"I am curious about the reason why, in a fully symmetric LPF filter design, the S12 and S21 values are the same, but the graphs of S11 and S22 appear different.

I would appreciate it if you could answer my question. I have implemented an LPF filter and designed it symmetrically. However, while the S12 and S21 values come out the same, the graphs of S11 and S22 show different shapes.

The circuit structure, port locations, sizes, and materials are all strictly symmetrical. The mesh and feed settings for simulation are also exactly the same for both ports, with the port impedance set to 50 ohms.

Nevertheless, I cannot clearly understand why the S11 and S22 characteristics differ at high frequencies.

Could you please explain whether this is caused by numerical errors, limitations in the simulator’s internal operation, or other physical or modeling reasons?

Also, if S11 and S22 differ in a symmetric circuit, what points should I check or what review methods do you recommend?

I would appreciate any advice to improve the reliability of my simulation results.

Answer 1

[Anumeha]

Hi 용모,

I understand you observe a significant difference (~6db) at higher frequencies. Ideally it should be the same. But any minor difference in the implementation (for example adaptive meshing) of both ports will have a significant impact on higher frequencies. Try swapping ports and see if the graphs swap exactly to pinpoint an implementation issue.

Moreover, both graphs could show better performance in terms of matching and losses. Check out this link below for details on LPF implementation.

https://www.mathworks.com/help/dsp/ug/lowpass-filter-design.html

Hope this helps!