Learning Forum

[Wan-Ling Tsai]

Hello all,

When I simulate a pure dielectric waveguide, I found out the waveport size does influence the results. The attachment shows the geometry simulated. Dielectric constant of the waveguide is 6.15, and the dimensions of the dielectric waveguide is 0.127 x 1.87 x 10 mm3. When I change the port size, the magnitude of S21 also changes. When the port size increases, S21 also increases and vice versa. I could not find the guideline to model the proper waveport size as to get a convincing result.

 

*Pink rectangular are the waveports, the purple object is the dielectric waveguide*

 

Does anyone have experiences in modeling dielectric waveguide in HFSS? Thank you!!!

[xuehewang]

if the strip inside is your waveguide, then your waveport is way too big, a waveguide typically uses a waveport that has the same size as its cross-section.

You may refer to the help file of HFSS, and input "waveport" in the search block.

Let me know of anything else that is unclear.

Enjoy!

[Wan-Ling Tsai]

Hello,

Thanks for your reply.

However this is a pure dielectric waveguide. It is not covered with PEC or conductive/metallic boundary so the waveport setting for my case should be different from the typical waveguide simulation in HFSS. 

Thank you still!

[asamaiyar]

Hi Wan-Ling Tsai,

Here's a suggestion:

1. Make Waveport using 8h x 10w rule


2. Request multiple modes from your wave port but do not assign any integration line


3. Plot each modes (by exciting them separately) to check which one is the one you want


4. Use that mode for all future analysis

Also, I feel that if you make your wave port large enough your results may start converging to an answer (not sure though, because this is pure dielectric waveguide).

Hope this works!

--AS