is this bpf design possible

[lhlbluesky]

i have to design a bandpass filter with adjustable center frequency f0=20khz, 26khz, 38khz. and for all the three f0, requiring that: Q>10, gain attenuation at 120hz is larger than 80db, gain attenuation at 50khz is larger than 70db. how to realize the bpf circuit? and is it possible?
thanks.

 

[lhlbluesky]

if it is not possible, what circuit should i use?

 

[BradtheRad]

From what I know about filters, these are demanding specs.

High Q would be possible with a biquad or state variable type. Each type uses 3 op amps. The capacitor and resistor values should be matched.

However you will need a different design than biquad or state variable type, if you are to achieve your stated dB attenuation. The order number will need to be high, meaning several capacitors and resistors, etc.

It will take some work to make it possible to tune to a center frequency. The traditional way is to use a ganged potentiometer which changes several resistor values simultaneously. If you're lucky there is a modern design which can make the job easier (maybe using varactors).

 

[LvW]

i have to design a bandpass filter with adjustable center frequency f0=20khz, 26khz, 38khz. and for all the three f0, requiring that: Q>10, gain attenuation at 120hz is larger than 80db, gain attenuation at 50khz is larger than 70db. how to realize the bpf circuit? and is it possible?
thanks.

I think, the specification is not complete yet.
1.) Do the attenuation requirements apply to ALL three center frequencies?
2.) Are there no other bandwidth requirements than Q>10 ?
3.) No gain requirements?
4.) Which parameters are to be kept constant for the three center frequencies (gain, bandwidth, Q)?

- - - Updated - - -

Supplement:

I am afraid, realization is not possible. An attenuation of 70 dB at 50kHz (center at 20kHz) requires (for a 2-stage filter) a quality factor above 100.

 

[lhlbluesky]

hi, LvW.
1) the attenuation requirements apply to ALL three center frequencies.
2) i want to suppress noise at 120hz and 50khz as much as possible
3) gain requirement is not strict.
4) Q to be constant here.

An attenuation of 70 dB at 50kHz (center at 20kHz) requires for each stage (of a 2-stage filter) a quality factor above 100., how do you calculate? thanks.

- - - Updated - - -

and BradtheRad, what do you mean by 'If you're lucky there is a modern design which can make the job easier'? can you explain more?

 

[LvW]

hi, LvW.
1) the attenuation requirements apply to ALL three center frequencies.
2) i want to suppress noise at 120hz and 50khz as much as possible
3) gain requirement is not strict.
4) Q to be constant here.

Hi,

Your specification is not yet clear. What means "Q to be constant"? Do you have one stage (2nd order) in your mind? This is not realizable as mentioned before.
Do you know the meaning of Q for a multi-stage filter (Q=Fo/B)? B=pass bandwidth.
Are there no specific requirements regarding width of the passband? Just smaller than B=Fo/Q=Fo/10 ?

As a first step I have assumed B=Fo/Q=20k/10=2 kHz.
For a damping of approx. 70 dB at 50 kHz you need a 6th order bandpass (3 stages) - the pole Q of each stage is between 10 and 20.
For a center frequency of 38 kHz (the damping requirements) and a bandwidth of 3.8 kHz you need at least an 8th order filter (4 stages).

I suppose, the design will be VERY critical (due to the high pole Q values per stage)
But - in any case - because of several stages tuning for the 3 different mid frequencies will not be possible.

 

[sabouras]

it may be possible to do with passive approach. A bank filter with pin diodes which open and close the proper passive filter.