How can i match SIW filter to 50Ω microstrip line?

[maryam sh]

SOS!

how can i match SIW(substrate integrated waveguide) filter ( with 7.5Ω impedance) to 50Ω microstrip line? which kind of transition is better? GCPW, tapered or quarter wave length transition?

 

[biff44]

Re: SOS!

bandwidth?

 

[burton]

Re: SOS!

Microstrip taper is the best. But SIW filter can be directly excited by 50 ohm microstrip and it is not necessary to use the transition.

 

[maryam sh]

Re: SOS!

bandwidth=40MHz
freq. : 8.2-8.4GHz

Width of SIW filter = 15.56 mm
Height of substrate= 0.5 mm
εr=2.2

with these information, the Impedance of filter is ≈ 7.5 Ω !! i think for matching need to transition. ??
I want to simulate this kind of filter with FDTD. first, i used only 50 ohm microstrip to excite filter. I examined different position of microstrip along the width of filter, but the result was very disappointing! S11 was very high.

 

[sundarmeenakshi]

Re: SOS!

bandwidth=40MHz
freq. : 8.2-8.4GHz

Width of SIW filter = 15.56 mm
Height of substrate= 0.5 mm
εr=2.2

with these information, the Impedance of filter is ≈ 7.5 Ω !! i think for matching need to transition. ??
I want to simulate this kind of filter with FDTD. first, i used only 50 ohm microstrip to excite filter. I examined different position of microstrip along the width of filter, but the result was very disappointing! S11 was very high.

hi which formula r u using to find the Impedance of filter

 

[Garbonzo]

Re: SOS!

Couple tips from my meager work on printed waveguide:

1. Microstrip -> Dielectric Loaded WG (SIW is a BS name for a prehistoric concept) tends to be pretty inductive, I've yet to stumble across a decent model for this transition.

2. If using HFSS to model your transition and/or filter (HPF in my case) feed it with a rough guess as to the width of your first matching section (or rough guess as to the finish width of your taper). Not necessary, but cut out an iteration or so. De-embed to the transition(s).

3. Put 2 extra lumped ports to your layer 2 (ground plane) right at the transitions to yield a 4 port set of s-parameters. Or if you are making just a transition and not a HPF, it'll yield 3 port parameters.

4. Back in your linear tool (MWO, ADS, etc) set the optimizer loose on a shunt capacitor on the 3rd (and 4th if you have it) and Zo of the microstrip port. It'll get clear quick that a pretty large capacitance is needed (realizable several ways, like big tabs on either side of your feed line), and that the resulting Zo for a good match is off +/-30% from your rules of thumb (the matching of the inductive transition is largely to blame).

5. For relatively narrow band matching you should only need a couple quarter wave sections. Getting that match good (~20 dB RL) over an octave and really close in to the cutoff at the same time requires a couple tricks I've yet to see published (hint: stupidly obvious once you see it, but strangely nobody seems to be doing it).

 

[Garbonzo]

SOS!

As a followup:

A 1964 paper I ran across discusses really broadening the match similar to what I empirically came up with:
https://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1125812

I found that by using 1 or 2 roughly quarter wave sections of slightly wider waveguide at the launch I was able to dramatically improve ease the task of matching to the launch. Specifically it really tames that very rapid Zo change near cutoff.