Edge Coupled Mirostrip Bandpass Filter

Hi,

I am trying to design an edge coupled microstrip bandpass filter, for linear analysis I used Genesys's synthesis tools but when I put the values into Sonnet, the output is very bad. Although I care the box resonances in sonnet, there are some resonances around my pass band. Do you have any suggestions, or which software should I use for this kind of filter.

Thanks...

Use ADS or MWO. Both can get linear and EM simulation in the same time.

first try your filter in schematic with coupled elements from ADS microstrip pallete, do the linear (s-param) analysis.

Once the linear response is Ok, convert the same circuit in layout elements
and do momentum simulation . Small tuning may be required in layout to meet the actual response.

We have to find out if there is a problem with the filter itself (e.g. sensitive to substrate loss) or if there is a mistake in the EM simulation model.

Can you upload your Sonnet EM model?

Best regards
Volker

I think my problem is due to box resonances, I made the box very large and made the top of the box free space and the response is almost OK. However, I must use a box and when I use a very narrow box, the lines are coupled to the walls and if I use a large box then cavity resonances occur. What shall I do to prevent resonances in the cavity?
Also what is the cavity resonance formula, or how can I calculate the resonance frequencies of a box ?

The simulation box in Sonnet behaves like a metal enclosure in reality. You will get the same results and possible box modes when you package your filter hardware with these box dimensions.

So the solution is the same for hardware and simulation: keep the box size under control (not bigger than necessary) and use an absorbing top cover (Sonnet: "Free Space") at a distance of lambda/2 or less, so that radiated energy can be absorbed.

For the side wall distance, 3*line width or 3*substrate height should be enough to minimize the influence of the metal side wall.

With these settings, box modes are no problem in most cases.

Use "Analysis > Estimate Box Resonances" in the Sonnet Editor.

thank you very much volker_muehlhaus

I wanna ask one more thing, since there are walls at the input and output, what should be the length of lead lines (50 ohm lines), they also affect the results.

The length of these lines is not critical because Sonnet always does a port calibration and removes these fringing fields at the feedpoint. You do not need to add extra length!