I’ve been playing around with some ideas for a VCA build and looking at different topologies and methods. Along the way I had the question of whether to do an exponential or linear circuit (or switchable) and was playing around with different exponential setups.
At one point I thought it was getting too much for my simple VCA, so I thought I’d scrap the exponential part. But then I thought “why not just make a module that takes a linear CV and outputs an exponential one?”
Hopefully someone on here could have a look and play (or tell me why it is a stupid idea).
It currently has input and output offsets, an “exponent” adjustment (really just a multiplier on the mixed inputs) and a normal and inverted output.
I thought it might also be useful as a utility module for interfacing with non V/oct gear, but might have to build a reverse version that outputs a log function for that instead.
Slightly updated schematic. Same circuit, just made the visualisation clearer. Also renamed the “exponent” adjustment to “multiplier” as that is more correct anyway.
Actually I had been working on mine a while ago (probably from similar sources) and his video prompted me to revisit it!
Great content he puts out. Even though I try to tweak things from various sources to make my “own” circuits (in so much as I understand what I’m putting together, not in terms of me developing anything novel) I really get a lot from following his logical building up from smaller parts.
From what I’ve seen what people usually do with exponential converters is turn a linear voltage into an exponential current. The reason this is done is because with a current you can cover a range from nA → mA, which is three orders of magnitude. I think that with an exponential voltage you’re going to run into trouble keeping within the supply rails of typical DIY synth circuits while at the same time having usable resolution for lower octaves/levels. An example would be maximum CV out @ 10V for a maximum input of 10V. For the lowest CV out octave, i.e. the linear input range from 0 to 1V, you’d be working from 0 to 20mV out or something along those lines, which doesn’t give much resolution.
I spent some time trying to understand and simulate exponential converters, I’ll try and dig up my annotated schematic later this evening when I get off my work PC and share it here.
That’s a great point actually! Running a few millivolts out through some long patch cables isn’t such a great idea. Still, it was a fun challenge to give the flexibility to the circuit that a hypothetical module would require. Might still be useful to someone for some less critical aspect such as taking an adsr and giving it exponential decay for instance. Not so critical to be exact then.
Let me know if you have any questions, I’ll do my best to answer, but must admit that these circuits are some of the hardest for me to get my head around.
Yeah - R8, connected to C2 is only really for the simulation, so I could measure current through a resistor.
C2 is where you’d hook up your load in a real circuit - like an integrator for a saw or triangle oscillator to control frequency, or into a diode ladder to control cutoff or something like that.
C1, on the other hand, is connected to the virtual ground that the op-amp will try and maintain on its inverting input. There should be 12V across R3 which will then set the reference current Iref. In this case it’s 12/3M = 4µA. You’d probably adjust this depending on the application. Remember, an ideal op-amps input impedance is infinite, so any current through R3 will be forced through C1, creating the reference current that will be sourced through C2 when B1 is equal to B2, i.e. 0V.
That’s where R2 comes in - you can change the voltage at B1 when the CV in (V_Oct) is equal to 0. In my VCO where I have this, R2 is a trimpot.
Ah, of course! The VCO I built I ended up going with a PNP based exp. converter (to drive the current of an LM13700), but I find the NPN based ones easier to understand for some reason.
