Wah annoying RF oscillation on breadboard. *Solved*

[anotherjim]

I've breadboarded a wah with a mind to making a Coqked Wah stompbox. Seemed to work fine... until I tried adding an output buffer as per...

When the wah pot wiper gets to the top, there are crackles, hiss & scratches out of the buffer which do not occur without it. They are not audible on the gate either!
So, I blame JFET bias - the source is only sitting at 0.8v. Spend a while with different FETs, Rs values and gate bias methods. None of that stops the noise out of the buffer. Decide to sleep on it.

The next day, with the scope on, I see a wave around 700kHz burst into life coinciding with the noises when the 100k pot wiper is 15k from the top lug and dies again when it's at max.
Found a couple of ways to stop it...
Q1 : Re increase from 470 to 1k. (reduces gain to stop oscillation)
JFET gate: insert a series resistor of 10k. (stops the noise on the output even though it doesn't stop the oscillation). I admit that I'm not sure how this works, only my intuition told me to try it. Any insights, guys?

As for the wah circuit, I've actually got the old Jen CryBaby circuit values which is mainly the 33k across the inductor. It doesn't work well with 100k anyway. Also, my inductor is an Eagle LT44 transformer primary which is supposed to be 550mH (I guess it's equivalent to the old Archer part RG recommended in his "technology of..." article).

I don't really want to reduce gain with the change to the 470 Re, has anybody any better ideas? Could it just be layout problems or should go away when boxed...
Or, they all do this but the Wah pot doesn't hit the top spot in a treadle mechanism so nobody noticed. 

[niektb]

I'm not exactly sure whats causing this but there are a couple option that you could try:
1: Can't you just tuck a 120pF or so from fet gate to ground? Considering that 700kHz is so high, i guess it could be easily filtered out.
2: 100nF power filtering from the fet drain to ground?
3: What's that 4uF NP thing? If it's an Elco thing than it might be it's poor High Frequency behaviour thats causing troubles. Maybe some 10(0)nFs in parallel would help?

[Vivek]

Would there be an improvement in bias of FET if end of 1M going to ground is made to go to a Vcc/2 instead ?

[antonis]

IMHO, that thing doesn't come from signal by its own..
(buffer Gate HPF corner frequency is at 3.2Hz..)

What way is +9V practically connected on your particular circuit..??
(it might need a LPF between stages..)

P.S.
Of course, as in almost all wahs, first suspect is Wah pot..

[Rob Strand]

Q1 : Re increase from 470 to 1k. (reduces gain to stop oscillation)
JFET gate: insert a series resistor of 10k. (stops the noise on the output even though it doesn't stop the oscillation). I admit that I'm not sure how this works, only my intuition told me to try it. Any insights, guys?

Try,
- a bypass cap on the supply to suppress coupling through the power rails.
- 1k to 2k2 on the base of Q2  in case Q2 is also oscillating.
- A 1k between the JFET source and the output cap, to prevent capacitive loads causing the JFET to oscillate.

Maybe any or all of  the above together with the 10k on the gate.

Gate/base resistors should stop the buffers oscillating.

I agree, be good not to increase the 470R emitter resistor.
I have a feeling some combination of the above will do it.

[anotherjim]

Have plenty of supply bypass capacitance, large electro & small ceramic on top and bottom rails of the BB. Also DC input via 470R for supply ripple filtering.
The 4uF NP is a McCoy thing. Normal Jen was a standard 4.7u polar which I have.


I wouldn't normally go with a 0v ref JFET. It doesn't give good headroom, a least not with the JFET's I've got. It's a follower and you give it 0v to follow? But I tried a Vref 1/2 supply (which I would go with in a build), fixes the headroom, but doesn't stop the noise which I suppose is the JFET input PN junction detecting the RF? But that junction ought to be reverse biased! Pretty sure I have legit & good Fairchild JFETs. Anyway, the series 10k in the gate stops the detection happening & I think I would keep that in a build.

Of course, I tried other 100k wah pots (trimmers actually) but no difference. The oscillation was occurring before I added the buffer but was inaudible by my test amp.

I also tried a small cap to 0v after the 68k input to check if rf was getting in - that made it audibly squeal. That's maybe a clue that it was moving the frequency of the feedback to Q1 base. Although on the face of it, Q1 base gets negative feedback from its collector and via Q2  - there may be enough phase shift and loop gain toward positive feedback when the wah pot resistance seen by Q2 base is a certain value. So...
Rob's suggestion of a resistor into Q2 base may be the cure!

BTW, Q's are 2N3904 with Hfe around 220 each. That ought to be ok according to my readings on the subject.

[antonis]

I also tried a small cap to 0v after the 68k input to check if rf was getting in - that made it audibly squeal.

Could you try that cap between Base & Emitter, insted of GND..??
(as a hopeless attempt..)

[anotherjim]

I just tried something stupid simple (as all good Jimi's should) - a 47pF from wah pot wiper to 0v. Kills the oscillation. Q1 Re now back to 470R.
There are probably other places that would work.
Think I'll call it solved (for now).

And thanks for all the replies - they do help.

[Rob Strand]

I just tried something stupid simple (as all good Jimi's should) - a 47pF from wah pot wiper to 0v. Kills the oscillation.

It's a good idea.

Did you end-up needing 10k on the gate of the JFET buffer?

I'm trying to work out if the buffers are oscillating ie. parasitic oscillation where a stage oscillates locally.  Wiring on the base or gate can make buffers prone to oscillation.  In this case we do have wires to the wah pot.   The 47pF would help that.

The other alternative is some unstable feedback around the whole loop.   The 47pF would help that too.  Even when the wah pot is maxed out there's still the 22k collector resistor feeding the cap.

For audio circuit the 1k on the collector of Q2 isn't so common.    It's on all wah's.   It can be used to stop parasitic oscillations similar to a base or gate resistor (to me it's always less effective).   The reason it is there is never questioned.

It's always amazing how small changes to these common circuits can trigger such evil behaviors.

[anotherjim]

The 10k gate resistor can be omitted. Or it could be 33k so you can say it's a tube amp input load simulation!

Q2 Rc 1k is a bit of a mystery. One thought I have is a practical reason. A lot of older wah boards were made with some transistor legs left long and crossed around to suit the pad layout. Although insulation sleeving should be used, a leg hard-wired to +9v could present a damaging short circuit hazard.

[antonis]

Just wonder if 47pF cap placement on Q1 Collector would result into the same effect.. 

[Elektrojänis]

Thank you very much for this!

I have had a plan for my owh wah build for a while and I have had a circuit on bradboard now for a while. I haven't had time for it for few months, but last thing I remember is that I had a weird noise/thump problem near one end of the pot travel. I have buffer on input and output and I remember that without the buffers the problem was not there...

[anotherjim]

Just wonder if 47pF cap placement on Q1 Collector would result into the same effect.. 

If you mean putting the cap on B-C, then yes it does work equally well. As it's close to the input there, it might be the best place as it also reduces rf entering the input.
Even 10pF works.

[Rob Strand]

The 10k gate resistor can be omitted. Or it could be 33k so you can say it's a tube amp input load simulation!

Thanks Jim.   So it's looking like the BJT buffer is the cause.  With the pot wires on the base it does seem the most likely culprit.

Q2 Rc 1k is a bit of a mystery. One thought I have is a practical reason. A lot of older wah boards were made with some transistor legs left long and crossed around to suit the pad layout. Although insulation sleeving should be used, a leg hard-wired to +9v could present a damaging short circuit hazard.

Yes, in some cases collectors are definitely used for protection.

[amptramp]

The 1K collector resistor for Q2 may have been to use Miller capacitance from collector to base to eliminate oscillation.  Maybe the 47 pF can be added there.

[anotherjim]

Maybe the 47 pF can be added there.

Probably.
Hey, this is starting to remind me of ...

[Rob Strand]

The buffer oscillating issue is a known thing.  Especially with wires on the input.

This paper does  good job of describing the problem and shows a number of solutions, one is a damped cap to ground,
https://www.hifisystemcomponents.com/downloads/articles/Prevent-Emitter-Follower-Oscillation.pdf

It was discussed a few months back in this thread,  and I did a sim with the base-cap to ground,
https://www.diystompboxes.com/smfforum/index.php?topic=124166.20

You can see the effect of capacitive loading in replies #30 and #31 and other comments in this thread,
https://www.diystompboxes.com/smfforum/index.php?topic=119501.20

The 1K collector resistor for Q2 may have been to use Miller capacitance from collector to base to eliminate oscillation.  Maybe the 47 pF can be added there.

I actually tried that idea some time back when I was playing with a wideband amp.   If you have a buffer and put a capacitve load on the output it creates oscillation issues.   You would think with a collector resistor the gain at the collector would go up when you add a capacitive load, like a CE amp with a bypassed emitter resistor, and that would increase input capacitance and perhaps prevent oscillation.   The weird thing was it didn't seem to help oscillation issues with capacitive loads.   It was only an idea I did in passing but I need to go back to it to workout why it doesn't work well.   It's more than just the transistor rolling off the gain at high frequencies.   It seems to be a devil in the details issue, perhaps related to phase shifts.   The analysis by Freucht kind of points to the collector resistor working the same as a base resistor,
http://www.audioworkshop.org/downloads/AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf