High pass sallen key filter

JingleJoe

A few weeks ago I started work on two sallen key filters for a device I'm building, a high pass and low pass, however I had some problems with the high pass filter. The low pass filter worked great, I used the standard arrangement with an op amp but adapted it for single supply use and added divided feedback to the inverting input for adjustable resonance (acctually it changes the filter type to bessel and chebyshev). However I found that the same method could not be implimented in the high pass filter to acheive resonance. Why not? Sad

it was so simple and effective! Damn you high pass filter! Can anyone reveal the mystery of resonant high pass filters?

Here is a diagram of the circuit I used, except R1 went to virtual ground (for single supply operation) and divided feedback (like this) was supplied to the inverting input via a pot and resistor, both 100k. This allowed for gain between 1 (unity, output shorted to inverting input to make it into a voltage follower) and 2, gain being approx 1+(Rf/Rin), Rf was the pot in the feedback path and Rin went from inverting input to virtual ground.

I gave up experimenting with this at the time because I had a good low pass filter working and other things to work on, but now the time has come for a high pass filter.

The Real MC · Joined: Jun 20, 2008 Posts: 62 Location: Painted Post NY

Check this wikipedia page

http://en.wikipedia.org/wiki/Sallen%E2%80%93Key_topology

If you study the formulas for Q, the sum of the resistors in the denominator is a factor in the LP but only one resistor in the denominator is a factor for HP. You may be scaling the wrong resistor.

The only resonant highpass I have seen in a synth is the SVF - the SEM and A6 had it. Voyager and Moog 904B has HP which is derived from simple manipulation of the 24dB ladder filter but resonance is disabled, not sure why.

JingleJoe · Posted: Tue Jul 31, 2012 12:44 pm Post subject:

Ah interesting, thankyou, I allways get put off by those transfer functions with lots of symbols I don't recognise Laughing

However I see what you're getting at there, in the high pass filter, the value of R2 (according to denotion in this diagram) is half R1, however if I have you right, altering that value will change the resonance.
By the same token, perhaps altering the other resistor will change the Q too? Because (I'm going to assume) certain ratios of the resistors will produce higher or lower Q ...
Experiments await Smile

Psyingo · Joined: Jun 11, 2009 Posts: 248 Location: Canada

Have you tried the synthacon style sallenkey? It has bpf, lpf and hpf in the same circuit.

Posted Image, might have been reduced in size. Click Image to view fullscreen.

Posted Image, might have been reduced in size. Click Image to view fullscreen.

No need to build two circuits. Unless you need hpf and lpf separately tunable, but you could still omit a bunch of those components related to the bpf and lpf for a resonant hpf.

JingleJoe · Posted: Tue Jul 31, 2012 5:19 pm Post subject:

Yes, however doesn't that one require a dual supply?

Psyingo · Joined: Jun 11, 2009 Posts: 248 Location: Canada

JingleJoe · Posted: Wed Aug 01, 2012 4:02 am Post subject:

I can remember this one was mentioned in the same thead in which LDR control was discussed, I couldn't locate that thread but nevertheless I've got a good working low pass now, so I'll stick with that and a separate high pass for the moment. Also this allows one separate control over cutoff of the high and low pass filters, whereas both are controlled in one filter there.
Anyways long story short, I'll experiment with that one another time if it works from a single supply.
I am estimating your bias additions were perhaps to R1 and R8 and maybe to the input signal too?

isak · Posted: Sat Aug 18, 2012 5:59 am Post subject:

JingleJoe · Posted: Sat Aug 18, 2012 6:44 am Post subject:

A Dual gang pot is the solution, altering the ratio of those two resistors will change the type of the filter. (butterworth to bessel to chebyshev)

Cynosure · Posted: Sat Aug 18, 2012 8:24 am Post subject:

I think the lm13700 can be used too as a voltage controlled resistor. Psy had linked me to some schematics in chat, but I don't have them now.

JovianPyx · Posted: Sat Aug 18, 2012 8:53 am Post subject:

isak · Posted: Sat Aug 18, 2012 10:27 am Post subject:

isak · Posted: Sat Aug 18, 2012 1:41 pm Post subject:

JovianPyx · Posted: Sat Aug 18, 2012 6:25 pm Post subject:

Not really a combination of state variable and Sallen-Key, they are two different filter cores. It may use similar techniques (subtracting one signal from another) to get a desired response, but that's where the similarity ends.

Look at the example circuits in the data sheet for a LM13700. It has a state variable filter that is setup for only lowpass. That is the core of the state variable filter, all that needs to be added for multimode is an opamp package that does the subtracting to get highpass, bandpass and band reject. You can see a full multimode schematic for state variable filters by using google on "state variable filter" - the wikipedia article is quite good.

In the case of Sallen-Key, it's a completely different core. While they are both 2 pole filters, they are very different. The Sallen-Key can be built from one opamp, the State Variable cannot. As previously mentioned as a Sallen-Key disadvantage, the Q of a Sallen-Key filter can't be pushed higher than about 3 without causing problems. A state variable will allow Q to be set much much higher. A state variable filter has an Fc input that does not affect Q.

One of the main things Sallen-Key is used for is when the filter doesn't need much Q and has fixed Fc (yes you can make it varaible, but it's just not very good) and it's an inexpensive product. It is not a particularly good synthesizer filter because of it's limited Q and I believe that changing Fc also changes the Q.

JingleJoe · Posted: Sun Aug 19, 2012 5:25 am Post subject: