Problems with S-Parameters Extraction using INTERCONNECT and FDTD



I am trying to extract the s-parameters for different structures. I am trying to get (S11,S12,S13,…,Snn); where n is the number of the ports in my structure as illustrated

I tried to follow the exact instructions illustrated in this example and apply it on a directional coupler (DC) that is symmetric about x and y axes.

I have done a FDTD simulations of that structure and the results can be found in DC.fsp (1.8 MB)
After finishing the s-parameters sweep, I extracting the s-parameter file DC_sparamaters.txt (571.9 KB)
an error message appears telling me that the device is not reciprocal image
Why this message appears to me?

When I tried to visualize the s-parameters matrix. The DC is designed to be 50/50 at 1.55um and when I viewed the s-matrix at that wavelength it doesn’t show me a symmetry as expected (i.e. S14 doesn’t equal to S14 or S23 or S32)
Why is not the s-matrix symmetric as expected?

After importing the extracted file of s-parameters sweep in FDTD into INTERCONNECTS and doing the following setup
Then I run the simulation and viewed the transmission at each port to get the following result that doesn’t support the FDTD results. The transmission of both input 2 and input 3 should intersect at 1.55um, but it gives me the following response. DC.icp (729.8 KB)


Why don’t the results from INTERCONNECTS follow the results from FDTD?


Hi @g15081147,

I will focus here on the FDTD simulation as it might be the root of the problems you mentioned. I had a look at the file, the first main issue is the simulation time: FDTD being a time domain method, we calculate the frequency/wavelength results by doing a Fourier transform of the time domain fields.

For this to be valid, we need the fields to have decayed enough by the end of the simulation. Ideally we would like the fields to be 0, but in practice, we evaluate the amount of energy left and stop the simulation when it reaches a threshold (by default, the threshold is \(10^{-5} \)). This can be checked during the simulation and in the log file by looking at the auto shutoff level.

In your case, the simulation doesn’t run long enough. At the end of the simulation, there is still a lot of energy left (over 90%) so the frequency domain results are not correct.
The first thing is to increase the simulation time to a value large enough to allow the simulation to reach the auto shutoff criteria.

Additionally, you can also:

  • Increase the size of the ports as they seem too small and are most likely truncating the fields (this will affect the mode calculations)
  • Reduce the length of the simulation region (so it ends a bit after the ports). This will reduce the propagation length
  • Increase the width and height of the simulation region (same reason as first point)

Once we are sure the FDTD simulation gives the correct results, we can then calculate the S matrix and compare with the INTERCONNECT simulation.

I hope this will help!