Discrete peaking/bell filter, fixed frequency and Q

[fryingpan]

I'd like to make a single band peaking filter using only discrete components. I don't need to be able to sweep the frequency, my design calls for a specific frequency (2kHz) and a single Q (0.70). On the treble side, actually, I'd like to add a treble shelf, but that can be done quite easily. Anyway, all the designs I've seen call for opamps (and they're much too powerful, I don't need all those features). I'd like to implement it with BJTs or FETs. Any ideas?

[fryingpan]

For example, I found this design:

https://qph.fs.quoracdn.net/main-qimg-39a1f20c8f373c9a1660d1bd0c816556-pjlq

which gives basically the kind of response I want, an octave lower, but are there any alternatives (maybe which employs BJTs instead of FETs)?

[Digital Larry]

Those FETs are running as source followers (no voltage gain).  They could be replaced by BJTs as emitter followers but you'll have to provide a DC bias current to the base.  You may also have to adjust component values to operate at a lower impedance while keeping the frequencies the same. 

It's at this point where I get off the bus and let someone with better analog skillz take over.

[fryingpan]

After simulating that circuit, I decided it's no good for audio purposes. It doesn't work as is (you have to add a coupling capacitor between the two FETs). And weirdly, it only provides the stated response (although with a very "linear" shape) if you feedback the output to the input, and accept -20dB attenuation, otherwise it's steep on the bass side and shallow on the treble side (which could work as a bell filter, mixing it with the clean sound, if the phase shift weren't too high and all over the place).

[iainpunk]

it needs 4.5v bias, and the other things like power supply decoupling to work well, just like an opamp needs.



this has a Q of 0.74

cheers

[fryingpan]

Great, thanks!

In the meantime I was fooling around with this bandpass filter:



and by varying R28 (the resistor in series with the bypass capacitor on the first transistor) you can effectively vary the filter's Q (the smaller the sharper, because of course, there's more gain and more feedback)... but I can't figure out how to mix it with the clean signal.