Adventures in buffers/odd question

[m_charles]

Hi everyone.
Had kind of an odd situation pop up that has me puzzled so I figured I'd throw it out there.
I've been plugging away at a Boss HM-2 clone for a while now. I have it laid out on bread-board, got it running, and was fiddling with it.
I was getting tons of high pitch squealing but I wrote that off as the result of a high gain circuit without any shielding because of the bread-board, etc.
Here is what puzzled me:
I took out my stock (real) HM-2 and plugged it in before my bread-boarded one just to compare the sounds. After putting the the real HM-2 in-front of my bread-boarded one, all the squealing stopped on mine...?! Even with distortion maxed, volume cranked, the whole deal, quiet as a mouse.
So my first thought of course was that the buffer from the Boss pedal was fixing some sort of issue with the buffer on the b-board circuit.
The in buffer on the schem is a 10k in series with the input then goes into a standard JFET buffer biased by a 1M going to vr (4.5v) , 10k to gnd on the source, drain tied directly to 9v, etc. Your garden variety JFET buffer. Checked and rechecked, all good.
Now I realize that once my bread-board circuit is properly made, in an enclosure, etc, the squeal may take care of itself, but the situation with the buffered pedal in-front of it totally removing the squeal really has me puzzled and I'd like to learn why this would happen just for sake of learning it.
Any thoughts?
Help is always appreciated.
-Charles

[WaveshapeIllusions]

Did you plug the guitar directly into the clone? Going from your description, it doesn't sound like there's an input cap. It should be in series with the signal before the 1M resistor. Without that, you have a high DC offset at the input, which could possibly interact with your guitar in unpleasant ways.

[R.G.]

It's probably the low output impedance of the pedal in front of it, not the fact that the pedal in front of it is buffered.

A low output impedance on the sending end of the cable between them lowers the apparent impedance the input of the oscillating pedal sees. This makes it much harder for that receiving pedal to pick up the stray feedback it needs to oscillate. You can conceptualize this as the driving pedal's output "shorting out" the feedback that makes it oscillate.

Try this. Put a 10K across the input of your oscillating pedal, from signal in to ground. It probably stops oscillating.

[m_charles]

Hi waves, yes, cap is there, all is correct.
R.G. Did you mean that 10k as an experimental thing, or a literal solution?
Unless I'm misunderstanding, a 10k to ground is way to small if used as a "pull-down" of sorts, right?
I actually had tried this and a 22k got rid of the oscillation, but also dumped so much signal to ground that it sounded completely different.
Here's another thing. I even went as far as building the output buffer of the pedal in front of the input buffer (so basically a standard BJT trans buffer, going into the JFET buffer) in order to mimic the other pedal's output, same result.
If you are now asking yourself why I don't just put it in a proper enclosure, and see if that does it, the answer is simple:
I'm in mad scientist mode, and really want to recreate this simply for the sake of doing it and learning something!
That's the only reason.
R.G. Any thoughts on this? :-)

[R.G.]

Hi waves, yes, cap is there, all is correct.
R.G. Did you mean that 10k as an experimental thing, or a literal solution?
Unless I'm misunderstanding, a 10k to ground is way to small if used as a "pull-down" of sorts, right?
I actually had tried this and a 22k got rid of the oscillation, but also dumped so much signal to ground that it sounded completely different.

Temporary solution, just to verify the problem. You're already there.

Here's another thing. I even went as far as building the output buffer of the pedal in front of the input buffer (so basically a standard BJT trans buffer, going into the JFET buffer) in order to mimic the other pedal's output, same result.

Then try shunting the input of the new buffer to ground. If it's floating around with a high impedance on ITS input too, you have just moved the problem. If the problem vanishes when the newly added buffer's input is shunted to ground, that's another verification of the suspected underlying problem.

If you are now asking yourself why I don't just put it in a proper enclosure, and see if that does it, the answer is simple:
I'm in mad scientist mode, and really want to recreate this simply for the sake of doing it and learning something!
That's the only reason.

That's a good enough reason. Pursuit of knowledge is valuable, more valuable than any partial pedal. I often refer to experimental prototypes as "victims".

I believe that under the several layers of issues that are obscuring things for you is a problem in wire routing or layout; possibly that it's lying on a metal table, or metal under some cosmetic surface. Oscillation requires three things: a feedback path, a phase variation in the forward amplification plus feedback path that makes the signal phase right to reinforce itself, and enough forward gain to make up any losses in the feedback path. You're using high gain so even high-loss feedback paths can do the trick. You're using a high input impedance buffer, so that even high impedance feedback paths don't suffer too much loss. All it takes from there is some phase shift to make the feedback positive at even one frequency, and you have an oscillator.

Given that you can't cut the gain (and sometimes follower buffers can show oscillation at a forward gain of less than one, if there are resonant things in the feedback path to make resonant gain) and that you want the high input impedance, you are left with finding out what path the feedback is coming through and making that path a higher-loss path so that the amount of feedback drops under what's needed to sustain oscillation.

The testing with a lower resistance made the feedback path not have enough signal level to sustain oscillation by forcing it to drive a lower impedance. You can't do that in practice. So you have to find the path for feedback and make it higher loss. The feedback path is probably stray capacitance since shunting the input removed oscillation. So start finding out what the stray capacitor is. Move wires around. In extremis, shield the input buffer wire to head off all the incoming stray signals. I mentioned the metal table because I once had an oscillator I could not kill that suddenly stopped when I moved the hanging garden of parts to a wooden table. The metal table was ungrounded and shunted signal out of one wire and into another. Could do magnetic coupling too, but that's not the problem here, I think.

[m_charles]

any other ideas?

[R.G.]

Read again. I hit "post" before I was finished typing.   

[m_charles]

Ahh, I thought maybe you hit the post button on accident! ha, ha.
But all of the above suggestions still wouldn't solve the mystery of the problem disappearing when the BOSS pedal is in front (meaning the metal table scenario, etc)? Right, or did I miss something?
I agree, wires hanging out everywhere, jumps on the bread-board almost surely are responsible for all the nonsense, but again, why is it gone when that damn pedal is in front?
Why can't I recreate this?
If I'm understanding your reply, you're suggesting that the possible reason is that the buffer being fed to the proto (inside the Boss pedal) is well shielded, so its sending in a nice tidy signal into the bread-board, thus helping it cope with all the susceptibility to oscillations?
Let me know if I'm on the right track and I'll dig in again tomorrow.
I'll come back then with some results. :-)
Thx for the input RG.

[m_charles]

BTW, RG. If you have a min, take a look at my post, " A proposed solution to germanium temp drift...".
I'd be curious what you thought of it.

[R.G.]

But all of the above suggestions still wouldn't solve the mystery of the problem disappearing when the BOSS pedal is in front (meaning the metal table scenario, etc)? Right, or did I miss something?

You missed something.
Putting the boss pedal in front is the same as connecting a small resistor across the input. The boss pedal has an internal "resistance" that is a part of its output buffer. That internal "resistance" decreases the feedback just like the real resistor does. Think of it as the Boss pedal trying to force the input NOT to listen to stuff coming in by stray capacitances. They're effectively identical situations.