Rules of thumb for gyrators, or better sweepable mid control?

[ElectricDruid]

Hi All,

Today I've been messing with LTSpice and a gyrator-based mid EQ design. This works fine, but I'm just plugging values in and seeing what happens. The design is far from ideal because of the variable Q that you get with sweepable gyrators. Are there rules of thumb that give a simple guide to the final boost/cut, or the final range of Q in this type of circuit? Is it possible to at least minimise the variability of Q across the frequency range? What are the side-effects of that?

The type of circuit I'm talking about is this one (From R.G.s site):



I know from having studied the Boss MT-2 that Roland's engineers decided it wasn't worth it for a mid control and adopted a Wien-bridge design instead (shown below). That fixed the variable-Q problem, but then requires a dual pot to control frequency.



Another (yet more complicated) solution would be a state variable filter, but again it's a dual pot for frequency control.

Does anyone have any better ideas or other possibilities, or am I asking for the moon on a plate again?...

Thanks,
Tom

[ElectricDruid]

I've managed to answer one of my questions. Doing some searches I came across this:

https://www.feucht.us/writings/graphic_equalizer.pdf

which gives an equation for the Q in a gyrator:

Q = 1 / (C1+C2) * sqrt((R2*C1*C2) / R1)

So basically, the Q changes as the square root of the change in R2. So you can't change the frequency without altering the Q, there's no getting away from it.

R.G.'s page on these things suggests that the amount the frequency alters is also the square root of the change in R2, so that'd make the Q and the Frequency directly proportional. Boo!

http://www.geofex.com/Article_Folders/EQs/paramet.htm

OK, still looking.

[PRR]

> you can't change the frequency without altering the Q

A naked gyrator, probably not.

You can make ALL parameters independent with more op-amps. Lancaster's Active Filter Cookbook has background. RANE had papers on mature EQ schemes.

[marcelomd]

I'm guessing you already checked ESP.

If not: http://sound.whsites.net/articles/gyrator-filters.htm.

IMHO, for what we do, there is no need to be surgically precise about Q.

[samhay]

>I know from having studied the Boss MT-2 that Roland's engineers decided it wasn't worth it for a mid control and adopted a Wien-bridge design instead (shown below). That fixed the variable-Q problem, but then requires a dual pot to control frequency.

I quite like the Wein bridge topology despite needing a dual gang pot.
If you haven't seen it, Rod/ESP has some description of these too:
http://sound.whsites.net/project150.htm

[ElectricDruid]

Ok, here's a couple of things I've got so far. I've been aiming for something that does a bit over an octave either side of 800Hz, and it should use easy pot values. The gyrator uses 1M, the MFB uses 10K.

Here's the gyrator circuit:



and here's its response:



And then I tried an alternative Multiple-Feedback Filter design:



Here's the response for that one:



Now the really interesting thing that jumps out for me is that one of them increases Q as the frequency goes down, and the other one increases it as the frequency goes up! That's pretty different, and is going to affect the sound significantly.