Deluxe Reverb strangeness

[Randall]

Hello shut ins. I have a DR that came in for a loud buzz, turned out to be reversed reverb switch connector with Tremolo. That problem solved, I checked out the other functions and noticed the reverb and tremolo didn't work. That led me to the tremolo tube with the first triode shorted cathode to filament. Now here is where the oddness come in. When swapping out the bad tube, the reverb and trem both worked. I don't yet understand why, so I put the bad tube back in just to confirm my findings. It was then that the sound of the amp got quite low and gravelly, and a couple of times faded to nothing, until I put a probe on the reverb channel plate, which made a pop, and then the sound returned to normal. I probed and resoldered around that area, but found no change.

This happened several times. I'm still collecting data, but here is what I think. The times it has done this is when the bad tube is in place when it is powered up after a long cool down period. Every time I touch a probe to a plate, it returns to normal. Once while it was in this condition I popped out the bad tube, and it stayed in this condition until I tried to measure with a probe. I even connected the probe to a few different plates before powering up, and in each of these cases it did not do it. As far as I know it has only done this condition when the bad tube was in place. Most of the time it powers up and sounds fine. Indeed, if it does power up in this condition, once I either probe it, or hit a solid guitar chord it will come back and as far as I can tell stay good.

Now I know someone is going to tell me not to put the bad tube back in, but when it is not in the odd condition, everything looks and sounds pretty normal. I don't know what is the deal here, and can live with tossing the tube and just accepting that as the problem, but I would like to understand what is happening here. How does a bad trem tube make the reverb not work?

http://ampwares.com/schematics/deluxe_reverb_ab763.pdf

[g1]

When it was working without trem and reverb, it was normal sounding? Or always the 'low and gravelly'?

[Randall]

Through the fog of war, I am not certain of the timeline, but I think it was both, yes and sometimes no.

[Randall]

Well dang, I thought I had it beat, but alas, it did it again just now.

At least now I can be more susinct about conditions, and hopefully less confusing. The bad tube in question has been sidelined before the most recent episode, so that rules it out. New tube in trem position, and two new plate resistors for V2. I left it on for about an hour and when I randomly checked it, it was doing the low fading out gravelly sound, but only in Channel 2, channel 1 was clear.

I measured good filament voltages on both channel 2 and the PI, and as before, the second I touched my probe to measure plate voltage, it snapped back to normal. The entire area around V2 has been resoldered on both ends of the wires, prior to this. Also, A strong signal from the guitar will put it back to normal as well.

This is a sticky and tricky one.

[Enzo]

Touching a probe to the circuit completes a circuit with the impedance of your meter. It suggests to me perhaps and open grid return resistor. WHich plate? And by channel 2 we mean vibrato channel?

Also, putting probe to plate pin pushed on the tube socket pin, which may be involved. Just for science, instead of measuring at the plate pin, measure at the end of the plate resistor or the coupling cap lead. Electrically the same but gets you away from the socket. Any difference?

Or instead of looking at plate voltage, try checking cathode voltage instead. Does that also bring it back, or does the cathode voltage look right?

[g1]

My only guess would be that it pulled down the heater voltage a bit, and one of the reverb tubes was more sensitive to low heater voltage than the other tubes.
The heater is the only thing I see in common between those circuits, and you said it was a heater to cathode fault so who knows.

[Randall]

"WHich plate? And by channel 2 we mean vibrato channel?"

Either one. And yes vibrato channel.

" Just for science, instead of measuring at the plate pin, measure at the end of the plate resistor or the coupling cap lead."

This is how I am measuring.

"Or instead of looking at plate voltage, try checking cathode voltage instead. "

Cathode voltage looks fine in normal operation. The next time I can get it to do it, I will put a probe on and see.

"The heater is the only thing I see in common between those circuits, and you said it was a heater to cathode fault so who knows. "

That's what I was thinking also, but remember the bad tube is removed from the picture when it failed this last time. And I measured good filament voltages on a couple of tube while in fault condition, so I don't think this is the issue.

edit:

OK, so I just turned it back on after a 40 minute cool down, and it went into failure mode, signal low and scratchy and fading out to almost nothing. Cathodes measure fine, and do not change anything when probed. Measuring the grids at the socket looked good, but made no change, but when I probed the resistors at the input jacks, it corrected. I do not know if this was from the inevitable pop this made. I hit them with freeze spray with no change. This is all at V2.

[Enzo]

OK, how about apaply a signal, and when this happens, leave it that way. Use a signal tracer (listening amp) and see where the signal collapses. Hey, if you turn up the reverb, do you get reasonable reverb signal? Or is the reverb the same garbled stuff?

You can probe grids without it resets? Fine, scope or listen to the grids of the two stages. SOund OK? Or do one or the other or both go bad?

Did we do the whack test? Whack the chassis with a mallet, any reaction?

[Helmholtz]

I may be wrong but I suspect that the amp is at the verge of instability and that the observed effects are caused by HF oscillation.

[Randall]

I'm not sure about the reverb, I missed it last time around. But have done the whack and probe test, no change.

However, following Hemholtz lead, I do find what I think is a 45K wave at the speaker jack, plugged into the vibrato channel (the one in question), and channel vol turned up to 4. If my calculations are correct, I measure around 22 uSec for a complete wave. 1/0.000022 = 45, 454 Hz. Correct? No such wave appears with the normal channel plugged in and dialed up. Keep in mind, this is after the condition had reset, and the amp sounding normal. I will try to get the same measurements next time it fails. My problem here is when I touch a probe to some things, it resets, but not others.

The wave disappears if the channel volume is turned full off, if the guitar is turned full off, or if either grid on V2 is grounded. So far moving leads has made no change.

Am I barking up the wrong tree here, or is Hemholtz onto something?

[Helmholtz]

Please always post scope settings with scope pics.

A 45kHz signal at the output is clear evidence of oscillation. Prime suspects are typically filter caps.

[Randall]

Scope settings are 5 uSec/division and 0.1 v/div.

I have more data. When I turned it on just now went into failure, and the reverb was also effected, scratchy and barely there. Also, in failure mode the 45K wave almost disappears into about 0.1v of hash. I probed the reverb send tube plates and it made no change, but it snapped back to life when I probed the B+ at the power supply side, and the 45K wave returned. I also swapped the preamp tubes before powering up to rule them out.

If it is filter caps, why does it only effect the vibrato channel, but not the normal channel? And anyway they have been done fairly recently.

[Helmholtz]

If it is filter caps, why does it only effect the vibrato channel, but not the normal channel?

Oscillation often involves more than one amp stage and I think that the reverb circuit is involved.
So especially check the screen filter cap and its ground connection. You could temporarily wire a known good cap in parallel.

When I turned it on just now went into failure, and the reverb was also effected, scratchy and barely there. Also, in failure mode the 45K wave almost disappears into about 0.1v of hash. I probed the reverb send tube plates and it made no change, but it snapped back to life when I probed the B+ at the power supply side, and the 45K wave returned.

One possibility is that the amp toggles between 2 different oscillation frequencies, the higher one being above the bandwidth limit of the OT.
What is the bandwidth of your scope?

[Chuck H]

I'm going to second Helmholtz line of reasoning here. I was actually thinking it earlier, but wasn't quite sure enough to post.

Notice that the vibrato channel has like phase gain stages sharing a filter node. The normal channel doesn't. This would explain why you're only experiencing the problem with the vibrato channel and support the probability of bad filter caps.

You can try paralleling a good cap with the last preamp filter cap. I'll bet it stops the problem. And...

I know it's not your MO to change caps that aren't detected as bad. But let's try to rationalize this. If it turns out that replacing the preamp filter solves the problem I think you should just perform a full cap job on the amp. Unless there's some weirdness in there like THAT particular cap is the only original cap.?. If the caps are all of the same make it's entirely probable the other caps are not sound. They can test fine on a tester and still not be snuff in the higher voltage environment of that amp. The amp may even perform ok even if the other caps are failing or close to failing since the other nodes share antiphase circuits.

It just seems counter productive to me to have an amp open with ONE of the filter caps is exhibiting failure and not replace them all. Including the bias supply caps since they're part of a critical circuit that prevents catastrophic failures!!! They ARE age sensitive parts. But it happens all the time. Someone finds 'a' bad cap and replaces 'the' bad cap. Amp "works" and away it goes. Chances are good the amp could "work" a lot better and more reliably if a full cap job were performed.

JM2C

[Randall]

The bandwidth of my scope is 20MHz.

And I stand corrected, the Sprague Atoms in the can are marked 9-99, so they are over 20 years old. I will lobby for replacing them. I already did the cathode caps, the bias cap is an F&T, I'm not too worried about it.

I hope this gets it because this condition only happens sporadically, maybe once every 5 times it is powered up cold, so it is a time glutton, and is clogging up my bench. Not that I have all that much work right now anyway.

I have only four 16mF/475v caps on hand, if you were to replace one of the Spragues with a 22uF, which one would you pick? Or does it really matter?

"Notice that the vibrato channel has like phase gain stages sharing a filter node. The normal channel doesn't. "

I'm not sure what you are saying here. It looks to me that both channels as well as the reverb tube recovery are served by the last filter cap D.