Cavity bandstop filter

[TheMaksimys]

Hi everyone!

My goal is to design the cavity bandstop filter with the 50 dB rejection at the 6.7-6.8 GHz. I found an attactive topology in the "Microwave filters for communication systems" book (figure attached). The coupling matrix for this topology can be estimated using various available CAD tools. Yet I have not found any papers where the physical implementation of such a filter was reported.

I have used Filter Designer 3D along with the CST Studio to compute the coupling matrix and construct the cavities with the desired couplings and quality factors. The simulation of the filter assembly, however, had demonstrated very poor perfromance and the S-parameter matrix decomposition signalized that the generated structure is severely detuned and it has numerous parasitic couplings ( which prevents layout optimization).

My question is whether the demonstrated topology is practically feasible? Also If someone could suggest the better method/topology for the cavity bandstop filter with the high rejection, that would be excellent!

Thank you for your help!

 

[biff44]

just like in the real world, your simulated filter needs to be tuned. Specifically, you need to place the stop band poles on exactly the right frequencies

 

[TheMaksimys]

just like in the real world, your simulated filter needs to be tuned. Specifically, you need to place the stop band poles on exactly the right frequencies

There is a little issue with tuning: Filter Designer offers the mathematical matrix decomposition technique to demonstrate the offset of the generated structure couplings , resonant frequencies and the quality factors. This method worked well when I was making a cavity bandpass filter. Yet for bandstop filter, the decomposed matrix is a coplete nonesence, which is not of any use for tuning. In theory, I could vary the physical dimentions of the filter (there are about 25 independant variables for the 4th order filter). But it is very time consuming process with no guarantees.

Therefore my question regarding the demonstated topology of the bandstop filter: Does it exist in reality or it is just a beautiful misleading mathematical model? I have found zero articles, publications and mentionings of this filter implementation in the hardware.

Thank you!

 

[biff44]

is there a better topology?

well to get 50 dB of rejection, you need high q resonators, and almost no leakage paths across the filter.

Maybe suspended substrate filter with multiple quarter wave open circuited stubs. you would have to carefully channelize the conductors so there were no waveguide modes generated.