Classic Microstrip/Stripline Filter Design Synthesis

(version 1.1 11/25/2001)

LLfilter.zip - this bundle includes a MathCad worksheet (and PDF format for viewing if you don't have MathCad) for synthesizing edge coupled bandpass filters (almost identical for the straight or hairpin shaped resonator cases) as well as stepped-impedance lowpass filters and lumped lowpass/highpass filters.

• Also included are some tools for synthesizing the coupled lines in microstrip or stripline. These are Excel Spreadsheets:
  • Stripline Version by Dr. James Rautio (Sonnet Software). This is mainly an analysis tool, one could wrap the solver around it in order to synthesize a desired coupled line section (physical dimensions).

  • Microstrip Version by Lance Lascari, using the excel-solver (suggestion By Dr. Rautio). The equations for synthesizing microstrip are far from closed form. The solution I chose, found in Brian Wadell's Transmission Line Design Handbook(original reference within this package) contained around 30 different sub-equations to make up the entire *analysis* solution. I devised a couple of lame metrics for meeting the goals of a coupled line section (based on coupling and impedance), and the solver will iteratively try to minimize the RMS error between these two terms. You should make sure and install the Microsoft Excel Solver "add-in" which is freely available for Excel, but may not be installed by default. This tool is mainly for getting you close if you don't have a better tool for synthesizing coupled microstrip. Something like sonnet lite or your favorite EM and/or linear simulator can perhaps get you a little closer. If you do not install the solver you are still able to analyze, you just can't have it synthesize dimensions based on your target coupling, electrical length, and frequency.

    Note: to install add-ins in excel 2002, you click Tools, and go to the "Add-ins" menu. Other versions may be slightly different. The "Solver" is then accessible on the "Tools" menu, and must be invoked there for it to run.

• My assumption is that people will choose the tapped filter design approach over using input/output couplers, but either choice is an option. Slight adjustments usually need to be performed once the design is entered into a simulator to account for the discontinuities (bends, end-effect capacitances, TEE junctions, etc) and any approximations in the coupled line design. Note: I've compared the coupled microstrip models between EM simulators and circuit simulators (both Libra/Eagleware).... there always seems to be a couple/few tenths of a dB difference in coupled line coupling. To obtain the best results I sometimes just load up a coupled line model and adjust the spacing and widths to achieve the prescribed coupling in the simulator, this usually only requires an adjustment of 1-2 mils (thousandths of an inch).
• Hopefully you'll get more than your money's worth....it's free!
• Please contact.html let me know what you think and if you have been able to use this successfully. Constructive criticism welcomed.
• Note: the lumped LPF/HPF synthesis for equal terminations has been included as an option in the Smith Chart program (version 0.725 and greater) as well.

This is a more compact "hairpin" type as described by Kenichi Takahashi, Morikazu Sagawa, and Mitsuo Makimoto in "Miniaturized Hair-Pin resonator filters and their applications to receiver front-end Mics", 1989 MTT-S Digest. I've had fairly good luck designing these. though I haven't really used the design procedure in this paper, it is where I got the idea.

This picture was generated using Sonnet Lite plus (a tool that is well worth the price tag for small to medium sized EM problems (of course these terms are relative... I consider most of the EM analysis that I do on the "medium demanding" level at most). One feature that is added with the "plus" over the free version is the DXF import capability, so you can use something like Eagleware Genesys schematic/equations/Layout to generate "artwork". I do this constantly because it's very easy within Genesys to setup equations that can adjust an entire filter structure (such as those above) with just a few variables (this makes it easy to reuse designs and also really helps when you're trying to make sure that all dimensions map nicely to the EM grid).

Eagleware also offers some really nifty synthesis tools for lumped and distributed filters (FILTER, M/FILTER, and S/FILTER) as well as an integrated EM simulator (EMPOWER). If you're on a budget and need to do some microwave filter work, I highly recommend Eagleware Genesys (basic package includes linear simulator, layout, schematic entry) coupled with something like sonnet LitePlus. With that as a very solid starting point, you can gauge which direction you might want to go (synthesis tools, different EM simulation capabilities, etc).

I just received sonnet lite plus version 8 (6/19/2002) and found that I could do a nice analysis of this the "compact" filter above in perhaps 3 minutes (on my newer computer). The new ABS feature showed exact agreement (visual on graphs), yet smoother curves with 5 frequency points (analysis frequencies). Previously I used 41 points to achieve resolution that wasn't even as good, which took at least (41/5) times longer! The particular sweep range was roughly 2.5 times the passband width, centered on the passband response.

I think we should commend Sonnet and Eagleware for keeping the "starting price" of their tools affordable and the quality exceptional.