Narrow Bandpass Filter

[jbushnel]

I wanted to build three narrow bandpass filters using the MFB topology attached below. I am trying to separate 3 different tones that are 225Hz apart each. I think this is do-able, I just need to know what quality factor would be sufficient? Maybe 30? I know the frequencies of each and they don't change so I can get away with a high quality factor. Any advice on how to approach choosing the right quality factor will help!

Also, I need help understanding the gains of these filters... Here is a link to a filter design website and I am not understanding the units:

http://www.analog.com/designtools/en/filterwizard/#/specifications

The drop-down menu shows "V/V," I have also seen "Vu." How does this relate to dB?

Thanks!

[T3sl4co1l]

Yeh... 225Hz apart at what frequency band?  It's a lot harder if the fundamental is 225 +/- 0.000225 MHz, and a lot easier if it's 225 +/- 225Hz.

If this is for a digital decoding thing, a LM567 tone decoder (or, I think that's deprecated now, and there's something better to replace it with) might be better.  Or even more application-specific, perhaps there's a pre-made chip, like a DTMF decoder.

There may be other methods, depending on the nature of the signals, phase or harmonic relation if any, etc.  Examples that come to mind: synchronous sampler / peak detector; PLLs (basically, what a tone decoder is*); PLL with frequency multipliers and mixers; tec.

*A tone decoder is basically a PLL with a 'lock' detect signal.  In turn, a PLL is basically a bandpass filter with certain characteristics that make it easier to use, or more powerful, for some purposes.

Tim

[tggzzz]

If you want a really high-Q filter, there's always commutating capacitor filters. I've built Q=4000 filters with 10% components

[LvW]

Perhaps it is wise to consider also other bandpass topologies than MFB?
One reason is, for example, that this structure (EDIT: for large Q values)exhibits a rather large component spread - if compared with some other realization alternatives. 
There are other filter design packages allowing different topologies.

[jbushnel]

The frequencies I want to detect are about 3.5kHz, 3.75kHz and 4.0kHz. Nothing as complicated as the MHz range...I am glad you brought up the LM567 though! I was going to use one of those for each frequency but I was not getting the right results. (I have attached the schematic below) I obviously changed the resistor value to the 3K range to get in my frequency range but here is what has been happening... I can get the LED to light up when I am around the tone but It seems as though the output is not stable when other frequencies are present. I am not sure why...

For example, I started to snap into the microphone when the tones occurred(it was on repeat) and my snapping would cut the output signal short and, if I timed my snap right, the LED would not even come on! I honestly don't think it should be doing that. The tone decoder should reject all other noises right?

So I thought if I built three narrow bandpass filters, I could eliminate this problem and get a nice output of the tone decoder for each tone. But if you guys don' think I need to filter everything out, I'm fine with that!

[nctnico]

Loud noises could overdrive the filter input signal. However I'd use a small ARM controller and the Goertzel algorithm or a bandpass filter + frequency counter algorithm depending on how accurate the tones need to be detected.