4072 VCF bleedthrough fix with VCA?

[minisystem]

Like gdavis, I'm adapting a Yusynth filter design for ultimate use in a polysynth project, the 4072 VCF. After much effort and troubleshooting (and lots of advice from the kind people here), I managed to get a nice 0-10V voltage controlled resonance response by replacing the resonance potentiometer with a LM13700-based VCA.

I get the oft-reported bleedthrough at any resonance setting above '0' with R43 present, so have omitted it from my subsequent prototypes. I'd like to compensate for the low frequency loss at high resonance and thought I'd try to implement a VCA-based equivalent of the dual-ganged pot method that Yves describes in this thread. The idea is that the compensating feedback through R43 increases through the VCA with the resonance CV. I naively thought that this could be done by simply replacing the connection between R43 and the inverting input of the output stage buffer (U2b) with another LM13700-based VCA. In LTSpice simulation (where the model behaves very much like the prototype), this just doesn't work and even results in more signal loss than if R43 is absent!

I'm getting close to giving up, but wanted to see if anyone here had any insight. Clearly, I'm not thinking about this the right way.

[gdavis]

Seems to me you should be able to make it work. Through the spice up here so we can take a look.

[minisystem]

Of course! Here it is (zip archive should contain necessary libraries and sub-circuits to run)

'4072-LM13700-GARY-2' is the simulation with VC resonance. It behaves the way I want it to and matches the behavior of the prototype nicely.

'4072-LM13700-RES_COMP' is the same circuit but with a LM13700 VCA between R43 and the output stage buffer (note: annotation is different in the simulation circuits!).

I've tried tweaking things to change Iabc for the LM13700 as well as the input divider for the LM13700, but increasing or decreasing the gain of the VCA doesn't seem to help much. I've tried using the output buffer stage and I've also tried using a more ideal VCA, the THAT2180, without success.

My theoretical knowledge is failing me here, so I'm really just takes stabs in the dark.

---

4072_VCF_RES_COMP.zip
Description:		Download (listen)
Filename:	4072_VCF_RES_COMP.zip
Filesize:	90.68 KB
Downloaded:	300 Time(s)

---

[gdavis]

Well, first of all your OTA input is connected to the virtual ground of the output opamp.

The way I understand the original mod, you should be hooking it up like the attachment.


See if that gets you going in the right direction. I tried it in the sim and it does, uh, something, but you're sim command is a bit to sophisticated for me to follow

---

filter.PNG
Description:
Filesize:	78.39 KB
Viewed:	311 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[minisystem]

Ha. That completely changes the way I was thinking about it - makes much more sense now. Of course, it makes sense that R43 is summing the feedback with the output at the buffer. I was imagining it the other way round - the input to the buffer was adding to the feedback signal through R43.

Alright, I'll keep chipping away at it and see if it gets me anywhere.

[minisystem]

So, yep, my misunderstanding of how the feedback compensation worked was the problem - thanks for setting me straight! Topology as gdavis indicates is definitely looking promising in simulation. Time to breadboard this sucker and see how it goes!

[minisystem]

In simulation, things are looking good. When R43 is present, I get the amplitude compensation at high resonance that I expect. When R43 is replaced with an OTA-based VCA, I can achieve the same compensation. However, on the breadboard, I'm not seeing the same effect. When R43 is present I see little to no compensation, but do get the expected bleedthrough. The VCA approach can compensate amplitude when self-oscillating, but doesn't seem to compensate at all when filtering an actual input signal; the amplitude decreases with increased resonance and compensation through the VCA has no effect.

At this point I think I'm going to give up. My voice board design will be modular, so I can always update later with a different filter card. I'm not that bothered by the decrease in signal at high resonance and was really more bothered by having half an LM13700 left unused!