.

[PaulUK]

the answer is in your question. circular waveguide requires non-orthogonal meshing

I've used the rectangular mesh with circular waveguide before, but I apply the appropriate circular waveguide port and had no problems.
I didn't use the cylindrical mesh because the slots in the waveguide walls of the polariser are not circular.