Calling all op-amp biasing experts

[Mark Hammer]

I tried twice today to make a 9v-powered octave doubling fuzz, using the Anderton schematic, but adapted in a couple of ways.  The original uses a +/-9v supply.  I had made that successfully once or twice before, but had cannibalized it for parts a while back and felt like making one again, only this time with a single battery and all the advantages that offers.  You can see the schematic I used here. 


My plan was to use a standard Boss-type input buffer stage, and co-opt one of the op-amps on a 4136 as an inverting unity-gain buffer so that I could add or subtract the doubled (FWR) signal for different effects (the proposed Octave Blend control).  So, it is slightly different from the original in that way as well.

But that's not the problem.  The problem is that I can't get any signal once things hit the FWR stage (U2).  First I thought it may have been because I fried the 4136, but I built another one using a verified good one and the same thing happened.  So what I want to know is whether I'm doing something wrong vis-a-vis biasing.  I have provided a Vref, using a pair of 22k resistors and a 10uf smoothing cap (not shown), and tied all the Vref points to there.  I have tried to assure DC-blocking at the output of all stages and re-biasing, but I'm missing something.

Have I tied the wrong things to Vref vs ground?  Should I have a current-limiting resistor from one or more of the noninverting inputs to Vref?  I'm stumped.  I've included the 4136 pinout for reference.

Here is the original.  The power connections are omitted for simplicity, but you can asume that all ground connections go to ground, with V+ going to the V+ pin and V- going to the V- pin .

[Gus]

From a fast look I would try a 100uf or so cap from the bottom of the fuzz blend pot to ground not direct to ground as drawn.

[Mark Hammer]

How would that affect the FWR stage?

[amz-fx]

I would put two paralleled caps on the output of U1,  one to drive the Fuzz Blend and the other to drive the 10k going to U2.

Also, the 10n going into U3 is too small... it needs to be at least 0.47uF.

Also the Octave Blend may be too small and the opposite phase signals cancelling...  you could try 100k...

regards, jack

[R.G.]

I'll throw it into the simulator, but there are a couple of obvious problems.

You're forcing the DC bias level off on both the rectifier and on U4 with the way the capacitors are arranged. For a single supply opamp to work, its inputs must both see the reference voltage. That can be by a tie to Vref, or for the - input, it can be supplied by the output if the -input is open to DC or opened by a cap in series.

You have the + inputs tied to Vref, but the inverting inputs are tied to ground on U2 and U4 by the Fuzz Blend control, as Gus mentions.

I would take the cold side of the Fuzz Blend pot to Vref if the Vref is low enough impedance, perhaps a 100uF cap to ground on it. That forces Vref DC levels on the whole Fuzz Blend circuit and should get the DC bias right. You could use a single big cap on the low side, as Gus mentions. That will let the Fuzz Blend pot float up to Vref by the current pulled from the outputs of U2 and U4.

You do need to calculate the time constants for the 10nFs and 10K's at the inputs of U3 and U4 as Jack suggests.

The circuit as shown has a high pass with a low frequency cutoff of F = 1/ (2*pi*10K*10nF) = 1592 Hz. Jack's 0.47 is a cutoff at 33.8Hz, which should be OK.

With the DC to ground broken at the Fuzz Blend, you could probably skip dual caps out of U1.

[toneman]

Try tying the 10K on the output of U3-12 to gnd.
Doesn't need to be in the feedback loop of U3/U2.
Both 10K pots are just mixers.  One fuzz; one FWR.
The 15K and 10K resistors to U4-6 input could be made equal.
Then, 10n cap can be made larger or smaller, since U would have equal mixing ratios.
U3 is an inverting buffer, so U might want to invert signal again, and add another mixer stage.
Then U would have knobs for Fuzz, FWR, & FWR-NOT.
What?  No tone stack???        T

[amz-fx]

Mark,

It's 3am and I can't sleep so I redesigned some parts of the circuit   



The all new Octave Blend section should give you an inverting output when the pot is towards the top and a non-inverting output when it is at the bottom.  I think this could be a better way to accomplish the signal mixing.

regards, Jack

[Mark Hammer]

Your insomnia is my windfall, sir!

Glad I spent the weekend cleaning, cooking and baking.  Now I can spend my evening following up on the very generous suggestions here.

[GFR]

From a fast look I would try a 100uf or so cap from the bottom of the fuzz blend pot to ground not direct to ground as drawn.

That's the first thing I'd try too.

[Doug_H]

From a fast look I would try a 100uf or so cap from the bottom of the fuzz blend pot to ground not direct to ground as drawn.

That's the first thing I'd try too.

You can do that, or you could connect the "ground lug" of the fuzz blend pot to Vr, which was my first thought. Either way should work.

The reason that might keep U2 from working, Mark, is because it incorrectly biases its input since it is referenced to ground instead of Vr.

[grapefruit]

Hi Mark and all.

If using a good low impedance Vref (like a rail splitter IC or op amp) I'd connect the CCW lug of the fuzz blend pot to Vref. If not, connect it to GND and put a 220n or bigger in between the pot wiper and the 15k input resistor. I've had problems with crosstalk when connecting things like that to a resistive divider Vref, even with big bypass cap.

I don't have any way of drawing it here at work, but I'd do the mixing differently....
Have 10k resistors coming from U1 and U2 (via coupling cap) going to either side of 10k linear pot. Wiper of pot goes to Vref (if you have a good one). Each side of the pot also goes to inverting input of U4. You can scale the resistors so you get even levels at either side of pot rotation. With this type of mixing if you have it panned to one side you get NONE  of the other signal, unlike a normal blend control.

I'd probably use 47k resistors and 50k pot so the coupling caps don't have to be so big.

Cheers,
Stew.

[Gus]

did you get it working?

[Mark Hammer]

I made some of the recommended changes but we're having a big lab party this weekend and I've been busy cooking and cleaning, so I haven't put the modded circuit to the test.  If my pedals came out as good as my salsa and lemon-rosemary chicken, I'd be a very happy man.  Maybe tomorrow I'll see if I can fire it up and report back.