Help me with a simulation of UHF bandpass filter in MWO

Hi guys

I am struggling with a UHF band pass filter. Here are the specs:

Passband frequency range: 416 - 458 MHz
Center frequency: 436.5 MHz
Stopband attenuation: 60 dB
Loss: 1 dB or better
Filter order: 3

I have done the simulation in Filter solutions and when I buid it it doesnt work. I please want if somebody can do the simulation in MWO office and see how it looks like. I am not experienced with MWO. Any suggestions are welcome.

Your comments and help will be highly appreciated.

Thanks in advance

60 dB attenuation at what offset from the center frequency?
This is very important to see if is possible to get these performances using order 3 filter.

Thanks Vfone for your reply The filter is for a receiver. The image frequency sits at 295 MHZ so it must atleast be attenuated by the 60 dB. Thanks

I would recommend to check the designs of 432 MHz filters in ARRL Radio Amateur's Handbook.
If you need to suppress 295 MHz by 60 dB, you can add a simple high-pass filter with a pole at that frequency.
I have made various similar filters by stacking two to four coaxial or stripline quarter-wave resonators with adjustable loop coupling. Each resonator is enclosed in a metal-sheet case approx. 12.5 mm (1/2") square and 2...3" long. Tubular trimmers are used to tune the resonators and wire loops for coupling. To adjust you need a sweep oscillator and a detector connected to an oscilloscope.
In a similar structure, with a different coupling I made also very good stop-band filters- with 4 resonators, the rejection was >60 dB at any chosen frequency.
My experience showed that it was easier to cut-and- try than to use any design computation method, and I had better results.

Thanks a lot jiripolivka for your reply. Sorry for not clarifying things from the beginning. I actually want to implement the filter with lumped elements (on a small PCB)for a compact receiver. I checked in the ARRL Handbook they used stripline for the band pass filters. Using the stripline for my application will not be possible due to limited space.

Can I still achieve these requirements with lumped elements?

Thanks

To my experience I do not think you can achieve the specifications with lumped elements or printed circuits.
If you need a small-size filter, another possibility is a SAW filter but the development would be costly.
I would suggest to try to modify an UHF TV tuner; their bandpass filters cover 470-850 MHz, so an addition of trimmers replacing varicaps can do. I would try to build a similar structure.
I believe no other filter technology can meet your specification than the coaxial or stripline resonators.
Please let me know if you find something better or if I am right.

Thanks a lot once again for your help. Where can I get the schematic for that tuner? I will then do what is possible. I will let you know if I come across something.

I have no schematic of a modern TV tuner at hand. Possibly a TV repair shop may have one.
I would suggest that you take a TV tuner from a discarded TV set or a VCR and open it. Then you can see the design of the bandpass filter and try to make a similar structure. Such coaxial or stripline filters enclosed in metal cavities offer an advantage of a high Q, therefore such filter has a low insertion loss. Lumped element filters are much more lossy above ~200 MHz.
The high Q of the quarter-wave sections also allows for steep response over a pass band as well as a good rejection at required frequencies.
You have mentioned that your model of a helical filter failed in rejecting the 295 MHz image band. I made also helical filters, with similarly good results like with coaxial or striplines. For the high rejection, you can add one-two reject sections at 295 MHz, also with helical structure. Use a half-wavelength section instead of quarter wave one.

Did you set realistic Q values for the inductors? Wirewound SMD inductors have e.g. Q values of 50. You'll hardly achieve 1 dB losses with it.

It might be possible to design a combline BPF with additional trap to meet your requirements but it won't be 3rd order. However a 5th order filter made on Rogers 4003 material and using ATC 600S caps might end up being about 25mm x 40mm on a PCB.

The insertion loss would be more like 2 dB though...

Is this too big?

Yes I tried, I even in ADS with real parameters but it didnt give promising results then I decided to physically buid it to no avail.

---------- Post added at 20:40 ---------- Previous post was at 20:35 ----------

I think that is promising but I havent worked with compline before. Please shed more light. I think 3rd order and 1 dB was too ambitious so anything less than 4 db will serve me well.

How big can it be?

Something in this range 25mm x 40mm is not too big or a bit small will be great.

One type of combline filter uses microstrip resonators on a PCB. The resonators are straight bits of microstrip that are much less than a quarter wavelength long. They get tuned to resonance by fitting capacitors at the end.

For your filter I would suggest two cascaded 3rd order combline filters with trap coupling between them. This would be about 25mm x 40mm. The PCB material would have to be good quality eg Rogers 4003 0.032" and the capacitors would need to be very high quality eg ATC600S series (typical 10pF?) Such a filter would achieve about a 2 to 3dB loss and would reject 295MHz and below by over 65dB. Note: you would also have to fit some tin screens over the filter to get the stopband performance.

There are RF tools that can design combline filters eg Eagleware Genesys but in reality this is just a baseline start. I find it best to do the design with an EM simulator like Sonnet as the design needs to be subtly adjusted to optimise the resonator lengths to allow a single value of capacitor across all resonators and also to get good match across the band. I think a typical VSWR for the filter would be 1.5:1 across 416-458MHz after optimisation in Sonnet. Once the optimisation is complete a filter like this would be easy to make in large qty as long as you always used ATC caps and the same PCB material and the PCB was made professionally.

So the filter would end up about 8mm in height with the screens fitted. To make this filter you would either have to get it milled on a CNC machine (I have one of these made by T-Tech) or get it made by a PCB company. You might be able to etch the PCB but it would take several attempts to get it right. Do you plan to make lots of these filters or just one?

Elliptic-Cauer type filter will serve you but you should also create a notch at 295MHz or you may arrange one of the transmission zeros at that frequency.Otherwise standard filter prototypes in 3rd order don't give you 60dB attenuation in stopband..

I think Tuyoleni will need to maintain the 60dB rejection below 295MHz for at least 42MHz as this will be the image reject band? So a simple notch will be too narrow on its own. If space permits, the combline filter would be my top choice for this filter. This could be made in volume with no need to ever adjust anything. A lumped elliptic filter would be very difficult to set up and also would be difficult to reproduce reliably IMO.

See below for s parameters of a simple combline filter designed for this frequency band.

It's just a small PCB with 7 x 0603 package capacitors on it.

Thanks a lot once again G0HZU. You have real put in your best effort to explain how I can achieve my specs. The problem is that I am running out of time for my receiver project and I have also to work on ther parts as well. However this is something I will definately take time to do with time. Please keep it up. My final resort if I dont come up with a quick solution will be a helical BPF by Temwell or Toko. Thanks man

---------- Post added at 23:52 ---------- Previous post was at 23:42 ----------

Would you please attach the the project file also as I am having dificult to view the s-parameter? That sounds pretty small. Is this IMO or what did you use to design? Thanks

Why not try to contact some dielectric filter vendor? I think it's very easy to order one and don't need to tune any thing.

Or you can use Mini-Circuits Bandpass filter RBP-440+, (9mmX9mm), passband from 410M~470M, IL<1.8dB. And at 320M, IL>30dB. So maybe you can use two RBP-440+, such as filter+Amplifier+filter mode.
You also can use Bandpass+HighPass+HighPass mode. Here the highpass you can select Mini-Circuits RHP-395+. The IL is 1.88dB@410M, and 28.6dB@290M.
Then you don't need tune anything, and the specification should be very stable.

Thanks Tony. The filter+Amplifier+filter set is my design spec but due to the issues with the front-end filter Iwas considering discarding one filter and just have one. I will definately also look at these Minicircuits. I am grateful to you guys. Thanks

This looks like a very easy filter with LC design. Use pole placer and put a zero at 295 MHz. 3-pole design gives you rejection of more than 70 dB at 295 MHz and insertion loss of less than 1 dB with coil Q of 120 and capacitor Q of 1000. It can easily be made with a package size of 0.75"x0.4"x0.25".