How to obtain the proper values of S-parameter for LPF Richard's and Kurodo identity

Enjunear sir,
I tried for the Richard's transformation method lowpass filter design , got the proper s-parameter values for the schematic simulation. But I didn't get expected s-parameter values for the EM extraction. Design values are as follows.

Z0= 50 Ohm, f = 4 GHz, H=1.6mm , T =0.036, dielectric constant =4.4, Tangential loss =0.001

I am expecting following values for both SCHEMATIC and EM Extraction


Frequency DB S(1,1) DB S(2,1)

4GHz -3.1084 -3.0946
Herewith I am attaching the present getting layout and schematic diagram.
Even I reduced the X and Y cell size 0.01 , am getting memory error for the MWO EMSIGHT, AXIAM, ACE AND VELOCE ACF.Please help me sir

It is frustrating that you did not even try to understand the answers.

Volker Sir,

I got the answer for the 180 degree hybrid coupler, even marked as solved. I don't have any problem with RAT RACE coupler.I changed the design values for the Low pass filter Richard's Kurodo identity, couldn't get it. Those simulator will take more than 4 hours take to process(simulator like AXIAM , EM SIGHT, ACE.......).

Sure, because you have set inefficient meshing. The model that you have shown should not simulate more than a few minutes. Please upload your AWR model, so that we can have a look.

Volker sir,
I uploaded , you can have look sir. I used all the existing simulator and tried .Some are showing error, are taking more time to process

Volker sir,
I uploaded , you can have look sir. I used all the existing simulator and tried .Some are showing error, are taking more time to process

Can you upload the *.emp file?

Volker sir,
o.k sir I will upload it sir

You are in for a rude awakening if you think that the simple electrical model will quickly match up with the EM solver solution on a coupled circuit of any kind. EM solvers take into account ALL of the coupling and edge effects of the metalization.

As a general process, you start with a simple electrical model, tune it up, then move to the EM solver. After you see the initial result, then you have to re-tune the design so that the full EM solution gives you the correct response (since you have introduced all sorts of additional interactions by solving for all of the cell-to-cell couplings).

That filter should simulate very quickly. You can probably reduce your mesh size to ~10 cells/wavelength as a start. Also, use conformal meshing, if you have that option. That will reduce the number of cells by picking different cell shapes for different metal shapes (curves for rounded edges, rectangles for long, straight edges, etc).

I really mean the AWR project file, not the screenshot that you have uploaded.

---------- Post added at 09:00 ---------- Previous post was at 08:57 ----------

From my experience, for this circuit, it should not be far off. Honestly, I don't understand why he is using EM extraction all the time, for these simple circuits.

Volker sir,

I changed the design values , got the expected results. Design values for the Maximally flat top Low pass filter as follows

f0 = 1.5 GHz , Insertion loss =20db, H=1.6mm, T= 0.036mm , Tangential loss =0.001.
Cell size for x& y =0.1mm

Ok, fine. These wider lines are what I had suggested here https://www.edaboard.com/thread238319.html#post1019809

Volker Sir,

I followed as you said, changed the width and design also. I got the expected results for both Low pass filter Maximally flat and Chebyshev filter for differnt ripples values.