Parallel-coupled bandpass filter

[miss_belle84]

parallel coupled bandpass filter

Dear all, please help me regarding to my final year engineering project. i am about to design a parallel-coupled bandpass filter that operates from 18.5Ghz to 18.6GHz with 0.1dB ripple. my problem now is that i don't know on how to select the number of orders, n . in my project there is no minimum attenuation that is specified.

1) Do i need to chose any minimum atttenuation in order for me to use that attenuation characteristic for 0.1dB ripple?

or

2) Is there any formula that can be used to calculate the minimum attenuation required?

3) Is there any book recommended that is good for me so that it can help me to design this type of filter?

Thanks for helping me.

 

[amu811]

you can read <Microstrip Filters for RF-Microwave Applications> by J.S.Hong

 

[biff44]

Well, where do I begin?

18 Ghz is a pretty high frequency. Your bandwidth is pretty small (<0.1 GHz). So you are going to have a whole bunch of practical implementation problems.

Suppose that you use an alumina substrate for a microstrip filter. If you simulate it properly, you will probably see that the predicted insertion loss will be on the order of 2 to 4 dB for a multisection filter. So the choice of ripple factor is not really all that important.

In engineering one designs to specifications. If your customer (or teacher) has not given you specifications, then you must postulate your own specifications, design to them, and see if your design makes sense.

Lets say that you want low loss from 18.5 to 18.6 GHz. You might figure out that the εr of alumina is not well controlled, perhaps +/- 0.2. You also know that you can not hold tight etch tolerances of the final artwork, for a student you might get processing as good as +/- 0.5 mils. So the first step is to design a "guard band" around you filter to take into account these two factors. You perhaps need to design a filter that has low loss from 18.4 to 18.7 GHz, or even broader!

Once you have figured that you, if there is no rejection requirement, pick an easy N number (like 3 or 5) and go ahead and design the filter. When you have the physically predicted dimensions, sketch it out to scale and see if the dimensions make sense. Do you have 10 mil wide resonators, but they are space 40 mils apart? Then you are going to have a lot of stray coupling, and the packaging is going to be a major part of the performance.

For such a high frequency, and such a narrow bandwidth, it would be good to take your final design and pass it through an emag simulator program to get a better prediction of the performance.

 

[miss_belle84]

thank you amu811 and biff44. both of you really help me through this project.

biff44, you said that my bandwidth is very narrow. means that i need to increase my bandwidth then. how much should i increase? actually my design is to be operates at K-Band (18GHz-26GHz). Between that ranges is necessary for me. And my teacher asked me to use RT Duroid 5880. Is there any problem that i will face in the future if i used RT Duroid 5880 subtrates?

i'm sorry for my bad grammar. im not good in English. =)

 

[YCChiou]

At first, I think you need to check your design specification.
You should make sure what fractional bandwidth is.
In your description, the center frequency seems to be 18.55 GHz and the circuit bandwidth is 0.1 GHz. It's too small to realize on PCB technology.
Even you can simulate your circuit, I doubt if you could successfully fabricate it.
The insertion loss will be too high.
I suggest you should read some books about microwave filter. For instance,
1. Pozar-Microwave Engineering
2. Matthaei, Young and Jones-Microwave Filters, Impedance-Matching Networks, and Coupling Structures.

 

[miss_belle84]

yeah, you're right. i should check on the design specification first. i'm thinking of to change my lower and upper frequency. i would like to make the bandwidth broader as everyone give comments on the bandwidth. it seems that there is problem with the bandwidth. so, i decide to make it operates at frequency 18.5GHz up to 19.5GHz. What do you think? Is it the best decision? Thanks for your help, anyway. i'm new to microwave design actually.

 

[viswatess]

Hi,
I want to design a X band -band pass filter having a 2% of fractional bandwidth.I want to realise on printed technology.suggest some topology like DGS,PBG,SICC ,aperture coupled etc.....

 

[nandopg]

Hi fellows,
Narrow band filters in microstripline or even stripline on a substrate ? Forget it !!! You simply don´t have enough unloaded Q´s resonators.

In a practical perspective, you can regard 5% of percentage BW as a practical limit for a microstrip filter and 2% for a strpline filter built on substrates. You can get a little better than these figures if you use a suspended microstripline, but at the expense of a much more complicated strucure.

To get a better filling on this, go to Sonnet´s website download their EM simulator -Student Version and run some simulations to get the unloaded Q of several different combinantions of physical structures.

To do something useful with these simulations, you may want to have a look on this paper:

**broken link removed**

NandoPG