Spring Reverb oscillations

[r080]

I have slowly been fixing a Peavey Classic 30. The reverb tank had been removed in the past, so I put in a replacement. After reading the ESP article on reverb tanks, I also changed the 4558 reverb driver/recovery chip for an NE5532. Everything was working except for a high pitched oscillation, which went away after changing back to the 4558.

Why would this circuit oscillate with the 5532 and not the 4558?

This amplifier is tough to open up, so I may not even bother, but I don't see why the "inferior" opamp would oscillate with this circuit.

https://www.bustedgear.com/images/schematics/Peavey_classic-30.pdf

https://sound-au.com/articles/reverb.htm

[ElectricDruid]

I don't know either, but I wouldn't say the NE5532 and 4558 are in many senses "equivalent parts". They share a pinout and a few other details, but they differ in major ways too, so it's not a huge surprise to me that a circuit designed for one didn't work with the other.

You have to be a bit careful with old gear that was designed with fairly slow, low-bandwidth op-amps. If you stick in something modern that has much wider bandwidth and a much better slew rate and etc etc, you might well find that the circuit oscillates all over the place. The original designers *didn't have to* fit any sort of limiting for that kind of thing because their chip couldn't do it anyway, so no problem. But you stick something modern in and it all goes apesh!t, and no wonder.

[Rob Strand]

There's nothing obvious that would cause the oscillation.    The wider bandwidth op amps definitely have more opportunity to oscillate.   

FYI:  The NE5532 has a much higher input bias current.   With a 1M input resistor you will find the DC offset at the output of the Reverb driver is somewhat higher than the 4558.   Also high 220k and small 1nF cap at the input cause the NE5532 to be a lot noisier than it usually is due to the high input noise current of the NE5523.

The question is what is actually oscillating?  The driver stage, the recovery amp, the reverb as a whole?
Is it audio band oscillation (ie. like a sine oscillator)?,  or HF oscillation (like something unstable)?
Is it oscillating with the reverb on and/or the reverb off? 
Does changing the level of the reverb pot affect the oscillation?

You might need to do some debugging.

The opamp has a 100nF cap across the PSU rails. So that should be OK.  Could try a larger cap in parallel.

The recovery amplifier has an input cap to ground and a feedback cap so the frequency response of that stage should be quite similar with either opamp.   Doesn't seem likely that would be the cause even with a wider bandwidth - not 100% ruled out.

The driver has a 15nF cap resonating with the reverb coil.  If the oscillation is at audio frequencies you might need to add a resistor somewhere to dampen the oscillations.   That could produce some high gain at certain frequencies.  The weird thing is that's common in both the NE5532 and the 4558.   

My gut feeling is the high impedance at the input of the reverb driver has got something to do with it.   You have 220k in series with the input line.   It's possible that the input capacitance, and possibly the PCB layout is promoting feedback to the op amp + input.   The capacitance of the 2x zener should help.   You could try adding 100pF across the zeners or from the op amp + input to somewhere near the ground side of the 22 ohm resistor (R34) - you need to pay some attention to the placement of the ground wire.   That's only going to work if you have HF oscillations.

If you can work out exactly what is oscillating and the type of oscillations it might help narrow things down.

Something supporting the reverb driver oscillating is while the 15nF cap rolls-off the HF gain, the resistor to ground is only 22 ohm, so the stage gain will quite high in the kHz region, approx 100 at 5kHz.

[r080]

Thank you both! My hope was getting better current drive for the tank would improve the sound of the reverb. Comparing datasheets, I wasn't see anything significant that seemed a likely cause here, though I was just using the first google hits, rather than the manufacturer specific datasheet.

It was audio range, probably around 10-15kHz. Reverb pot affected the volume of the oscillation. If I remember right, it was still there when I disconnected the tank, so I was looking at the recovery circuit capacitive load, thinking of adding a series resistor to C30.

At this point, it might be more work that it is worth to modify the circuit, as there is significant physical feedback/oscillation, especially with a lot of bass or mids.

[Rob Strand]

It was audio range, probably around 10-15kHz. Reverb pot affected the volume of the oscillation. If I remember right, it was still there when I disconnected the tank, so I was looking at the recovery circuit capacitive load, thinking of adding a series resistor to C30.

At this point, it might be more work that it is worth to modify the circuit, as there is significant physical feedback/oscillation, especially with a lot of bass or mids.

They are odd symptoms.  It's like the source signal causing the feedback is somewhere between V1B and the power amp.  In which case the place it's getting back into the circuit would be the reverb recovery circuit (R25, U1B).

Check that reverb spring itself connects the metal tank chassis to the ground side of the reverb receive coil (ie. the tank output).   I'm pretty sure all reverb tank used on Peavey's connect the receive coil to the chassis.  That also mounted the reverb tanks via rubber grommets so the tank floats from the chassis electrically and mechanically.  If you can keep the wires from the reverb receive coil away from the PCB.   If they rest on the  PCB near the higher level signals (from V1B through to the power amp) it could inject unwanted feedback.  It wouldn't hurt to check the shield of the reverb cable has a low impedance back to ground.

[Nasse]

I have small collection (2 pieces) of non-working Peavey reverb tanks, must check someday whats wrong with these, anyway was lucky twice just plug in replacement tank made two amps reverb alive again. On solid state redstripe Studio Pro I first oriented input and output socket same way as old tank, but there was extra cable length and turned the tank 180 degrees and got much less hum. You can buy aftermarket padded bag for some tanks.

[r080]

Yes! That was one troubleshooting step I forgot to mention. It was much improved after I rerouted the cables, but still there was whine that only went away after putting the 4558 back in. I also made my own bag with the leg of an old pair of jeans and some tayda ESD bubble wrap bags.

I did try playing the amp a little bit yesterday. I am no spring reverb connoisseur, but it didn't sound awful with the 4558. If I set the tank out of the cab, I can turn up the reverb with no mechanical feedback.

Rob - I did verify all the grounds. I am curious, what is making you think the feedback signal is starting between V1B and the power amp, rather than just locally in the recovery circuit?

[Rob Strand]

Rob - I did verify all the grounds. I am curious, what is making you think the feedback signal is starting between V1B and the power amp, rather than just locally in the recovery circuit?

It's because the Reverb level, Bass and Mids have an effect.   Now you mention re-routing the cable makes me a little more suspicious.   If I didn't know you checked the grounds I'd be thinking the wires to the recovery amp in / reverb tank out might have been reversed.

There's two ways signals in the amp can couple back to the recover amp input:  capacitively and inductively.  Capacitive coupling is fended off with the with coax on the reverb tank out and the grounded tank chassis.   Inductive coupling would only come from the output transformer.

If the receive coil and drive coil are on different ends to the original tank it's possible the receive coil is now close to the o/p transformer.   You can get similar issue with the reverb tank output cables being forced to be close to higher level signals than they were before.    Another thing is the metal top of the tank should be between the tank coils and the amp circuits - If you placed the tank in a "bag" it's possible the open/unshielded base of the reverb tank is facing the amp electronics.

I can only air some possibles.   It's not normal for a circuit to be so sensitive without something being wrong.  The fact the 4558 fixes it means there's probably a real problem but it's right on the edge showing itself.

[r080]

It's because the Reverb level, Bass and Mids have an effect. 

Okay. Clarification - bass and mid only change the mechanical feedback issue, which is solved with the tank out of the cabinet.

I will definitely look at which coil is closest to the OPT. There were some instructions I found somewhere that said to face the side with the RCA jacks forward. I blindly followed that without thinking about why. I would not have come up with inductive coupling, and would have assumed the reason was to keep one cable shorter, like the output coil cable.

[Rob Strand]

Okay. Clarification - bass and mid only change the mechanical feedback issue, which is solved with the tank out of the cabinet.

OK got it.   Yes mechanical isolation is the only way.  The most crude is the rubber grommets and screws but you can do more like put a sheet of soft rubbery material underneath.  Some of those reverb tank bags helped as they were soft and padded.

The old reverb tanks had floating tray for the spring and coils and the designers made some effort to choose the mounting springs to avoid mechanical vibration issues.   In a few cases I even saw foam poked in the hole of the springs to dampen oscillations.    Maybe the new tank looks the part but the details aren't implemented well (due to lost knowledge).

So is the oscillation affected by the reverb level? or did the reverb level only affect mechanical feedback?

I will definitely look at which coil is closest to the OPT. There were some instructions I found somewhere that said to face the side with the RCA jacks forward. I blindly followed that without thinking about why. I would not have come up with inductive coupling, and would have assumed the reason was to keep one cable shorter, like the output coil cable.

Once you know the possible issues like that you can only try a few things.   Sometimes deliberately making it worse can help identify the cause.   If you disconnect the reverb out from the recover amp and just leave the wire hanging, can you get it to oscillate.

Something else that came to mind.   A lot of reverb springs have 1nF or 2nF caps soldered across the reverb output.  The caps are placed inside of the tank across the output connector.    Those caps shape the high frequency response.   They don't appear on many schematics, even if they are there. Without that cap you only have the 100pF input cap on the recovery amplifier.   When you have inductive pickups you can get very large resonant peaks in the response.  These often go unnoticed since the peak is at the top of the audio band or outside of it.   However, the large peaks can promote HF oscillation due to stray coupling - the type of thing that you are seeing where moving the cable has an effect.
EDIT:
I should be more precise about what is going on.  The larger the cap inside the reverb the stronger the peak *BUT* the extra capacitance reduces the ability for capacitive coupling to get into the recover amp.   The cap also changes the frequency of the peak.  No doubt there's a magic value that reduces the overall susceptibility for oscillation.   (You can do more but see how you go for now.)

Perhaps you should just try adding a cap across the reverb output terminals to see if it has an effect.   Also, check the old tank and see if there is a cap inside.