cascaded filter design

[ucah]

You can passivelly cascade the three RC sections if you do something for the interaction between them. One way is to have each resistor value be ten times larger than the previous one; the capacitors values are shaped accordingly.
You will of course end up with x100 output resistance, and you need to simulate using e.g. LTSpice.

 

[LvW]

You can make a second-order Sallen and Key highpass or lowpass active filter with equal values for the two R's and the two C's. The Q can be adjusted by changing the gain of the opamp. If the Q is higher than 1.6 then the circuit will have a peak in its response near the cutoff frequency.

Right, you can make .......with equal values for R's and C's, but you shoudn't !
That's - I think - the worst (but the simplest) design as it is very, very sensitive against part tolerances.

Audioguru, are sure that a peak occurs if the Q values are above 1.6 ?
Up to now I was of the opinion that this limit is Q=o.707 (Butterworth).
A pole Q=1.6 leads to a Chebyshev response with a peak more than 3 dB.

 

[keith1200rs]

If you don't mind the 6dB loss you can always do passive LC filter, although care would need to be taken to allow for the series resistance of the inductor.

Keith.

 

[Audioguru]

I made a bass-boost circuit using an equal-values Sallen and Key highpass filter with too much positive feedback.

I also made a 3.5kHz peaking circuit for a teleconferencing system using an equal-values Sallen and Key lowpass filter with too much positive feedback.

 

[FvM]

Up to now I was of the opinion that this limit is Q=0.707 (Butterworth).

Of course, if you understand "peaking" as any kind of non-monotonous amplitude characteristic. So any 2nd order Cheybyshev filter (Q > 0.707) shows peaking. For higher order filters, there's no simple criterion of complex pole Q, because the combination matters for the overall characteristic.

If you don't mind the 6dB loss you can always do passive LC filter

You can actually design a "0" dB loss passive filter, by either using zero source or infinite load impedance.

 

[keith1200rs]

If you don't mind the 6dB loss you can always do passive LC filter

You can actually design a "0" dB loss passive filter, by either using zero source or infinite load impedance.

Yes, I realise that, although I often find that even a 1:10 ratio of source to load resistance results in problems - either impractical values or inductor series resistance exceeding the source resistance, or excessive sensitivity to component tolerances. Depending on the frequency I usually find practical values end up between 100 ohms and 1k with a 1:1 ratio. It depends on the frequency and other parameters though, I guess.

Keith.

 

[LvW]

More than that, a double terminated passive LC filter has the optimum (that means: the least) sensitivity to parts tolerances.