[SOLVED] The simplest and tunable microwave filter

[baby_2]

Hi,
For a specific application I need a microwave filter(micro-strip or wave-guide) which works at 10Ghz with 20Mhz bandwidth. So for this little bandwidth I need a microwave filter which has a good tunability to compensate fabrication's errors.

what do you suggest?

Thanks in advanced

 

[BigBoss]

Re: the simplest and tunable microwave filter

You will work relatively narrow bandwidth so Microstrip Structures can be an option but tuning will be a bit troublesome.( Trimmer Capacitors ended Microstrip Line Filter )
I suggest you to use mechanical filter structures such as Cavity Resonator Filter with Tuning Screws.You can adjust easily the Filter Response with mechanically tightened screws.
But designing such filter is not easy at all.You will need a strong 3D EM Simulator knowledge..

 

[SLK001]

Trying to achieve a 20MHz BW at 10GHz is going to be problematic for you - especially since you don't seem to do this on an every day basis. What is your application? Is seems like you are trying to achieve your selectivity at the RF stage - not a smart approach. Your better approach would be to perform your filtering at a lower IF stage (much lower).

There are times when you do need to filter at the RF stage, but when you do it there, your sensitivity will suffer greatly. Also, you will need equipment in the $100k+ range to be even able to attempt and produce the design. At your desired bandwidth at 10GHz, probably a 2nd or 3rd order coupled coaxial cavity resonator filter would be the better approach - however, the design would occupy a fair amount of space.

 

[baby_2]

Thanks for your contributions
It's related to my university project and my professor idea and maybe a challenge to design microwave filter with low band-width. would it be possible to introduce me to some good reference that cover cavity filters full and with some good examples in designing?

Thanks

 

[SLK001]

One of the best resources for microwave filters design is Microwave Filters Impedance-Matching Networks and Coupling Structures by Jones, Matthaei and Young. It should be in your university library. Just know that this book is NOT for the layman. If this one is too complex for you, you might find design info on cavity filters in college level Microwave Engineering Electromagnetics textbooks. Also, search the IEEE database (probably MTT-S) for helpful info on the design - this also should be available at your library.

 

[baby_2]

Thanks,
Yes, in our university we have learned some part of Matthaei book(for lumped element and also microstrip). But I can't find a good reference that covers cavity filter design with good details and step by step designing procedure.

 

[biff44]

that is very narrow band. I would recommend a waveguide bandpass with susceptive irises spaced at half wave, with a varactor diode right in the middle of each cavity. The varactor WILL lower the Q, so it will be a banpass with rounded corners for insertion loss.

- - - Updated - - -

The problem is you have a very high center frequency, and a very narrow passband. Hence the percentage bandwidth is very small.
If you first downconvert your signal, to maybe a 1 GHz frequency, filter it THERE for 20 MHz bandwidth, and upconvert again to 10 GHz, you will have a very easy to do tunable filter! The local oscillator to do the up/down conversion should be simple, since most of its phase noise or frequency error subtracts out in the up/down conversion.

 

[baby_2]

Thanks a lot for you explanation,
Actually we have down-converting and filtered our signal and after up-converting we need a narrow band filter to reject f1+f2 and save f1-f2. Could you introduce me a good reference about designing IRIS and your proposed structure?

 

[volker@muehlhaus]

Actually we have down-converting and filtered our signal and after up-converting we need a narrow band filter to reject f1+f2 and save f1-f2.

Two stage conversion would be the better design, e.g. going from low first IF to 100MHz second IF, then converting 100MHz to 10GHz. Then you don't need these extreme 0.2% BW filter specs.

 

[baby_2]

Thanks for you comment,
we first down-convert 10Ghz to IF and after some processing we need up-convert to 10Ghz again. In addition as you know the output of mixer (by neglecting intermodualtions) has two different frequency and by knowing that F1+F2 and F1-F1 are close to each other a narrow band-pass filter is required.

- - - Updated - - -

Thanks for you comment,
we first down-convert 10Ghz to IF and after some processing we need up-convert to 10Ghz again. In addition as you know the output of mixer (by neglecting intermodualtions) has two different frequency and by knowing that F1+F2 and F1-F1 are close to each other a narrow band-pass filter is required.

 

[volker@muehlhaus]

Thanks for you comment, we first down-convert 10Ghz to IF and after some processing we need up-convert to 10Ghz again. In addition as you know the output of mixer (by neglecting intermodualtions) has two different frequency and by knowing that F1+F2 and F1-F1 are close to each other a narrow band-pass filter is required.

Everyone here understands the mixing products, and that is why we keep telling yout about the standard solution: up-/downconvert in multiple stages, with filtering at each stage. For 20MHz to 10GHz, your single stage conversion is a bad concept, and that is why you need this extreme filter spec.

With dual conversion and a 500MHz first LO, you then have 500MHz +/- 20MHz in the first IF (easy to filter) and up/downconvert the filtered signal to 10GHz, where you get 10GHz +/- 500MHz. Both image signals are easy to filter. Do you get the point?

 

[baby_2]

Thanks volker@muehlhaus for your good explanation.
unfortunately, In our laboratory we don't have many mixers and also signal generators to up/down convert in different stages. You are completely right it might be difficult way but I will very thankful if you can help me in deigning microwave filter design with appropriate band-width.

 

[BigBoss]

unfortunately, In our laboratory we don't have many mixers and also signal generators to up/down convert in different stages

Since your laboratory is poor in term of equipment,component; have you ever considered the cost of such ( cavity resonator filter) filter ??
Believe me, it will cost you much.It's absolutely precision mechanical production with CNC /3D.

 

[baby_2]

Thanks Bigboss for your comment,
Fortunately mechanical faculty has good equipment in designing mechanical parts and I hope they can help me in fabrication and I hope again my professor's grant would be enough. So don't we have any other ways to design a narrow band microwave filter easily?

Anyway, all of the people in this post helped me a lot to be aware of designing problems and also I've learned more.

 

[SLK001]

Well, if you're not locked into a cavity filter, look into "evanescent mode waveguide filters". These are resonant structures built using waveguide in the cutoff region that can produce extremely narrow bandwidths - and the insertion losses are quite low. They are also relatively easy to hand construct. There have been articles about these filters in ARRL and IEEE MTT-S publications. To me, these articles are about as "step-by-step" as they get.

 

[biff44]

well, it seems you have skipped the SYSTEM DESIGN part, where you compare block diagrams and frequency plans to minimize output spurious products. So...i guess adding a tunable waveguide filter might work best.

IF you are willing to entertain a block diagram change....i would take my baseband signal, upconvert it to say 1.9 GHz, pass it thru a saw filter, THEN upconvert that signal up to 10 GHz with a tuneable LO.
Use a filter like this one:
https://www.digikey.com/product-detail/en/abracon-llc/AFS20A20-1910.00-T3/535-12067-2-ND/4036384

but before you rush off and change the system, do a frequency plan/spurious output analysis. there are plenty of online programs that will predict spur outputs.