Hi all,

I need to design a bandpass filter using a Tow-Thomas biquad. The filter must work with center frequency at 1000Hz, and bandwidth of 100Hz.
Using a resistor and a switch, I must be able to shift the center frequency over to 1300Hz, but keep the bandwidth the same.

I can design them separately, but I can't seem to be able to find a resistor position that will allow for switch from one to the other.

This is the circuit: http://en.wikipedia.org/wiki/File:BiquadFilter1.PNG
I tried putting parallel resistors across R1, R3, R4. I also tried changing gain of the inverter at the end. I could try doing a feedforward variation, but that will give me a 2nd order transfer function in the numerator, which I don't need anyway (so I don't think it will work).

Any hints will be appreciated!

 

http://www.google.com/#hl=en&output=search&sclient=psy-ab&q=%2B%22Tow-Thomas+biquad%22&oq=%2B%22Tow-Thomas+biquad%22&gs_l=hp.3...1047.4110.0.5547.4.4.0.0.0.0.188.640.0j4.4.0.les%3B..0.0...1c.1.z6H2-etpkMc&pbx=1&bav=on.2,or.r_gc.r_pw.r_qf.&fp=3fc67df3026004fb&biw=1171&bih=941
might be of assistance.

 

I don't see how that is helpful. Most of the links are already purple, because I have looked through it before posting here. I understand how to make a filter centered at one frequency using the biquad.

I need to make two center frequencies realizable with one circuit, an extra resistor and a switch.

 

Did they mention the two frequencies you need for center?

 

Yes, this is mentioned in my original post. Nevertheless, I will reiterate:

Center frequencies: 1000Hz and 1300Hz. Bandwidth must stay the same at 100Hz.

 

Yes, this is mentioned in my original post. Nevertheless, I will reiterate:

Center frequencies: 1000Hz and 1300Hz. Bandwidth must stay the same at 100Hz.

How tight a spec is there on any bandwidth changes - ±1% .... ±0.5% ...??

 

I don't see how that is helpful.

In one of the references listed by Joe, it gives the frequency as a function of the resistors. You can "vary" one of the resistors just by connecting a second resistor around it and switching it into parallel. That should vary the center frequency. See below:

http://en.wikipedia.org/wiki/Electronic_filter_topology

 

In one of the references listed by Joe, it gives the frequency as a function of the resistors. You can "vary" one of the resistors just by connecting a second resistor around it and switching it into parallel. That should vary the center frequency. See below:

http://en.wikipedia.org/wiki/Electronic_filter_topology

Problem is that [as the link also indicates] the Q and hence bandwidth also change as either R2 or R4 are changed to vary the center frequency. The OP's question states that the bandwidth must remain the same at 100Hz. That's why I asked whether there was some "slack" on this requirement.

 

If we use the approximation for bandwidth, B = ωo/Q mentioned in the Wikipedia article, the bandwidth is given by B = 1/(R3*C1). Hence, changing R2 or R4 will change ωo, but if we don't change R3 or C1 the bandwidth won't change.

 

Good point Electrician,

I thought I'd checked the ratio ω0/Q but messed up the math.

I did a simulation based on an ideal op amps. It works fine for changing only R2 say. The difference in B is very small within the limitation of picking the -3dB point on an output graph.

This should re-assure the OP it does work.

 

I did a simulation with E24 capacitors and E96 resistors and using OPA627E opamps. It indeed does work

Below is the graph ...