I am designing a high frequency filter using a loop resonator, one of the main goals is to achieve a reflection loss (S11) of -20db in a bandwidth of at least 500Mhz around 13.5 Ghz.
I used CST studio suite to simulate and design the filter, I used a frequency window of 0 - 20 Ghz at first, and achieved the desired filter (picture 1). I tried to simulate the same filter using a window of 0 - 30 Ghz(picture 2) and observed a harmonic at 27 Ghz, which is natural, but the first peak is now reduced to only -11db, the goal is not achieved. the question is, why is my first peak affected when I only changed the the frequency window ? which one of the two simulations would match measurements in case of a real produced and measured filter ?
The second run has not finished to the required accuracy. It needs to run longer. The ripple at the top which is the size of the remaining error is about 0.2 dB which is about 97.5% +/- 2.5% of the power. The ripple signal is about 16 dB down which is dominating in the region where you want your match.
Ask for a higher accuracy.
The mesh will be different for the two runs, because of the different top frequency, (unless you make sure it doesn't change - can't remember how to do that) which always gives slightly different results.
Interesting, thank you. the frequency window and the top frequency affected the mesh which affected the simulation. When I tried to simulate using the same frequency window but different number of meshcells, I obtained different results, which one of both is more accurate ? is the number of meshcells a factor of accuracy ?
- Badges
- Science topic