Op amp variable high and low pass filters

[Rutherberg]

Hello,

I would like to design an audio bandpass filter, obviously composed of a low pass and high pass filter, both variable.
Nothing apparently difficult. My problem is that most of the time, each filter cutoff frequency is controlled by a stereo pot, like this http://sound.whsites.net/p155-f3.gif http://sound.whsites.net/p155-f1.gif

I would like to control each filter with a mono pot, and eventually change the filter slope with a switch.
I also would like to avoid OTAs.

I searched a lot, without success so far. First of all, is it achievable ?

[Zero999]

You need to provide more information.

What's the cut-off frequency? What sort of roll-off do you require? Does it need to have any gain?

A basic variable cut-off integrator (low pass) and differentiator (high pass) filter can be made with a single potentiometer and capacitor but the roll-off is only 20db/decade.

[magetoo]

Why mono pots?

Do you want to avoid OTAs specifically, or active variable gain elements more generally?  If the problem is just distortion from the OTAs, you could use VCAs instead.  I wouldn't be surprised if e.g. THAT had some app notes talking about just that.

Edit: Douglas Self has a single pot circuit in the second edition of "Small Signal Audio Design", pages 438-439, intended for a swept-mid in an EQ.  (Single reverse-log pot for frequency, plus a pot for level.)  It might be possible to tweak that to do what you want.

[Rutherberg]

What's the cut-off frequency? What sort of roll-off do you require? Does it need to have any gain?

A basic variable cut-off integrator (low pass) and differentiator (high pass) filter can be made with a single potentiometer and capacitor but the roll-off is only 20db/decade.

About cut-off frequencies, low pass settable from 1000 to 4000Hz and high pass 60 to 800Hz.
About the slope, -3dB/octave slope is not enough, -6 or -12 dB/octave.
Gain is optional, but if implemented, settable from 0.5 to 5.

Why mono pots?

Because more expensive, less elegant and there is a space constraint.

Do you want to avoid OTAs specifically, or active variable gain elements more generally?  If the problem is just distortion from the OTAs, you could use VCAs instead.  I wouldn't be surprised if e.g. THAT had some app notes talking about just that.

Again about the price, and I don't understand well OTAs (this shouldn't be an excuse, but well..). Distortion is not a problem. Anyway, I'll check THAT ICs.

Edit: Douglas Self has a single pot circuit in the second edition of "Small Signal Audio Design", pages 438-439, intended for a swept-mid in an EQ.  (Single reverse-log pot for frequency, plus a pot for level.)  It might be possible to tweak that to do what you want.

Thanks, I forgot about that book. We don't have the same edition, I find the circuit you mentionned pages 278-279.

[Kleinstein]

In the usual implementation a mono pot can control a 3db/oct filter, a 6 db/oct filter would need stereo. As stereo pots are still regular available I would tend to avoid the extra effort. However it gets more difficult for higher order. 12 dB/oct would need 4 pots. Also with higher quality factor (though likely not an issue here) the pots would need to be good quality to make them run in parallel.

For a large frequency range one might have to switch capacitors too. A pot usually only works well for something like 1 decade or a little more.

One option would be a switched capacitor filter: There are special filter chips (e.g. Maxim) for this - the filter frequency is controlled by a clock frequency.

If control does not need to analog, one could also use digital pots.

Today it would even be possible to go the digital way - so ADC - DSP/µC - DAC. This is especially true is a little delay is acceptable.

[Zero999]

One option would be a switched capacitor filter: There are special filter chips (e.g. Maxim) for this - the filter frequency is controlled by a clock frequency.

I agree. I was just about to suggest this but you beat me to it. A switched capacitor filter could easily be controlled using an astable multi-vibrator who's frequency can be adjusted with a single potentiometer. The only issue is you may need some additional filtering, to remove the clock but hopefully that will be outside the audio band anyway.

[DrGonzoDK]

You could try looking at voltage-controlled Sallen/Key filters in a bandpass configuration.

I've been doing some research into this so I would politely show you to https://www.eevblog.com/forum/projects/voltage-controlled-sallenkey-filter-using-jfets/

At the moment, either an OTA - or using reverse biased NPN transistors as current controlled resistors. You can set the control current using that mono pot and use it to control however many filter stages you desire.

Now, you could stage a Sallen/Key low-pass block with a high-pass block to get a -12db/oct resonant band pass filter; or, you could use a single filter block and get a -6db/oct band pass filter.

However, BP filters aren't very useful unless you can control the band width (i.e. the Q) - and you might want to consider that when wanting to create a "one big knob" filter.

[DrGonzoDK]

One option would be a switched capacitor filter: There are special filter chips (e.g. Maxim) for this - the filter frequency is controlled by a clock frequency.

I agree. I was just about to suggest this but you beat me to it. A switched capacitor filter could easily be controlled using an astable multi-vibrator who's frequency can be adjusted with a single potentiometer. The only issue is you may need some additional filtering, to remove the clock but hopefully that will be outside the audio band anyway.

If the desired application is audio, then switched-capacitor filters may be a bit of a dud. They don't sound terribly good - which is a totally subjective measurement, yes, but so much in audio is. They aren't used for audio stuff generally due to all sorts of intermodulation distortion (even if the switching freq is way up) and other funky artifacts that aren't immediately apparent.

Maybe they're acceptable for the OP's target application (of which we don't know anything of course), but you rarely see 'em in audio applications for a reason.

[danadak]

State variable one approach, requires a dual pot. I used one where integrator input R's were
replaced by R2R DACs, controlled by UP, worked great.



http://sound.whsites.net/articles/state-variable.htm



Regards, Dana.

[Rutherberg]

Thanks for all the inputs.

I would like to stay all analog, but it has limitations.

I've been doing some research into this so I would politely show you to https://www.eevblog.com/forum/projects/voltage-controlled-sallenkey-filter-using-jfets/

Yes I saw it, and read it with great interest 
BTW, where did you find your LM13700 spice model ?

I will explore all the possibilities during next weeks.

Thanks again !

[DrGonzoDK]

Thanks for all the inputs.

I would like to stay all analog, but it has limitations.

I've been doing some research into this so I would politely show you to https://www.eevblog.com/forum/projects/voltage-controlled-sallenkey-filter-using-jfets/

Yes I saw it, and read it with great interest 
BTW, where did you find your LM13700 spice model ?

I will explore all the possibilities during next weeks.

Thanks again !

The LM13700 model is on TI's web page - http://www.ti.com/product/LM13700/toolssoftware (it's the PSPICE model). It's a pretty good simulation, too!

[DrGonzoDK]

Thanks for all the inputs.

I would like to stay all analog, but it has limitations.

I've been doing some research into this so I would politely show you to https://www.eevblog.com/forum/projects/voltage-controlled-sallenkey-filter-using-jfets/

Yes I saw it, and read it with great interest 
BTW, where did you find your LM13700 spice model ?

I will explore all the possibilities during next weeks.

Thanks again !

Just a quick aside - currently building the photoresistor+LED-controlled Sallen/Key style filter (see attachment). It's quite elegant (even though the "voltage controlled resistors" up top aren't too pretty). The resistance of both photoresistors are about 1 megaohm when no light is on; and down to about 150 ohm with maximum intensity from the LED's. I dremeled the LED's down so they were flat, and superglued them to the photoresistor surface; hot-snotted them, painted this black, and added two layers of heatshrink insulation.

At the moment, I'm just wiring up the filter opamp (TL061CP in the middle); the voltage buffer for the voltage control is in place (cheap ass 100KHz STMicro opamp up left), as is the negative rail charge pump (a standard 7660 since I'm not gonna need more than 10-20 milliamps).

I tested the LED-photoresistor voltage control with just a simple passive RC filter, and it worked beautifully. The only important thing is to tweak the LED resistance - and thereby, the voltage control slope. That is still ongoing, but, pretty cool. And, of course, I threw in an NPN LED driver after the voltage buffer to allow for more current output than the 10 cent opamp can handle...

EDIT: Also - the photoresistor/LED combo has a bandwidth of about 100 to 200 Hz, so if you need quicker cutoff modulation, this is not a great solution. Also, it consumes more current than other solutions, but I just loved the Rube Goldberg style of it, so had to build at least one. The mounted potentiometer is an optionally connectable (single female pin) voltage controller; alternatively, you can connect an external source. The input is going to be a single RCA input - so I still gotta drill the holes for that one. But, as I said, I love the sort of quirky idea behind it.