Sonnet co-calibrated internal ports

Hi

I have a question on Sonnet’s co-calibrated internal ports.

Sonnet can predict box resonances and if they are present the software usually provides a warning during simulation. My experience is that the presence of box resonances only influences the simulation time (for ABS) and not the accuracy of the s-parameters.

My question is: If Sonnet predicts box resonances during simulation and prompts the user with a warning, are the s-parameters associated with co-calibrated internal ports still valid?

I am currently designing a stripline filter and I have inserted co-calibrated internal ports in the resonators. The simulator warns me of possible box resonances. When I simulate the multi-port network, Sonnet requires a lot of frequency points to converge. If I pull the resulting multi-port s-parameter file into Microwave Office and connect circuit-theory components between the ports to “repair” the resonators, I get the expected response. However, when I tune the circuit-theory components, I find something similar to high-frequency noise on my frequency response. Could something go wrong with the de-embedding of the co-calibrated ports if possible box resonances are predicted?

I am by no means an expert and would appreciate any thoughts.

Kind regards
tg_brand

tg_brand said:

Could something go wrong with the de-embedding of the co-calibrated ports if possible box resonances are predicted?

Click to expand...

Yes. Port calibration is based on the assumption that lines are single moded. Double the length of the cal standard and you double the phase, etc.

Now if there are box modes, this means multi-mode propagation, and port calibration can indeed fail. Besides the obvious (make box dimensions smaller), you should contact Sonnet support to discuss possible solutions.