Svf control voltage

[Kipper4]

I would like to cv an op amp svf
I'm using a cd4093 Lfo pics later I'm at work right now.
How do I hook the two up so the Lfo volts cv's the svf
More later.
Cheers

[Fender3D]

svf = Stevie Vai Faughan??

[lapsteelman]

State Variable Filter

[lapsteelman]

Isn't the usual way to control an op-amp SVF to use photo resistors and an an led.... like some of the Mutron effects?..

I just completed a voltage controlled SVF, but I went with a LM13600.

[Kipper4]

You wait till Stephens see that fender . Lol

Normally I'd use opto too but id rather do it another way.
Trying to avoid opto.
And ota filters too.

[Kipper4]

What about a jfet method?

[R.G.]

There are some pre-canned versions of the SVF made using the LM13700 OTA. You feed this a >>current<< to set the frequency, which can easily vary over a range of several hundred to one. The control node of the LM13700 must and will sit at a voltage of two diode drops (1.0-1.4V), so it's handiest to feed it from a PNP current mirror at V+, and suck current from the mirror input to ground. Easier to set up cleanly. But a simply voltage input through a resistor to the OTA control input will work. You just have to never let it go below two diode drops, nor high enough to make the current be more than 1ma, which will kill the ota.

... er, did I say all that out loud?? 

[Kipper4]

... er, did I say all that out loud?? 

'fraid so.

I've killed more ota's on the breadboard than I have used in boxed builds.

I've been having a run of good luck this week though I managed to cludge together an ota VCA (lm3080) I started out with an lm13700 that went well too.
Thanks to a handy dandy Geofx page.

http://www.geofex.com/Article_Folders/VCA%20Applications.pdf

I even managed to get the series resistor down to a 4k7 without killing it.

I've yet to do some current mirror experiments on the breadboard. It's inevitable since I ordered a couple of hundred BC547's.  Ooops looks like I need some PNP now too.
What better time to hand match transistors than on a cold winters night in a nice warm place listening to some tunes ?
Thanks for that R.G. and everyone.

I'll go do some reading.

Lapsteelman. I could go the vactrol route. In fact it would be an easy option to use some of the breadboarded indicator leds placements. The thing is my boards getting a bit crowded right now
(2x 14pins 1x 8pin about 10 pots 5 switches)
I've got a nice svf from a previous project (vactrol based wah) I'd like to use (TL074).
So I got to thinking of ways to get round having to use yet another chip.
I had a slow start to the work day so i thought I'd pick you guys brains.
All that soon changed when one of the other old guys couldnt cope with what he was asked to do, So muggings came to the rescue and we swapped jobs for the night.



Sod it you may as well see the scruffy drawing.



The Escabedo VCF is capable of some nice sounds. Some great formant sounds. Unfortunately I didnt catch them on the demo IIRC.
I might well stick with it If I cant manage to use the svf. (Harold Sabratone's version)



Edit;
I'd like to have the possibilty to hook the Modulation lfo up to the VCF as well as the AR out.
I've reposted the drawing

[Kipper4]

Something like this for simplicity?
Some values may vary.

[Fender3D]

You wait till Stephens see that fender . Lol

I thought you were looking for a new logo for your strat's pickguard 

[ElectricDruid]

Why so determined to avoid the venerable and trustworthy LM13700 SVF?

I mean, it has a few faults like most circuits, but it works and its many variations are widespread. You can have a simple version with no input mixer which does just BP and LP, or the full version with a mixer which can do HP, BP and LP. You can use the cr*ppy on-chip darlington buffers with a cap to ground, or you can upgrade to external op-amp integrator stages.

My biggest problem with it is control feedthrough, which I've never been able to get low enough. This isn't a problem for low frequency modulations, but if you whack the control input with something spiky or at an audio rate, you'll hear it on the output, even with no input. If anyone's got any clues on that one, I'm all ears...

Otherwise, it's a great circuit any way you do it. It even distorts nicely, especially if you ignore the "linearisation" diodes and stick with the native tanh response - trading distortion to improve S/N.

Tom

[Kipper4]

Well I tried the bjt idea and it had the effect of turning the svf into an vca.
No go.

I tried the tl074 svf with a vactrol from AR out and lfo out and it works well. vactrol has a variable CLR. Some more options there too. Reso.  Another vactrol (or variable R) for the lpf.


I've had a look at the options for an LM13700 vcf. I'm not sure If I'm gonna go with the mixer yet.
I want to hear it without and make further desicions based on that.

Heres my redraw of one of Ray Wilsons filters.
As yet untested on the breadboard.

I put the 2x    10k > +9v (pins 2 & 15) Just cause I read somewhere it might help with distortion.
Not sure If they're needed....




What do we think?

[diffeq]

Looks about right to me. There is another option: MS-20 VCF variant, based on LM13700. Just as simple (if not simpler) schematic. Quick search reveals this schematic:
http://www.diale.org/tpe123/sk.gif

By the way, what software do you employ for such neat schematics, Kipper?

[ElectricDruid]

I put the 2x    10k > +9v (pins 2 & 15) Just cause I read somewhere it might help with distortion.
Not sure If they're needed....

They're not "needed" in the sense that it'll work without them. They're intended to provide bias to the pre-distortion diodes. These provide an "anti distortion" which helps compensate for the nonlinear response of the OTA and allows you to put a larger signal level in before you hit 5% or 10% THD. So far, so good.
The downside is that when you *do* hit distortion, it comes on much more suddenly and harshly. Whereas without them, you get a much softer curve (tanh, in theory). So it depends what you're after. If you want maximum clean signal level and best S/N, the resistors are a good idea. For better sounding overdrive characteristics and more "vintage character" in general, better leave them out. Or at least, that's my view of it, for what it's worth.

Tom

[PRR]

> I've killed more ota's on the breadboard

The usual way to do this is to put more current into the Iabc pin than it can stand. Unlike nearly any other chip pin, the Iabc pin is Max 2mA.

For 9V work: Nail a 3.3K resistor to the Iabc pin and do NOT go near the actual pin, only futz-around with the other end of the 3.3K.

(9V-1.2V= 7.8V, 7.8V/3.3K is 2.36mA, and maybe tolerable 99% of the time.) (For 9.3V wart and '3080, 4.35K or 4.7K is a shade safer.)

This slightly limits your upper gain when added to a proper control circuit. But once you pooch the Iabc pin, there's no gain at all, so it is a win.

[Kipper4]

Brilliant thanks guys
So the 1v2 is my double diode drop.

[ElectricDruid]

Exactly. You lose 1.2V from the double diode drop.

T.

[Kipper4]

"By the way, what software do you employ for such neat schematics, Kipper?"

that's eagle cad my friend and thanks.

[R.G.]

Lurking behind a lot of this is exactly HOW you drive the Iabc pin.

It's one reason I like a PNP current mirror from the + supply. The PNP approach lets the current mirror make the exact voltage on the Iabc pin not matter at all, as the PNP takes up the whole of the power supply >voltage< variation. This produces a very linear sweep. It also lets you control the current very well from an LFO >voltage< with a single opamp and NPN transistor.

And you get right about to there (well, OK, I do  ) and remember that things that deal with audio loudness and frequency sound more linear to the human ear if they're actually exponential. That's why volume pots are most natural feeling if they're log taper.

The actual voltage on Iabc is not fixed. It's two diode drops, which is logarithmic compared to the incoming current. It actually varies from zero to burn-out at 2ma as the logarithm of the applied current. Looked at backwards, the current going in there is exponential with the incoming voltage, which is handy in doing volume and frequency things. If you can drive the Iabc input with a linear VOLTAGE that never lets the current get higher than 2ma (for the LM13700), you get exponential change with a linear LFO voltage. And you never have to worry about that two-diode-drop offset from zero in generating your current into the node.

It is possible to design a drive circuit that uses an opamp or two to translate a linear LFO voltage into a small linear voltage across the Iabc pin and ground that causes an exponential current to flow, and give you the exponential movement in the OTA that you'd really like. The real trick in this is to make the current resulting from the linear control voltage top out at just under 2ma.

And of course, the demon of complexity rears its ugly head right about here. The circuit to do this requires a couple of opamps and one or two transistors, and so is as complex or more so than the OTA you're trying to control. It's not all that bad, but adding an opamp and some transistors seems to be a big stumbling block to many people. It's probably better suited to synth hackers than guitar pedal hackers.

The bane of LFO anything to many pedal makers is that the idea of an LFO changing some feature is easy to get as a concept, but every LFO voltage or current needs adaptation to the circuit being controlled. The circuit being controlled nearly always has special needs in terms of minimum DC voltage and peak DC voltage, idle DC voltage, and funny current needs to get the voltage you're trying to get to work. It's like one of my favorite lines from Peter Sellers in his Charlie Chan imitation role in "Murder by Death" - "Answer easy - question, very hard."   

[PRR]

> 1v2 is my double diode drop.

No. It is the input to an internal Current Mirror.

The '3080 used a basic 1-Q 1-D current mirror.

The '13700 designers were breaking-in a new layout person, working at a time when added transistors cost almost nothing. They shifted to the slightly more 'accurate' 2-Q current mirror, which by then was all-over any analog chip, good practice for the new kid.