Why is the S21 of this 20 GHz BPF periodic??

[giorgi3092]

Hi,

I am working in AWR and realized a BPF with coupled microstrip lines.
So, I have two schematics:
a) Ideal (designed with AWR iFilter tool)

b) Realized with microstrip technology (values were optimized)

Now, here is the resulting graphs

dashed regions are optimization goals that were used for optimization. You can see that the transmission coefficient (S21) of the microstrip version (last red line) goes back up to 0 dB at the end of the graph. Meanwhile, the ideal version does not do that (pink line).

Can anyone explain to me why S21 goes back up to 0 dB at the end of the graph? What is the phenomenon at work? How can this be improved?


edit:
here's the layout as well

Thank you,
Giorgi

 

[vfone]

This is an issue of the coupled-microstrip line bandpass filters, which suffer from the presence of the parasitic harmonic resonances.
There are many topologies on the net how to reduce these parasitic resonances, especially the 2nd harmonic.
One option to minimize the 2nd harmonic resonances is to use the topology (b) in the picture attached.

 

[volker@muehlhaus]

What is the phenomenon at work?

Unlike your LC prototype, the lines are also resonant at 3*f0, 5*f0 etc
That's n*360° additional phase = additional full circles in Smith chart.