Is there an oscillation faq/de-oscillation how-to around?

[phaeton]

Hey everyone.

Is there a thread, faq, or ready-made-article available that deals with silencing oscillations in medium-to-high gain circuits?  I'm having quite a time with a couple variations of a circuit lately.

I could post specifics if you all would rather do it that way, but I figured I'd ask first if there were any self-help/just-add-water resources available.  I've searched here, i've googled there, etc. (it's a bit of a tough topic to search on, because there are so many applications for circuits that oscillate on purpose).  I've even read on intentional oscillators to see if I could gain any insight, but either i'm not that talented or capable.

Thanks for any and all!

-phaeton

[R.G.]

It helps to know what makes an oscillator.

An oscillator is an amplifier that produces an output (a) which feeds its own input and (b) in the correct phase to reinforce what the output does and (c) with enough level delivered to the input to have an overall gain from output to input back to output of greater than one.

There are two kinds of oscillators: relaxation oscillators and phase shift oscillators. Relaxation oscillators have an inherent non-inverting gain, so that the output signal phase is already correct for driving the input into oscillation, there just needs to be a path back to the input with enough signal getting there to make up for any feedback path losses.

Phase shift oscillators are amplifiers that have inherently inverting gain, but something about the feedback path or the natural phase shift of the amplifier itself at either high or low frequencies causes a phase shift of 180degrees at some frequency, and then there is also (c).

The problem with giving guidelines is that ANY feedback path that meets the oscillation criteria will make it oscillate. That can be
(a) unintended resistance in a ground wire
(b) a power supply with too high an internal impedance
(c) stray capacitance
(d) wires just a few millimeters too close to one another
(e) the SHAPE of the wires on a PCB
(f) overlapping loops of wire carrying input and output current and getting magnetic coupling
and many others.

The cures are always
(a) identify the feedback path that is providing the oscillatory feedback (aye, there's the rub!!!)
(b) make the feedback path attenuation larger (less signal gets back to the input)
(c) make the feedback path phase shift different
(d) lower the forward gain of the amplifier so there is not enough feedback loop gain to oscillate (no one ever does this, but it does eliminate the oscillation)
(e) make the amplifier forward phase shifts different
(f) both D and E - this is classic negative feedback phase compensation

You're after A and B

[phaeton]

It helps to know what makes an oscillator.

Right!

An oscillator is an amplifier that produces an output (a) which feeds its own input and (b) in the correct phase to reinforce what the output does and (c) with enough level delivered to the input to have an overall gain from output to input back to output of greater than one.

Yes!

The problem with giving guidelines is that ANY feedback path that meets the oscillation criteria will make it oscillate. That can be
(a) unintended resistance in a ground wire

I will have to check this.  Remember once upon a time you suggested that my breadboard could be 'flaky'?  It very well could be adding some variables.  (doesn't explain probs I'm having with a soldered-up copy, though).  I guess I'm actually have TWO different issues, possibly related, or possibly not.

(b) a power supply with too high an internal impedance

In this case, a 9V battery.  Unfortunately, you can lump me in with the folks that are still struggling to understand impedance.

(c) stray capacitance
(d) wires just a few millimeters too close to one another

Are these two related?  For soldered perfboard circuits, I've shortened leads on everything, tried asbestos I could to point all wiring away from all other wiring (in a star pattern from center).  Sometimes it helps a little bit, other times it makes no difference.  So far all my enclosures are PVC or some other form of plastic.  I haven't put anything in metal yet.

On the breadboard I wiggle wires around, shorten them, lengthen them.... it makes little difference.  I do the same with components too.  I've moved stuff around (on the board, increasing/decreasing physical distance of components) to see if it makes a difference, but it doesn't seem to.  Back to the breadboard flakiness, i had actually discovered that i can change the pitch of the oscillation by tapping on the board itself. *tap* and it's an E, *tap* and it's a C, *tap* and it's a really high F, *tap* and it stops.  *tap* and there it goes again.  One day the circuit is completely trouble free, the next day it's not.

It might point to breadboard issues (dust in the holes or something).... but doesn't really address the issue in the soldered circuits.

(e) the SHAPE of the wires on a PCB

No drawing cutesy-wutesy pictures of Tom Servo or Crow T. Robot with the circuit traces? 

(j/k)  so far all mine are perfboard (ptp) stuff, but the PCB stuff i've drawn out and 'planned' to create, i've been pretty careful (or so i think i am) about routing traces and wiring and stuff. 

(f) overlapping loops of wire carrying input and output current and getting magnetic coupling
and many others.

The cures are always
(a) identify the feedback path that is providing the oscillatory feedback (aye, there's the rub!!!)

Indeed.  Can it travel through the ground circuit from the output of one transistor stage to the input of another transistor stage?  Since transistor output is 180 degrees out of phase with the input, I assume that this is what could be happening.  Otherwise I'm at a loss.

(b) make the feedback path attenuation larger (less signal gets back to the input)
(c) make the feedback path phase shift different[/quote
(d) lower the forward gain of the amplifier so there is not enough feedback loop gain to oscillate (no one ever does this, but it does eliminate the oscillation)
(e) make the amplifier forward phase shifts different
(f) both D and E - this is classic negative feedback phase compensation

You're after A and B

I am, but don't think I haven't considered (f).  Instead of having two transistor stages going balls-to-the-wall, maybe having 3 transistor stages with less gain each is a better idea?  Thing is, there are sure a lot of two-stage distortion/fuzz effects in existence.  They must be doing something right while I'm doing something wrong.  The schematic is this thing if it helps.

(Unless high-gain transistors like the 2N5089 would *never* fly in a FuzzFace.....)


Once again, thanks for a great, succinct and informative post, (as always), R.G. 

[R.G.]

The schematic helps. I can see a couple of things right away that **might** be helpful.

- C5, the bypass cap; if it'sreally wired that way, this maximizes the feedback through the ground wire resistance. Move C5's negative end over to near the place where you have the ground symbol
- If that doesn't help, use C5 to bypass the power supply for the second transistor, then insert a 100 ohm resistor in the +9V line to the first transistor and bypass the circuit end of that with another cap, 10uF to 100uF.
- Assume that your wires are in fact low value resistors. Especially the wire that runs from the ground symbol in the middle of the circuit to the ground lug of your input jack. 100% of the current drawn by the circuit runs down that wire, into the sleeve contact of the jack, from the jack to the sleeve of the plug, across the sleeve to the ring contact, out the ring contact to the battery. So the ground of the input stage is being wiggled around by all of the current flowing through the (resistive) wire from the circuit to the input jack ground lug. This can easily be enough to cause oscillation, especially if C5 is wired up the way it's shown.
Try hard wiring the battery - lead and see if that helps.

There are others, probably, but that's where I'd start based on the schemo.

[phaeton]

The schematic helps. I can see a couple of things right away that **might** be helpful.

- C5, the bypass cap; if it'sreally wired that way, this maximizes the feedback through the ground wire resistance. Move C5's negative end over to near the place where you have the ground symbol
- If that doesn't help, use C5 to bypass the power supply for the second transistor, then insert a 100 ohm resistor in the +9V line to the first transistor and bypass the circuit end of that with another cap, 10uF to 100uF.
- Assume that your wires are in fact low value resistors. Especially the wire that runs from the ground symbol in the middle of the circuit to the ground lug of your input jack. 100% of the current drawn by the circuit runs down that wire, into the sleeve contact of the jack, from the jack to the sleeve of the plug, across the sleeve to the ring contact, out the ring contact to the battery. So the ground of the input stage is being wiggled around by all of the current flowing through the (resistive) wire from the circuit to the input jack ground lug. This can easily be enough to cause oscillation, especially if C5 is wired up the way it's shown.
Try hard wiring the battery - lead and see if that helps.

There are others, probably, but that's where I'd start based on the schemo.

The short story:  No Dice.  I'm giving up on this idea, at least for now.

The long story: I did all the things stated above.  I re-routed C5 several different ways, even pulled it out completely once.  I created separate power connections for the two 'halves' (at least as well as I could- in theory they're still electrically common at the battery terminals).  Gone over the schematic many times, and

1) The oscillation is still there, usually (it comes and goes on its own whim, mind you)
2) I have yet to have this circuit sound the way it did the very first time I built it and got all excited about it.

For #1, I'm wondering if my breadboard setup is injecting all sorts of opportunities for bizarre behavior.  Remember my tribulations with the Quick N Dirty Oscillator?  Tapping onto the breadboard surface still changes pitch and intensity of the oscillation.  Yet, moving components around and jiggling wires doesn't.  Perhaps I should disconnect my two halves and use a smaller breadboard.

For #2, either I built the circuit differently than what I intended and wrote down (i.e.. misread components), or it goes back to #1.

Thanks for the help R.G., but as best as I can figure, even if I *do* solve the oscillation problem, this circuit probably won't sound very good anyways (simply due to its design and what it is).  You can confirm that if you want, I won't get offended

Instead of trying to polish this turd, maybe I should build a few 'tried and true' circuits next instead of getting all cute and 'creative'.    It's a problem I have with just about everything.  Oh well.

Thanks again.  I'd invite you over for beer and brats, but then everyone would think I'm trying to schmooze or something. 

[TELEFUNKON]

have you tried the circuit without using a bypass switch,
i.e.: guitar direct to input on the left side,
and amp direct at the right side of the device?
(without combining the weak inputsignal and the hot outputsignal
within the 1/10th of an inch distance in the footswitch)