Biasing differential amplifier from single supply

[electrosonic]

I am working a geofex type wrapper, with a ground hum eliminator (differential amplifier). I have a TL072 buffer into a 4066 bypass setup (the switching in incomplete) then in to the hum eliminator. I can't see how to bias it though. It seems to me that once it is plugged into another pedal DC current will flow from the bias network to ground. Do I need an output cap there (or two) ?

SND is the send to the effect and RET is the return. VR is the bias voltage from a voltage divider. The SND is shunted to the bias when bypassed.

I will breadboard it tomorrow and see how it goes, but any input is welcomed.

Thanks,
Andrew.

[Pablo1234]

OK VR being the Bias Voltage, why is it biased in between the 4066n chip, their is no blocking capacitor in their. Sorry I'm drinking heavily  ATM lol.

[electrosonic]

The idea is to keep all switch points at same dc level so there's no popping when switching.

Andrew.

[PRR]

This really would be easier with a bipolar supply.

Single supply with lifted reference is simple for simple circuits. Here you are going in and out a LOT, forcing many-many in/out coupling networks. You drown in R-C parts.

The coupling between IC2A and the switches is not needed. (I ass-ume the switchs are never both open.)

The output to SND must be DC blocked,

Likewise the input at RET.

The output really does need some ground return to protect against truly floating connections. In the most likely case of hard-bonded grounds at slightly different potential, the resistor does not add to ground-loop area.

The output, as you say, needs two caps. The shell reference cap must be quite large; not 70.7% (3db) at 82Hz, but well under 10% reactance at 50/60Hz.

The 47K resistors must be better than 10% match; use 2%.

The VR must be fairly low impedance.

[electrosonic]

OK, so the output from IC2A is already biased, so the cap and resistor are not needed.

My plan is that this would be a "housekeeping" PCB, used in a few different pedals, the blocking caps on the SND and RET would be incorporated into the other boards as necessary.

You have crossed out the bias resistor R5, what is your reasoning? It is not obvious without the switching but the top left and top right switches of the 4066 will alternate. (pins 3,4,5 and 1,2,13), the SND branch will be biased from either side in the steady state (not sure about the switching transitions though)

The same logic would apply for the removal of R9 bias also I guess.

Andrew.

[Gurner]

I've never worked with CD4066 directly....but I'd imagine that shorting together two of it's 'output' pins (9 & 10), might not be optimal .....those are presumably lowish impedance outputs, so bettter to have some resistance in there.....at probably go with 1ko the output of both  pins 9 & 10 .....with the junction heading off to you final opamp.

[electrosonic]

The shorting of the pins should be ok I think. I am taking that part directly from the Tone God's "Wicked Switches" - converting the 4066's 4 SPSTs to 1 DPDT.

Andrew.

[PRR]

> I'd imagine that shorting together two of its 'output' pins (9 & 10), might not be optimal .....

The switches are 100Meg off, 300 ohms on.

If the switches are worked alternately (9 .OR. 10) there is no conflict. Same as a mechanical switch.

If 9.AND.10 are on, then IC2A fights RET. The stronger source mostly wins, with increased distortion and clipping. This would probably be un-intentional program bug, but not a smoke-maker.

[electrosonic]

How to you calculate the output impedance of the differential amplifier?   Just wondering what the low frequency cut off of the output would be with the 10uF caps.

Andrew.

[electrosonic]

A little more work done on the schematic, taking into account most of PRR suggestions.

I have decided to skip the batteries in my builds, so I skipped the power switching. The stereo jacks are not necessary, but I have some to use up. I will short the input to ground when no jack is plugged in to quiet it down though.

The 2nd half of the 4013 is for if I make a 2 foot switch pedal. I might add another LED in series to be wired off board in case I want to use an LDR for switching.

My plan is to have two boards per build - this utility board and the actual effect. The inputs and outputs are biased, which will be taken into account in the other board.

Not verified, hope to get to that soon.

Thanks to Geofex, The Tone God and Processaurus for the ideas used here.

Andrew.

[PRR]

> How to you calculate the output impedance of the differential amplifier?

Zero to a dozen ohms.

> what the low frequency cut off of the output would be with the 10uF caps.

The cap against the amp output +plus+ the load impedance.

Load is unlikely to be under 10K. 10K with 10uFd is 1.7Hz.

Why so low? Because the ground-compensation scheme needs *precise match* of the transfer functions of the two networks. If the cap adds a 10% error, that's only 1db frequency response error but at-best 20db ground-compensation.

At least I think so; and the cost of your two 10uFd caps is less than the cost of me thinking if 1uFd might be good-enough.