[MOVED]Filter design with two center frequencies

[anex16]

Hi

I want to design a circuit to which I would feed in a sinusoidal signal of any frequency but always the same amplitude say 1V. The output should be 1 V for only two input frequencies (f1 and f2) and zero for any other input frequency. Can anyone suggest how I should go about it? How do I create such a sharp filter and that too with two peaks at f1 and f2 ?

Thanks

---------- Post added at 11:59 ---------- Previous post was at 11:44 ----------

What I can think of is to design two very sharp bandpass filters centered at f1 and f2 and then cascade them. Whats your take on this?

 

[keith1200rs]

Re: Filter design with two center frequencies

Sounds reasonable, but they need to be in parallel and the outputs summed, not cascaded.

Keith

 

[FvM]

Re: Filter design with two center frequencies

Real filters will have non-zero pass band width and finite stopp band attenuation. In so far, you'll need to supplement the problem formulation with a real specification in terms of pass and stopp band characteristic.

Technically, the problem is referring to two bandpasses with their outputs added.

 

[anex16]

Re: Filter design with two center frequencies

ah yes..they would need to be added not cascaded..thanks..any other way out?

 

[FvM]

Re: Filter design with two center frequencies

any other way out?

Out of what? The basic problem is simple and can be tackled straight forward. For a real implementation ... depends on your specification.

 

[anex16]

Re: Filter design with two center frequencies

I need to use only passive components.. I am not sure if I can get such a sharp response curve out of them.. am pretty new with this but I think active filters would be better suited for the application ?

 

[FvM]

Re: Filter design with two center frequencies

but I think active filters would be better suited for the application ?

Also the answer to this depends on the specification. Passive LC filters are particularly preferred for the high MHz range and almost unsuitable below 1 kHZ.

 

[permute]

Re: Filter design with two center frequencies

you can also use a simple fir filter.

the most basic method is to make a filter with coef's of cos(f1*t)+cos(f2*t).

this won't exactly meet the requirements if f1 and f2 are not related. This is reduced when the filter length is increased. increasing filter length also sets the selectivity.

windowing might also be useful. Linear algebra can also be used to find a filter that has a gain of 1 for two frequencies. such a filter would have a form a*sin(f1*t)+b*cos(f1*t)+c*sin(f2*t)+d*cos(f2*t).