'Constant Q' SVF parametric equalizer

[domingo]

Is there anyone aware of a model or design of state-variable filter that compensates Q when changing the gain (boost/cut)?
The design published by @JohnRoberts in Popular Electronics for example does something similar, it compensates gain when Q is changed. I think figure 3 here by Ethan Winer came to do the same with a dual gang pot for Q. But I'm wondering for the opposite. Rane calls it Constant-Q here, where changes in gain affect the 'skirt' (bandpass basal width) but not Q. The concept is opposed to Proportional-Q, where the skirt remains constant and Q changes. All PEQ designs I'm aware of are Proportional-Q. Constant-Q seems only present in graphic EQs. But I wonder how difficult can it be... in my Spice simulations I failed of course but that's me.
Anyways, I wonder if any clever compensation of that kind was ever tackled by someone–in the non-digital domain.

Cheers,
Domingo

 

[JohnRoberts]

Rane's constant Q is something else related to Graphic Equalizers. I published the parametric equations based on SVF in my 1979 kit article (i have them somewhere in my old papers).

To vary the Q with boost/cut I took advantage of the relationship between Q and bandpass gain that moved Q in the direction I wanted (narrower bandpass for higher boost/cut). FWIW I recall seeing one commercial parametric back in the 70s that used a dual pot to vary Q. One pot section increased the BP gain while the other pot section trimmed back the boost/cut. It was a popular EQ back then but the approach cost headroom when commanding narrow Q.

I only did this on consumer (hifi) equalizers operating on wideband signals. For professional applications I held the Q constant with boost/cut as EQ would be more targeted.

JR

 

[domingo]

Thanks a lot for your answer John. It makes me very happy to read the author himself.
I hope you don't mind, but I made a quick simulation of your circuit. I see that the gain is increased when Q increases (about 6dB). I can imagine this compensates the change in gain of a wider spectrum really nicely.
But I was referring to compensate Q when gain changes—instead of compensating gain when Q changes. A normal behaviour I would expect is that Q stays the same when boosting or cutting. In my simulation for example the maximum Q is around 0.2 octaves at maximum gain (20dB). But when the gain is reduced to 1/4 (5dB) Q changes to 1.5 octaves.

 

[domingo]

Ups, sorry, I think the right term for I was talking about is 'Constant Bandwidth'. SSL calls it that way in their E-Series, I'm coming to realise now.

 

[user 133392]

What about the Orban 642? See pdf page 19-20.
https://umlsrt.com/wp-content/uploads/Studio Documents/Orban_642B_EQ.pdf

 

[JohnRoberts]

OK I am repeating myself but I wasted a couple years lobbying the AES standards committee to come up with a sensible definition for Q/bandwidth in the context of Boost/cut equalizers. Obviously for small amounts of boost/cut the classic -3dB half power points do not apply.

Most instead use the Q/bandwidth of the underlying BP filter for BP filter based EQ topologies.

I lost energy and gave up years ago.

JR