How to properly create stripline designs surrounded by dielectric substrate (not air)?

Salim_21845

I'm trying to simulate the stripline parallel-coupled bandpass filter design presented by the attached slides (pg. 28 and 29), but I keep getting incorrect results - I've tried this implementation in CADFEKO (finite substrate) and EDITFEKO (Greens function/inifite substrate) but they don't give me the corresponding results. I have also found a forum that discusses implementing stripline in FEKO - "https://community.altair.com/community?id=community_question&sys_id=b8b680ba1b2bd0908017dc61ec4bcb35" (uses air as dielectric), which I have read and followed but I still don't get the correct response. I have attached my CADFEKO as well as my EDITFEKO designs and the POSTFEKO results, any help to push me in the right direction or to point out what I'm doing wrong is greatly appreciated.

Torben Voigt

Hi @Salim ,

Please note, FEKO gives these warnings:

WARNING 510: Imaginary part of the impedance is large thus the real part will be inaccurate
WARNING 4722: It is highly recommended to use double precision accuracy (solution settings)
WARNING 509: Reactive power large thus real power is inaccurate
WARNING 4723: It is highly recommended to use double precision accuracy (solution settings)

I would recommend activating double precision and also I would suggest to use a finer mesh on the striplines:

I didn't run the simulation, but this is where I would start. The edited CADFEKO model is attached.

You may also play around with the mesh at those closely coupling edges:

 

Best regards,

Torben

Mel

@Salim ,impressed that you ventured into EDITFEKO.

You can add the infinite substrate and infinite ground planes in CADFEKO:

These kinds of filters require very fine meshing as Torben also pointed out, exacerbated by the 0.62 mm spacing between two pairs of those lines. Sometimes one needs to add the finite thickness to really properly model the coupling between the parallel lines. A compromise is to add vertical surfaces equal to the thickness on the edges of the lines as shown below:

This gives the result below. 

My model attached. You will need to use Feko 2020.1 to open the CADFEKO file.

Note I did not calculate the full set of S-parameters, just the VSWR according to your reference.

It does not look like the reference you provided is a full wave solution. Would be interesting to see a result from a 3rd party full wave solver too.

Salim_21845

Thank you, I got almost the correct response, but I'm assuming the deviation is due to the inaccuracy of the design equations, because the referenced slides don't run the model on a full-wave EM simulator as was pointed out by @mel, and so the filter would require tuning. I also noticed that I wasn't defining my Greens functions correctly for a two layer dielectric substrate in EDITFEKO so thanks a lot for that well, I was able to setup the infinite planes as desired in CADFEKO and read the associated EDITFEKO file that FEKO generates to see how to implement what I needed.