Vari-Q Bandpass Filter With Constant Gain

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Samuel Groner

Well-known member
Joined
Aug 19, 2004
Messages
2,940
Location
Zürich, Switzerland
Hi

I'm looking for a bandpass filter (2nd order) topology which has the following two features:

* orthogonal control of frequency and Q
* constant max. gain

So far I've only found topologies which fit either one or the other requirement. Any suggestions?

Thanks,
Samuel
 
I'm not clear about the terminology but if orthogonal means independent control of Q, freq, and gain, and 2 pole bandpass means one pole each for HP and LP skirt, the simple 3 opamp state variable parametric can deliver independent control of all three.

You can vary the center frequency by tweaking the R feeding the HP and LP integrators. You can vary the Q by skewing the Hp and LP resistors apart, or by tweaking the noise gain of the first opamp (HP output).

n122fig24.gif

here is an img I lifted from Rane website.  http://www.rane.com/note122.html

You can also tweak the bandwidth with a pot to ground from the - input of the same opamp, but it shifts the Q in the other direction, so depending on whether you want to tweak the bandwidth wider or narrower than 1 octave can direct where that pot is located. I have even put the pot across both inputs with the wiper to ground (through a resistor) to get both very wide and very narrow Q.    

Somewhere I have an old handout sheet from my kit company days with all the equations. A lot of my old kit customers were college students who needed to modify a circuit for their term project. I gave them a cheat sheet showing how to change it, because so many asked me the same questions.  

I hope this helps.

JR
 
I have few. Unfortunatelt I am very busy within the next two days but can scan/post some schems over the weekend.

I also have look up tables for component values for the topology John posted.
 
Thanks for your responses.

The simple 3 opamp state variable parametric can deliver independent control of all three.

Yes, but it does not fit the second requirement; the gain at the center frequency varies all over the place with Q. In the case of infinite Q (i.e. an oscillator) the bandpass gain is also infinite; for Q = 1 bandpass gain is unity, and for Q < 1 less than unity.

The dual amplifier bandpass topology keeps bandpass gain fixed at 2, but does not offer orthogonal f/Q control... It occured to me that maybe by using an all-pass filter things could work out. Will need to look into this.

Samuel
 
Yes you can vary the Q independently of the center frequency, perhaps not the exact image shown. Clearly if you vary the Q by shifting the two integrators apart that doesn't change the gain.

The simple Q control usually changes the gain of something (perhaps the HP and LP outputs). A common design flaw in parametric EQs IMO is headroom, as these other outputs can clip several dB before the direct path when higher than unity gain. This shows up as a subtle increase in distortion based on EQ settings, depending on the implementation. Some variable Q topologies just increase the noise gain of first stage so less (different) linearity issues. 

I have seen one commercial design where the Q control was a dual pot with one section compensating for the gain change (I'm not advocating that).

Note: the audio input to that filter can come into either the + or - input of the left most opamp. Imagine Vin from that drawing grounded and another similar value resistor feeding the audio input signal into the + input.  Now the bandpass output is inverting overall but BP gain doesn't change with the Q pot as drawn.

I have seen and used several variants on this common SVF.

JR
 
Good morning everyone, sorry to revisit such an old thread but it's very close to what I'm on the hunt for today. Unfortunately electronics isn't my first or even second language, so although the answer is probably already clearly stated I'm just not picking it up.

I also am looking for an audio bandpass circuit that has fixed gain but variable Fo and Q. It seems like it should be possible with some version of a state variable filter, but each that I've tried allows the gain to vary with Q. Ultimately I'm wanting to build a 2nd order all-pass filter with adjustable Fo and Q, but to do that using a 1-2BP method the BP has to be very consistent. I've built fixed Q and Fo versions of the all-pass already with great results, but would like the ability to tune by ear.

The only version I found that did come close was altering the gain on the first op-amp to compensate for the gain change with Q, but I can't come up with any way to do that with a ganged pot, and I don't know if digital pots would maintain the high level of sound quality that I'm looking for. I just don't know enough about them and what they can do. Here's that circuit (courtesy of www.next.gr):

135-8528.png


Does anyone have some circuits they can point me to that might be able to do this? Or if it's already been solved in this thread maybe you could push me a little in the right direction. Thanks in advance,

Chris
Kansas City
 
I'm mildly interested in EQ designs although I have little to no experience with them but does it have to be analog or can it be digital?
 
mojozoom said:
Does anyone have some circuits they can point me to that might be able to do this?
You have to do a small mod; inject the input signal to the non-inverting input of the leftmost opamp (instead of the inverting input) via a 10k resistor and ground the former input resistor. You will see that the boost is pretty much constant over a wide range of BW (notice I don't use the term "Q" deliberately).
 
efinque said:
I'm mildly interested in EQ designs although I have little to no experience with them but does it have to be analog or can it be digital?
Although it is conceivable to transform the SVF into DSP, it is not considered a good solution. The SVF relies heavily on "instant" NFB, which is not feasible in digital. The SVF is essentially an analog way of realizing bi-quads, for which there are easier ways to do  in digital.
 
abbey road d enfer said:
Although it is conceivable to transform the SVF into DSP, it is not considered a good solution. The SVF relies heavily on "instant" NFB, which is not feasible in digital. The SVF is essentially an analog way of realizing bi-quads, for which there are easier ways to do  in digital.

What's NFB?

Anyway, I've read there are SHARC processors (and other solutions) that could be used to do the task but they are expensive and complex (they require a bit of knowledge to configure).. CMOS, basically programmable logic/FPGA.. I've been tinkering around with the Arduino board and I think it could do something like this with a potentiometer (or two) in an analog input, although I don't know how to program it and I think the sound quality would be questionable.
 
mojozoom said:
Good morning everyone, sorry to revisit such an old thread but it's very close to what I'm on the hunt for today. Unfortunately electronics isn't my first or even second language, so although the answer is probably already clearly stated I'm just not picking it up.

I also am looking for an audio bandpass circuit that has fixed gain but variable Fo and Q. It seems like it should be possible with some version of a state variable filter, but each that I've tried allows the gain to vary with Q. Ultimately I'm wanting to build a 2nd order all-pass filter with adjustable Fo and Q, but to do that using a 1-2BP method the BP has to be very consistent. I've built fixed Q and Fo versions of the all-pass already with great results, but would like the ability to tune by ear.

The only version I found that did come close was altering the gain on the first op-amp to compensate for the gain change with Q, but I can't come up with any way to do that with a ganged pot, and I don't know if digital pots would maintain the high level of sound quality that I'm looking for. I just don't know enough about them and what they can do. Here's that circuit (courtesy of www.next.gr):

135-8528.png


Does anyone have some circuits they can point me to that might be able to do this? Or if it's already been solved in this thread maybe you could push me a little in the right direction. Thanks in advance,

Chris
Kansas City


Isn't this circuit similar to the NetEQ circuit?
Thanks!
Best,
Bruno2000
 
efinque said:
What's NFB?

Anyway, I've read there are SHARC processors (and other solutions) that could be used to do the task but they are expensive and complex (they require a bit of knowledge to configure).. CMOS, basically programmable logic/FPGA.. I've been tinkering around with the Arduino board and I think it could do something like this with a potentiometer (or two) in an analog input, although I don't know how to program it and I think the sound quality would be questionable.

As *bit* of knowledge is an understantement. Plus the tool chain isn't cheap.

An arduino is basically useless for this application unless you pair it with a good A/D and D/A. Even then, it is not an environment designed to run dsp code....

A better option is the sigmadsp from analog devices.

bruno2000 said:
Isn't this circuit similar to the NetEQ circuit?
Thanks!
Best,
Bruno2000

Yes. BUT most of the common SVF implementations all look pretty similar, the differences are in the subtle details, which I will let more expierence posters pontificate about.  ;D

efinque said:
It looks like a Baxandall circuit to me...

No, it's a state variable filter. The baxandall circuit is very different.
 
> ... ask Google first.

National Film Board of Canada
National Federation of the Blind
National Federation of Builders
No f***ing Bueno; When something is REALLY bad
National Research School in Business Economics and Administration (NFB)
NFB File Extension - What is an .nfb file and how do I open it?

Building a bad film for blind cellphone-using businesspersons?
 
That is a pretty common SVF (state variable filter). I published a parametric EQ kit article back in 1978, but I intentionally incorporated a first order interaction between bandwidth and boost/cut so wider bandwidth was compensated by less boost/cut for a more natural sounding EQ to complex final mixes.

Professional parametric EQ designs of mine did not interact on the ASSumption EQ was being applied to individual tracks or for surgical EQ.

I am unclear about your "second order all pass" terminology...  the schematic shown has conventional HP/BP/LP outputs.

I have done some work with DPOTs (digitally controlled pots) and their audio quality was certainly adequate. In fact with independent control of the two frequency resistors you could spread them apart to make wider Q.

Hope this helps

JR
 
Back
Top