Problem with signal splitting

[poiureza]

Hello,

So I thought I could get away with a simple bypass/FX switch at the end of the signal chain.
Turns out I was wrong, the distortion is bleeding in the dry signal when the 100k gain pot is not fully turned down.

I thought the amplifier op amp would also provide the buffering
A spice simulation did not detect the problem, dry out is clean regardless the amplifier settings.
Yep, real world is different ...

I don't really understand where the problem comes from (those few picoamps going through the non-inverting input ? should I add a voltage reference there ?) 

In case I'd need a real buffer somewhere I'm scratching my head to where I should put it ?
In front of the amplifier ? Or in front of the 10u/1k part ?

The input buffer provides the mid-voltage biasing etc ..
But I simplified the picture to show only the split signal

[PRR]

The opamp is often clipping violently. The low-low (1K) impedances are throwing that crap into ground, supply rails, and even through the air.

There are many possible fixes but the few I tried need more pins on the switch.

[Dormammu]

Try using an inverting amplifier circuit after stage 1. They are much less sensitive to interference.
What kind of power is used, 1 or 2 rail ?
If it's not too much trouble, post some sound examples.

[ElectricDruid]

Can we see a full schematic with the biasing drawn out properly please?

The output buffer op-amp has its inputs grounded on this one, and that's not going to work.

And having the biasing *before* the switch is a waste of one resistor. Put it from the +ve input to Vref *after* the switch, and then you only need one resistor to do the job, instead of two - and the input doesn't float while it switches either.

[poiureza]

Ok here is the complete schematic.

The dummy buffer to the left is just for providing the out of phase signal towards the inverting input (it's a microphone).  In spice that is, not in my hardware.
The 2k at the end of chain is my "supposed" mixing console impedance.

FX path is drive + filter

Before you mention it, no I did not put caps across the 100K/100K voltage divider for VR supply as I'm using a 9V battery. I'm still debating if it's worth or not since the DRY out is pretty clean (when the gain is turned down in the other path).

[antonis]

I'm not sure I can follow that schematic (no output for any switch setting..)

P.S.
A capacitor on VR setting IC non-inverting input should be suggested..

[poiureza]

So I investigated FET switching instead of my simple hard switch selector at the end of both chains.
My reasoning is that the distortion path still "operates" even if it's not fed to the output signal.
And that's because the distortion output is connected to ground and somehow modulating something somewhere ...

So I thought that I'd connect its output to the source of a FET i.e. hitting an infinite impedance (when the FET is depleted) instead of the tiny 1k impedance that it sees in the above schematic.


Long story short, I put one FET in the bypass path and one at the end of the distortion path, and well, the FET switching obviously works.

And this raises a slew of new questions :

I attached an Ibanez TS9 schematic as I found this is almost the same circuit configuration that I am trying to get to work.
Notice Ibanez only puts a FET at the end of the distortion path while BOSS always puts one in front of the distortion as well.  Yet I hear nobody complain about the Ibanez bypass ...

Anyway, I am wondering about the caps related to the FET biasing.  I read it's to isolate the DC condition in the FET from the rest of the circuit so the gate condition works nominally.
OK, but why and how ?
Could someone explain to me why the parts in red are needed?
When the signal comes directly from a VR biased emitter-follower (left part of the circuit) or hitting the gate of the exit emitter-follower (right part) aren't we totally sure to have VR voltage regardless of the rest of the circuit ?
FWIW if we remove those then we still have the FET's properly biased no ?

[ElectricDruid]

Could someone explain to me why the parts in red are needed?

The 510K resistors are for biasing, and the caps are for DC blocking. It's as simple as you think it is.

When the signal comes directly from a VR biased emitter-follower (left part of the circuit) or hitting the gate of the exit emitter-follower (right part) aren't we totally sure to have VR voltage regardless of the rest of the circuit ?

No, we're totally sure we *don't* have 4.5V, because there's a diode drop across the transistor BE. The emitter will be at 3.9V or so. Hence the cap and the extra biasing.

HTH

[antonis]

Points A & B sit on different DC voltage levels (despite they are fed from +4.5V via same value resistors..)

Could you figure out why..?? 

[poiureza]

Ouch ... I completely forgot the silicon drop
It would look different with IC buffers for in and out instead of transistors no ?