Reverb oscillation fixed

[Enzo]

Just a crazy whim. The cold side of your speaker appears to be grounded. Try a small film cap from the speaker cold terminal to its frame with a clip wire or something. ANy change? Sort of a death cap for speakers, so to speak. Not so much as a solution, but as a find out something.

[Chuck H]

Chuck,

First off, nice work on the schemo and agreed the "BOING" component name is a nice touch. Can you say more precisely at what frequency the oscillation is/was?

The oscillation is around/above 15k if I had to guess. When I put the amp back in the cab I can induce the problem and try to read it on a scope, but I don't think that's necessary. For those who can still hear that high through a guitar speaker it's the classic oscillation that is often heard with high gain builds, but this is only on the reverb. It doesn't reverberate either. Which tells me it's the raw gain of the post pan recovery circuit (interacting with the speakers some how?)

[Chuck H]

Double check all the grounds, and that the caps are good on nodes C and F?

Thanks Tim (can I call you Tim?). The amp is only a couple of years old and I used some pretty bomb caps for this one. I'll check them anyway

[Chuck H]

Just a crazy whim. The cold side of your speaker appears to be grounded. Try a small film cap from the speaker cold terminal to its frame with a clip wire or something. ANy change? Sort of a death cap for speakers, so to speak. Not so much as a solution, but as a find out something.

I'm all ears! Because I did that and it did stop the problem. I just grounded the speakers with straight lead since I didn't figure there was any harm in it and it's simpler than including the cap. But I still don't know what it means that a .022uf cap to the negative terminal (ground in this case) stopped the oscillation.?.

[Enzo]

Dunno either, other than it demonstrates that somehow the frame of the speaker is picking up a field or signal somehow and in some way getting it back to a gain element in the amp. My cap shunts any hf on the frame to ground, just as does your ground wire. Just thinking out loud, perhaps there is a capacitive coupling from the hot side of the speaker to the frame, and the frame then acts as a much larger antenna than do the wires, transmitting the whatever so the reverb return can pick it up... somehow...

Don't know if it is a coincidence, but on some speakers where the rectangle fiber board for the terminals is, there is a strap from the negative terminal over to the nearby frame rivet. Have you seen that?

[Tom Phillips]

...on some speakers where the rectangle fiber board for the terminals is, there is a strap from the negative terminal over to the nearby frame rivet. Have you seen that?

I have seen that. I've also seen the negative terminal mounted directly to the speaker frame. The attached photo of a 1960 Jensen speaker shows one such example.

Tom

[trobbins]

To my mind the observations of importance are:
- grounding (AC or DC) the speaker frame stops the feedback
- oscillation is higher in frequency
- proximity dependant

The speaker coil (and speaker wire) carry a large signal voltage. The coil is in close proximity to the magnet system, and would have quite a large surface area (both sides of coil) for capacitive coupling. So I reckon it is capacitive coupling to the basket, which has a large floating surface area which can again capacitively couple to any high impedance signal node that is close enough.

Electrostatic shielding at the source is the best place to start, and works! Trying to shield a reverb pan coil, that possibly doesn't have great shielded cable going to an RCA socket, with the shield connected to the metal work, and then through open wiring to a coil that may be in the proximity to that metalwork, but is certainly pretty open, would make it difficult to fully shield from capacitive coupling.

Is the speaker winding in the amp grounded to amp side ground, perhaps somewhere near the reverb circuit ground?

Did you check if the oscillation was just at a single high audio frequency. Sometimes a spectrum analyser display can very easily identify the early onset of a parasitic feedback oscillation as its popping up out of the noise floor.

I've certainly noticed capacitively coupling with PA style output winding leads that get brought out to terminal strips, or impedance switches, or line terminals. They can be a long distance from preamp circuitry, but still cause noticeable coupling. I'll often unsolder the unused speaker line taps within the transformer, or coil up the flying leads and tuck away within a bell end. 100V outputs are pretty similar to output stage anode signal levels!

[Chuck H]

The idea of capacitive coupling to the frame is a good one. Also the proximity of in/out circuits, as one would typically avoid for the dry signal chain, IS sort of undone with most reverb circuits since the pan output coil that feeds what amounts to a high gain amplifier is likely to be close to the output leads and speakers. In the case of a dry signal chain though, reversing the phase of the input or output device (but not both) will often stabilize the circuit. Which is not my observation in this case as I did try reversing the speaker leads with no success. It should be noted again that the oscillation only occurs near the speakers themselves and not the speaker leads. This observation was crude as I arrived at it by sliding the pan from near one speaker to the other. In between the two, and nearest to the speaker leads, the oscillation almost stops. Regarding the possibility of capacitive coupling to the speaker frames I'll note again that the oscillation is stopped when I ground the frame OR the magnet assembly. One speaker has a resistance from frame to magnet assembly of over 100k and the other only 7 ohms. And if I ground only one speaker frame (or magnet) sliding the pan near the other speaker returns the oscillation. I'm sorry I don't have a spectrum analyzer to run a proper test. I may be able to determine the frequency with my scope if I induce the oscillation, dial my scope into focus on it and then manage the control settings division math correctly.

OH! Another note I forgot to mention... The oscillation only occurs in the VOX mode when the NFB is shunted to ground down to about 15Hz.

It seems to me that this is a more common problem than I knew since grounding of speaker frames for one reason or another isn't unusual and I'm now noting more and more occasions where this reverb/speaker oscillation problem was fixed by grounding the speaker frames. Whether anyone, from the old days to now, knew precisely WHY remains a mystery.

I want to note again that as far as I'm concerned the "problem" is fixed. So this post is strictly an exercise. So if no one is having any fun with that I hope they won't trouble with it. I'm having some fun and will run any tests requested that my limited bench gear allows for as long as the amp remains on my bench. I ordered some new power tubes for it that I expect in a couple of days. After that, it's back with it's owner.

EDIT: The pan is steel, it is grounded to the amplifier chassis via the shielded cables as is typical in these circuits. The amp uses three star points for grounds. One for the preamp, one for the PI, NFB, reverb and trem circuits and one for the power amp and power supply (with the exception that HV rail filters are grounded with their respective circuits). All grounds are run without daisy chaining.

[Tom Phillips]

...I may be able to determine the frequency with my scope if I induce the oscillation, dial my scope into focus on it and then manage the control settings division math correctly. ...I'm having some fun and will run any tests requested that my limited bench gear allows for as long as the amp remains on my bench....

Hi Chuck,
Given that, have you considered using the Lissajous method to measure the frequency? It's an interesting technique that requires only a two channel scope with X-Y mode capability and a signal generator. If you are not familiar with it then just Google or YouTube search "Lissajous Figures." It is fun and generates interesting scope displays.
Tom

[g1]

One speaker has a resistance from frame to magnet assembly of over 100k and the other only 7 ohms.

Interesting. Does connecting magnet to frame of the 100K one have any effect?

[J M Fahey]

Hi Chuck,
Given that, have you considered using the Lissajous method to measure the frequency? It's an interesting technique that requires only a two channel scope with X-Y mode capability and a signal generator. If you are not familiar with it then just Google or YouTube search "Lissajous Figures." It fun and generates interesting scope displays.
Tom

I'm a big fan of Class B movies , including Mike Hammer style detective movies and of course, cheesy "Monsters from the Moon" type ones.
In almost all, when showing the Spaceship or Earth Control you see an oscilloscope with a slow moving Lissajous pattern in the background.
High Tech at its peak

[Chuck H]

Interesting. Does connecting magnet to frame of the 100K one have any effect?

I'll have to check that. Not tonight though.

[trobbins]

I'd think that the magnet and basket act as one large floating conductor at high audio frequency, even if the DC resistance between them is high, as the contact surface area between magnet and basket would be large, and with little separation.

On the observation that reversing the speaker wire polarity made no change - I would think that the phase reversal would need to be done at the OT winding terminals, or if easier then in the output stage, but that would impact any amp feedback circuit.

[Chuck H]

On the observation that reversing the speaker wire polarity made no change - I would think that the phase reversal would need to be done at the OT winding terminals, or if easier then in the output stage, but that would impact any amp feedback circuit.

I thought about that. I couldn't reconcile the design reasonably just for an experiment. Being a Fender/VOX switcher the NFB is wired a specific way to be a characteristic of each "mode". And I hear what you're saying about the coupling between the magnet and basket assemblies. Even at a high-ish DCR there is probably enough capacitive coupling to act as a dead short to a high frequency oscillation. Clever summation. I'll try coupling the magnets to the baskets anyway just cuz. Not until I get the power tubes in and the amp reassembled though.

[Tom Phillips]

Hi Chuck,
How about posting some photos of the Double Agent before you send it back to the purchaser? I'm not asking for detailed gut shots. Maybe just a "catalog page" type shot of the amp. I always like to see amps that are unique and created from scratch.
Cheers,
Tom