Problem with bandwidth in combline round rod filter

hi every one , i have designed a combline round rod filter for 112 mhz band width at 4.45 ghz center,
but im getting the band width of 220 mhz and the return loss also very poor that is 5 db with in pass band ,the pass band is from 4.4 to4.6 ghz


is any solution so thst i can get the 112 mhz band width and y the response is so wide like that?
any one clear my doubt plz?

thank you

Re: band width is more

Hi,
Try to design a higher order filter.The BW may reduce owing to the increase in selectivity.
bye
Bhanumurthy.

Re: band width is more

Check the spacing between rods. It is possible that they are over coupled.

BR,
RF-OM

Re: band width is more

thank you for urs suggesion i will try both of ur advice, but
my designed filter is 7th order and it is working for good(3 db loss in passband) for different tappings but returrn loss is bad(5 db ) in pass band and the band width is 200 mhz only , do i have to improve the order ?

i cross checked the spacing values they are same as i desiognd only is there any way (reducing length of resonator or b depth of box or endwall space)so that i can reduce my band width ,

only with tuning only can i reduce my band width ? is it possible if yes?
how much i can reduce the band width by using tuning ?


thank you

Re: band width is more

I a little bit confused. Let’s do it one step in a time.

1. Required filter order. If you provide me with your spec I may check this number for you.
2. Walls can affect your bandwidth and you need to find right distance to walls.
3. You may reduce the length of the resonators and then compensate it with capacitance. But you will reduce the L-to-C ratio and, therefore the Q-factor of the resonator(s). Play with these parameters and find the best combination for you.
4. Insertion loss of the band-pass filter is determined primarily by two major factors. The first one is required bandwidth. As more narrow the filter as higher the insertion loss. The second factor is unloaded Q-factor. As it higher as lower will be insertion loss. If I remember it right, insertion loss can be approximated as IL=4.34*(SUMgn)(QL/QU). Here IL is insertion loss in dB; SUMgn is the sum of filters g coefficients, QL is loaded Q-factor, it depends on required bandwidth and QU is unloaded Q-factor. Practically QU is the quality of the resonator you can actually build.

I hope this can help you.
Best regards,
RF-OM

Re: band width is more

hi thank you for ur all advices its useful for me ,and
this my filter spec 4.4 to 4.5 ghz and 0.1 ripple or 0.01 also 40 db @ 50 mhz away
using combline round rod filter,
i dont know whether u have information about combline round rod filters if u have means,
tell me the steps u told already i.e
any corrections and my end wall space is 7.41 mm from the end of the last resonator rod

nothing confusion i can design this filters but tuning is very difficult and dont know how to do ? thats the problem so i want to know the basics of those filters sothat i can tune it ......
thank you

Re: band width is more

The minimum required order is 6, but it is even number, so there may be a problem with equality of input and output impedance. I think 7th order is better as an odd number. Unfortunately I cannot tell you about the best spacing or distance to the wall. For this it is necessary to run actual EM simulation. The tuning is a hard task. There is one old method: you turn off all resonators except one and tune this one to required by design frequency. Then turn off all resonators except the next one and tune this one. Then repeat this for the rest of the resonators. The idea is to turn one resonator at a time when all other are tuned off (far away). The main problem is how to turn off the resonators. It really depends on frequency, type of filter and so on. The idea is to detune the resonator far from filter's band so it is not react with other resonators any more. You may do it by shunting resonator with cap or resistor, may be even short it with wire. Try to watch the filter on Network Analyzer screen and turn off the resonators. You probably will find a good way to do it. There is no one method that fit all cases. After you tune all the resonators there may be necessary to slightly tune them again. It is good idea to remember or mark all tuning element states and then slightly tune them in order to get required characteristic. As closer your spacing to the right value as less tuning is necessary after individual resonator tuning.

Sometimes it is possible to tune the filter with EM simulator and get rough values of tuning elements. It may help with final tuning of the real filter. You need to know that despite the fact that filter is passive element oscillation is possible inside the filter. It is hard to believe but it is the true. When high Q resonators interacts there are very steep slopes of energy exchange are involved. Actually it means very sensitive and sharp stable points in resonators interaction. Small shift and they are jumped well away one from another. You may think about it as a pencil that put vertically. It is definitely passive element, but it was installed in conditions that close to unstable. Small force that inclines a pencil a bit will be enough to abruptly move it to the horizontal state. This is the one shot oscillation actually and this is roughly what is going on into resonators when they are exchange the energy due to mutual coupling.

I hope this information will help you.

Best regards,
RF-OM

Re: band width is more

ok i will try your procedure waht you have given here, tomy filter
thyank you

Re: band width is more

RF-OM said:

Check the spacing between rods. It is possible that they are over coupled.

Click to expand...

here you mean that the more spacing than the required or designed may lead to over coupled... and band width may increase thats what you teling here or other wise.........
any thing else
and i have another idea about this that is
due to the very low loss also may band width high and by reducing cavity depth we can reduce band width .... is it right?
just clarify this...

Re: band width is more

Yes, I mean the distance between roads. When it is increased, coupling is decreased and bandwidth become narrower. I do not know your exact design. Of course, cavity dept may change the coupling too, but if everything is the same, changing the distance between rods will work as described above.

Best regards,
RF-OM

Re: band width is more

when i want to increase the spacings the length will decrease and b depth will increse as per my design equations ......
but your words telling that with out any change in those terms just by changing the spacing and remaining things as it is we can get reduced band width.......
im i right.

but how much i can increse the spacing can u tell me any ratio.........
like that
here is my design 4.4 to 4.5 ghz , 0.01 ripple, 7 thorder,40db@50 mhz away,
112 band width , this is the spec of the combline round rod filter
.....................................................................................................................
spec........*GPS..*dia....*RL......*spacing.................*TL........*tap......*EWS........
....................................................................................................................
as above.*13.8..*5.5....*13.36..*14.51,16.17,16.5...*108.65..*1.39....*9.89............
.....................................................................................................................this is the filter i designed all are in mm
GPS=GROUND PLANE SPACE
RL=resonator length
TL=total length
EWS=end wall spacefrom rod center
tap= tapping from gronding end of the resonator

could you tell me the spacing how much i can improve
thank you

Added after 2 hours 20 minutes:

and also even im getting 200 mhz band width but return loss is very bad as 5 to 23db in pass band (how to improve that ....... can i add cross coupling in between two resonators so that the tuning is possible for thae improvement of return loss........)and is there any thing close relation to these two factors that is
1>band width redeuction
2>return loss improve ment

Re: band width is more

Unfortunately, I cannot check this numbers right now because software what installed on my machine have no ability to design combline filters with round rods. I need to check other programs and if they can do such a work install it and then check your numbers. I am pretty much busy these days, but will try to check the programs. For me it is rare case because all our today designs require very small size. We have no luxury to use big or tunable filters, 5x5mm is the maximum allowable size (of course for small power) and there must be no tuning at all.

Best regards,
RF-OM

Re: band width is more

ok i was waiting for u r response so that i can get any solution, even now also when ever u have time dont forget to post this ans if u know ........

thnak you