Bandpass Filter Design ?

Which is one better butterworth or Chebyshev to design bandpass filter for channel 44 UHF ?

butterworth's response is flat in its pass band but it's transition band is not much abrupt,where chabyshev has ripples in pass band but its transition band is abrupt so it signal dies out quickly

Now it's up to you what exactly you want out of your filter.

What your mean out of filter? Can you explain briefly about output from both of them? Thnx before bro,,.. :idea:

i personally thinks that u should use Butterworth if possible because in butterwoth group delay is almost contant in passband and it is also easier to make than chebyshev

Hello,

the chebyshev design gives steeper skirts, so you need less components to get a certain out of band attenuation. The disadvantage is the pass band ripple and more signal distortion. If number of components is not a major issue, I would go for the Butterworth (with flat pass band) option. .

From my experience, the multi-pole Butterworth filters are easier to tune.

if you want steep response then use chebyshev but at the same time you have to consider the fact which is said by WImRFP. I would also go with Butterworth as you can also get steep response by incresing order of the filter.

TV Channel filters are not done with discrete components because the stopband attenuation will be poor even you use nth order ..
I suggest you to use coupled helical coils in plated resonators with apertures.Realization and tuning will take time at the beginning but later you will see that performance of these filters are very good..

Hello,

You may visit The Utah ATV page - Putting together an interdigital filter

This shows an example of UHF filter design as done by radio amateurs (HAM).

well suggestions from WimRFP & BigBOss are quite good... but keep in mind these filter are mechanical so u need to have some good mechanical workshop....
but with helical filter definitely u will get very good response...

hello,

If there is time to experiment, you can make such filters from scrap FR4 copper glad material. You use the FR4 material for making the housing. The resonators you can make from metal strip also. Such materials are easy to machine with hand tools.

Without access to a VNA or equivalent setup, design on your own is very time consuming and requires good RF skills. If the specs are not that tight, a 2 pole filter (two resonators) just over critically coupled can be an option. You can do this just with a signal source and (diode)detector.

If small space is important, the helical wound resonator, as suggested by ashishchandra, is a good alternative.

If specs are not too tight, and porangan has access to EM simulation tools, a microstrip implementation on Teflon or equivalent material may be feasible.

By the way, butterworth or chebyshev that need less money to create Bandpass Filter ? And which is ones that need fastest time to create it ? because I have no longer time bro,.. :sad:

porangan said:

By the way, butterworth or chebyshev that need less money to create Bandpass Filter ? And which is ones that need fastest time to create it ? because I have no longer time bro,.. :sad:

Click to expand...

please provide with the all the specification u need for the application like rejection at stopband,stopband frequency etc. so we can provide u with some good topologies for that....

I don't know your specs for the filter, but when it is for digital TV (OFDM, multi-carrier system), significant group delay is allowed. This allows large pass band ripple Chebyshev approach.

As you know, a Chebyshev filter with large ripple, requires less poles (w.r.t. Butterworth approach). When, based on the specs, a 2 or 3 pole Chebyshev filter is possible, it may reduce the component count (if this is of importance). Off course good RF tools (at least means to determine |S21|) are required.

Note that large ripple Chebyshev has bad S11 (or return loss), even in the pass band.

It is my problem now, because I have not determined it's spesification. Could you give me some source about specification of BPF for channel 44 UHF (654 - 662 MHz)? like bandwidth -3db, bandwidth 50db, insertion loss pass band and stop band, VSWR and impedance ?

For your application you need to know the rejection wanted at the Local Oscillator Frequency and Image Band Frequency. I assume this is for pre-selection at the channel of interest. Thus if the vendor is using a high side Local Oscillator then the filter needs to suppress the Local Oscillator Frequency on the High Side of the Channel and the Image Frequencies on the High Side.

To determine your Local Oscillator frequency identify the Intermediate Frequency in the receiver IF strip. I do not know what is normally used in current receivers but 44 MHz was a nominal IF in past years in home television markets. In set with a 44 MHz IF Frequency using high side conversion your Local Oscillator would have been a nominal 44 MHz above the lower edge of the channel in use. The Image Band would then occur at a frequency corresponding to 2 times the IF frequency plus the Lower edge Frequency of the channel you are using.

Then you determine the suppression that is needed for these two frequencies. In other words just how far down does the Local Oscillator Amplitude have to be. In many cases this is driven by national regulatory rules. In the US the FCC mandates that receivers cannot emit RF signals from receiver Local Oscillators above certain levels. So what you have to determine is what the level of the Local Oscillator signal is at the input port of the mixers. Your mixer ideally would not have any Local Oscillator signal at the input port but mixers are not perfect and part of the local oscillator signal will appear at the input ready to travel back out your input rf amplifier to the receivers antenna terminals where it will jump off the receiver antenna into the ether just as if it was designed to be a transmitter.

Determining the image band rejection is another matter. Frequencies appearing in the image band can mix with the local oscillator to convert down to the IF frequency. Thus the objective is to suppress the signals in the image band. The objective is to suppress signals in the image band enough so they do not produce signals large enough in amplitude to interfere with your signal processing system.

With that said I suspect 25 to 30 dB of Local Oscillator suppression will be adequate based on previous general receiver design work I have done. Better receivers typically have 70 dB or more of rejection in the image band but your signal processing system may have the ability to deal with higher levels and you may not need nearly that much image rejection.

The other thing to realize with your filter design is that to achieve a narrow bandwidth filter in the 600 MHz band can be a tough problem. "Loaded Q" of the circuits comes into play. If your passband is 6 MHz and the nominal frequency is 600 MHz for the center frequency of the filter then you are dealing with a filter that will need a Q of 100. It will almost surely require the use of helical resonators. Discrete components such as capacitors and inductors will not likely provide the needed Q at those frequencies. Other choices include using sections of transmission lines (coax resonators) etc which bring other problems at what are sometimes called re-entrant frequencies.

Most likely you can widen the response of the filter as most regulatory agencies do not allocate frequencies on adjacent channels int he same geographic areas. Thus instead of trying to achieve a filter bandwidth of 6 Mhz
(if that is the bandwidth of the channel) you could design the filter for 12 Mhz thus relaxing the Q requirements by a bit.

1 Determine your Local Oscillator Frequency
2 Determine your IF Frequency
3 Determine the LO Frequency Suppression needed
4 Determine the Image Frequency Suppression needed.
5 Determine if you are using high side of low side conversion
6 Determine the ripple you can tolerate in the passband (this is driven by the systems tolerance to amplitude changes and signal delay tolerance as a function of frequency)


Then get back with me here and I will see if I can be of help. Hopefully this is not to much to late.

Harry