Microstrip bandpass filter design

EDAboard.com Forum Index -> RF, Microwave, Antennas and Optics -> Microstrip bandpass filter design

Author

Message

CZa

Joined: 11 Mar 2009
Posts: 4

26 Oct 2009 16:15 Microstrip bandpass filter design

Hello everybody,

Fist off, I'd like to thank this community for all the valuable information and the answers to many of my questions I've found here so far. This is my first post.

Although microstrip filters are not the main part of my job, I need to design one every now and then. Up until now, the design flow I've followed is described below.

I design the filter using the ADS Design Guide for my particular requirements, I export the design to layout and perform a full wave electromagnetic simulation (Momentum). Normally, the behaviour is not satisfactory so I modify the parameters I input in the Design Guide and repeat the process.

After a few iterations, once I've obtained the desired performance, I manufacture the design and measure it. Sometimes the results are not acceptable so, once again, I have to modify the design parameters and repeat the process.

This has always been a quick and easy way to design my filters (even though I've never had particularly stringent requirements). However, I am currently working on a filter with the following specifications:

Pass band: 14-14.5 GHz
Max. attenuation pass band: 3 dB
Rejection bands: 13-13.5 GHz and 15-15.5 GHz
Attenuation rejection bands: >20 dB

Previously, I had to design a filter with the same specifications but centred 1 GHz below and I didn't have any problems. However, I've been trying to design this one for weeks and I haven't been able. The electromagnetic simulations are not similar to the expected behaviour, the attenuation in the pass band is unacceptably high and the lower skirt is too mild, regardless of the order of the filter.

I've read many comments in different fora and I don't know what to do. I'm thinking about trying to design it manually using the equations shown in Hong's book.

Any ideas / suggestions? Do you think there might be a problem with the models to calculate the impedances of the odd and even modes? Would it be a good idea to try and determine these values through full wave simulations? Any suggestions on how to do that?

Thank you very much in advance.

VSWR

Joined: 07 Feb 2002
Posts: 767
Helped: 72

26 Oct 2009 22:19 Re: Microstrip bandpass filter design

Could be many reasons. Zap your project files and upload here. Someone may take a look.

biff44

Joined: 24 Dec 2004
Posts: 1835
Helped: 244
Location: New England, USA

27 Oct 2009 3:08 Re: Microstrip bandpass filter design

the higher the frequency, the more important secondary effects are. Is your housing resonating. Are you experiencing dielectric dispersion that is screwing up your center and coupling factors, etc etc.

CZa

Joined: 11 Mar 2009
Posts: 4

27 Oct 2009 11:50 Re: Microstrip bandpass filter design

The file is very simple, that's why I didn't upload it in the first place. I upload it here anyway in case somebody wants to take a look.

I know that things generally become complicated when you go up in frequency. However, my impression is that the 1GHz step between 13 and 14 GHz is not that significant as to cause such levels of damage in the performance. Furthermore, the designed filters at both frequencies for the same requirements are qualitatively very different (number of stages,...) so I'm affraid the models used by ads might be wrong at some point.

As I said in my previous post, the lower skirt is too mild regardless of the order of the filter and I don't know how to tune it. I've unsuccessfully tried to run a global optimisation (keeping the symmetry in the filter stages). The problem seems to be that, if I want higher attenuation of the lower rejected band, then I must assume a higher attenuation on the lower part of the pass band.

Any ideas/suggestions?

Thanks in advance.

Sorry, but you need login in to view this attachment

CZa

Joined: 11 Mar 2009
Posts: 4

27 Oct 2009 19:52 Re: Microstrip bandpass filter design

The performance of the schematic simulation of the filter is compared to the electromagnetic simulation in this figure.

Any ideas/suggestions/references on how to tune this design?

Thanks in advance.

Sorry, but you need login in to view this attachment

tomtomson

Joined: 11 Apr 2008
Posts: 11
Helped: 2
Location: California

07 Nov 2009 4:52 Microstrip bandpass filter design

Can you post a picture of your schematic and momentum layout?

Google
AdSense
Google Adsense

07 Nov 2009 4:52 Ads

kspalla

Joined: 19 Aug 2008
Posts: 848
Helped: 140

09 Nov 2009 12:55 Microstrip bandpass filter design

It look like you are designing with 3 resonators.
In schematic the results look like satisfactory to you. But in schamatic it accoins the coupling of individual coupled sections only.
It will not account the interaction between the couplers 1 to 2 ans 1to 3 etc. which will be accounted in momentum and hence the bad results.

You need to re do the optimization in momentum.

Schematic is only a startup for you to guess the initial values of the coupler/s and their parameters.

Element_115

Joined: 23 Mar 2002
Posts: 359
Helped: 23

09 Nov 2009 16:45 Re: Microstrip bandpass filter design

Have you looked in the design kits?
Also if you have Gennesys you should look at their filter wizard, it is really
helpful and easy to build filters.
@ 14 GHz you might want to use something with a low εr. because the
trace widths might be so small that you board maker can not make what
you need i.e. if you need a trace that is 1mil but the board house can only
spec a ±2mil tolerance you're in trouble.

Also make 3 design one centered, one tuned lower, and one tuned higher in
frequency. This way you will know which way to tweek things, and make 3
50 Ω 1λ traces, one 40Ω, 50Ω, & 60Ω. So you can see what is the impedance
discrepancy. Do this all in one file/board so it's all made at the same time.

Good Luck