Some questions about microstrip filter.

I have designed a LPF microstrip filter with a stop band of 2.2GHz.
It consists of a microstrip main line and several open-circuited stubs that are vertically connected to the main line.
The impedance of the microstrip main line is 100 Ohm in order to reduce the size of the filter.
What's the best choice of the impedance of the microstrip main line?
Is open-circuited stubs commonly used in passive circuit? I heard that short circuit are usually used instead of
Is open-circuited stubs, But 1/4 wavelength is too large for my circuit.

Can anyone give me some advice on the filter design?
Thank you.

Dear htforever,
there is a lot of designs for microstrip filters, i advise you to read Colin's book "Microwave foundations", or other related books. bytheway, i think a good choice for your case is to use gap filters. this topic is widely considered in all microwave circuit books, however, i can post some papers about it if you can not find anybook about.

marti

Hi,mamali
thank you for your reply.
I have a book : Microstrip Filters for RF/Microwave Applications. Jia-Sheng Hong, M. J. Lancaster
The filter design is not my major task, so I don't want to waste time on it.
I just want to design a filter, LPF or BPF both are ok.
It should pass the signal of 1.8GHz to 2.2GHz, and reject signal of 2.8GHz to 3.2GHz. -20dB rejection is ok.
I think it's not a difficult job.

hi htforever
you can use interdigital tapped filter.This filter will fit your requirements.The best way to "design" filter is to use Filter synthesis software if you don't want to waste time on it.

I use ADS filter DG to synthesize.
Then convert the LC to microstrip lines, use momentum to validate the performace.
Is this ok?

thats ok. bytheway, i can help you if you let me know the substrate parameters you use, and your design restrictions such as riple you can allow, the minimum distance between microstrip lines or space limitations.

marti

Hi,Mamali:
I need a LPF or a BPF, both are ok.
I have designed one using ADS.
For LPF, pass band:0--2.2GHz, out band:2.8GHz--3.5GHz. In band loss<3dB, out band rejection>20dB(the bigger the better).
For BPF:pass band:1.5GHz--2.2GHz,
out band:2.8GHz--3.5GHz, In band loss<3dB, out band rejection>20dB(the bigger the better).
minimum distance:0.2mm
Prefered distance:0.4mm
Using FR4,thickness=0.8mm, Er=4.5, tanD=0.0001

htforever wrote:
I use @DS filter DG to synthesize.
Then convert the LC to microstrip lines, use momentum to validate the performace.
Is this ok?

I use this method too and everything is OK.I think that tanD = 0.025 for FR4 this is very important for insertion loss and slope of filter.

tipical FR4 loss tangent: tand=0.022
(er~=4.2)
marti

Thanks to pva and mamali!

Now I changed my substrate, A substrate for high frequency circuit.
Er=2.8, H=1mm. But I don't know the tanD, is 0.02 OK?

Hi, pva:
I use ADS filter DG to synthesize a LPF, the result of lumped component circuit is good.
But when I use Transformation Assistant of filter DG to convert the LC to microstrip lines, the simulation result is very different from the lumped component circuit.
So, I tried to change the Fp, Fs, Ap, As manually to adjust the design, because I think I should believe the result of microstrip circuit because the filter is made of it.
The simulation result of microstrip circuit is a little different from the EM simulation, and I think I should believe the result of EM simulation.
This is right?

you should use a resonable substrait, which is available at the market, you may use this:
RT/duroid 6002 ; er=2.94 +/- 0.04; loss tangent: tand= 0.0012 (at 10GHz). but this is an expensive material :). dont forget to search for an available substrait, it is very impotant, and you can not gain a practical design, if your substrait is not achievable.

and, the simulation results (MoM is better, especially Zelan d's IE3 D), should be trusted, not the lumped elements models (such as what is available in Micr0wave 0ffice).

marti

My director provided me the material, I have got it.
But he only knows the Er and H, don't know the exact the model.

thus, er=0.01 is ok, but ask your director for exact value. (or name). bytheway, if you can reach IE3 D, it has a FilterSyn, wich is very nce program, and easily u can simulate it with the most precise moment code available. very reliable package.

marti

I am familiar with ADS, it's 2.5D EM simulation.
tanD=0.01 or er=0.01?

sorry , tand=0.01

Would you like to take a look at my simulation:
Please check the attachment, a PDF file.

i think the results are acceptable, u should trust simulation results of MoM code, frequency shift is resonable, try to enlarg your design for about 5%~10%, (shifting frequency to left), then acording to microstrip model (which you have to gain a frequency shift to left), your final results will be acceptable.

marti

Thank you, I will take your advice.

hi htforever
this is for example in my last design of LPF-elliptic (cut off F=1.5GHz ),frequency shift between measurements and simulation was -10MHz for EM simulation (Zelan d IE3 D) and +40MHz for circuit simulation(Seren@de).The material is Rogers4003

Hi pva:
Your design is successful.
Frequency shift of my LPF between EM simulation and circuit simulation is about 400MHz. Oh, my god, I can't even image the result of measurement.
Pass band0--2.2GHz, stop band is 2.8--3.5GHz