Help with voltage-controlled distortion circuit design

[p4rsley]

I'm trying to make a voltage-controlled distortion circuit (see attached schematic). The idea is similar to PAiA's Quadrafuzz module:
http://www.paia.com/ProdArticles/quadrafz-design.htm

..except I'd like to add voltage-control (or CV in the synth world) to each channel. I'm not ripping off PAiA's design, so I wanted to re-design it based on my understanding how the circuit actually works - which isn't too hard (besides the extra circuitry at the output for feedback and loop send/receive which I don't understand).

So the idea.. the input signal splits off into 4 frequency bands/channels, then each one's distortion is independently voltage controlled, and then all channels are summed and re-scaled at the output. Note, my attached schematic has the input/output section on page 1, and only one of the four freq + distortion channels on page 2. There's no global feedback or anything either (don't really need it ..I think?) - so it's a pretty straightfwd circuit.

The part of the circuit which adds the distortion is a variable feedback resistor based on the LM13700, wired as a floating resistor per the datasheet's cryptic wiring (see figure 28 - bottom of page 16 on the datasheet: http://www.ti.com/lit/ds/symlink/lm13700.pdf).

The circuit actually works, but there are several problems. Here are just a couple for now..

1) Distortion is subtle and not very prominent. I'm using LEDs in series with 4148 diodes to do the clipping - which is actually asymmetric because of the different forward voltages on each LED (one is red and the other is blue). But I want the clipping to be at a higher voltage. Should I just add more diodes in series with the LEDs? Also, I'd like the LEDs to be brighter when they’re clipping (right now they're quite dim).

2) The wet/dry circuit (on page 1) doesn't really work, because the signal is always pretty wet. There has to be a better way of mixing wet/dry signals.

Any advice, suggestions, or comments are welcome

[gdavis]

1. The low level of distortion isn't because the clip voltage is too low, it's that it is too high. Adding diodes will only exacerbate the problem, which already may be happening because you've added a diode in series in each direction.

You have a couple possible solutions: Remove a diode (either the LED or 4148) or increase the gain so that the signal is large enough to clip sufficiently with the extra diodes.

BTW, are you sure you don't want R124 going to the negative opamp input? Actually, maybe this is the real source of your problem. R101 and C50 also looks weird.

2. Maybe this will help https://electronics.stackexchange.com/questions/194180/crossfading-between-2-audio-signals-single-supply-opamp-circuit

[p4rsley]

1. If I remove the 4148 diodes from the circuit (which removes the LEDs too), the maximum gain on each channel is 5 - which means the output of the amp is slammed to the power supply rails (for a normal 10V_pk-pk sine wave input signal, I’m measuring almost 22V_pk-pk at the output). If I add only the LEDs (and put a zero ohm resistor in place of the 4148 diodes) I get about 8V_pk-pk. Ideally it would be nice to get like 10V or 12V_pk-pk, I’ll guess I’ll just add the extra gain on the next stage. But the distortion is still a pretty subtle effect - I was hoping to hear a "completely clean" to "extremely grungy big distortion" sound. It just goes from kinda distorted, to a little more distorted with a barely noticeable increase in volume.

I included C50 as an option for high pass frequency roll off, but I’m actually not using it now (it’s just a zero ohm resistor). And R101 is 1k now, which gives a maximum gain of 5.

It’s a non-inverting amplifier config, so R124 goes to the non-inverting terminal. I can also short R124, but it doesn’t make much of a difference. Just added it so the low-resistance output of amp U6D doesn’t see a huge input resistance for U6B, and also to give extra test points on my PCB.

2. About the wet/dry mix circuit.. I should clarify that ideally I’d like 100% clean or 100% distorted when the POT is turned all the way to either end. So that first circuit in your link might work - which couples each signal through either side of the POT (with wiper grounded). I’ll have to give it a try today.

Thanks for the great feedback and suggestions!

[gdavis]

The PAiA circuit you are borrowing from uses an inverting configuration. You can't simply swap the input to the non-inverting input as you've done and expect it to operate the same. Inverting and non-inverting configurations operate differently.

The clipping relies on the the non-inverting input being at virtual ground. When you flip the input as you've done, the non-inverting input is now at the same potential as the input signal, therefore your output is always going to be within 1 diode drop of the input signal level (or 2 with 2 diodes in each direction). Actually I'd expect the output to be higher than the input, not lower, but anyway... This is a simplified explanation that assumes everything is within the operating limits of the circuit.

Ditch R101 and C50, put it back to a non-inverting configuration. Get the basic operation working, then start adding stuff.

The output from the clipping circuit is going to be small (because it's being clipped to +/- the diode drop). If you want 10Vp-p at the final output, you'll need to add gain after the clipping.

[p4rsley]

Oh wow, you are so right about the inverting configuration (forehead slap) I can’t believe I missed that. Thank you! And you’re totally right about the 10k series resistor.

Btw, I opted for the non-inverting config because it has minimum gain of 1. And it’s funny, it actually sounded better. Because now I wired the op-amps (with the clipping diodes) as inverting config and the circuit works, but now I have weird phasing issues, an echo/delay at the output, and LOTS of distortion. I guess I’ll probably need some kind of feedback? Not sure where though - global/local? And it still would be nice to know how to increase dynamic range for LM13700 as a floating resistor.

Also, I had to re-wire the wet/dry mix circuit as inverting config op-amp, otherwise I'm summing an inverted signal with the original.