Reson for loss of highpass flter

[amicloud]

I designed a high pass filter on a substrate of Eps=3.38 and Heigh=0.5mm,
as shown in the attached graph.the loss in the passband is about 2.5db.and the
s21 curve is not smooth in the passband.especially when the frequency goes high.
The design is hope to has a highpass band to 13GHz.
Any one can give me some advice?
thanks.

 

[flyhigh]

Hi amicloud,

you didn't mention if you used circuit simulator only or you used EM simulation as well. Having in mind the size of a filter it might be a chalenge for EM.

Anyway, your filter seems to be mismatched meaning that positions (frequencies) of transmission poles and zeros are not where they should be. Filter has a lot of Ts and opens, and is electrically very long. There is an relatively large error in modeling Ts in cicuit simulators (2-3%). This make a small difference between stub reactance value expected in the simulator and a value in real circuit at every T. The other possible source of error is the open end of the stub. Error is smaller here, but it exsists.

When the small difference in a T is transformed over a large length of transmission lines (towards the input and towards the output), and when you take all Ts and add opens in consideration you will have large difference in overall filter response as a result. This is my explanation.

All I can think on to make your life easier is to use EM simulator. If the structure is to large to simulate in whole, you can at least encounter for every T and open in detail and than tweak a length of stubs and lines between the stubs to reachieve specifications. So, simulate Ts and opens in EM simulator and transfer them in ciruit simulator for tuning/optimization. I think this should work better than your first result.

flyhigh

 

[amicloud]

hi,flyhigh,


If it is because of the mismatch,the S11 must be poor,but here,the s11 is more than 12dB.you know sqare(S11)+square(s21) is approximate 1,when loss is very
limited,but here it is not the case.so there must be some reason else to incur the
loss.

thanks

 

[flyhigh]

Hi amicloud

one reason for loss apart from omic loss can be radiation effect. Try to put enclosure and see the effect of it. Your filter might be an antenna at some frequency.

flyhigh

 

[biff44]

I am afraid I do not understand. What is at the end of each of those stubs? What is the electrical length of the stubs at 13 GHz?

My general comment is that you show a lowpass or stopband structure, not a highpass one.

 

[amicloud]

I am afraid I do not understand. What is at the end of each of those stubs? What is the electrical length of the stubs at 13 GHz?

My general comment is that you show a lowpass or stopband structure, not a highpass one.

the stubs all are short lines to ground.
and you can read about J.S.Hong's book <microstrip fillters for RF microwave application>

 

[biff44]

And the electrical lengths of the stubs are what?