Blue flash on V9/V10 of Fender Twin Reverb leaving S/B

[nevetslab]

I have a relatively new Fender Twin Reverb (65 reissue) in the shop. It came in with the mains fuse blown, and after checking the amp out, not having found anything specific, with the shop lights dimmed, the left pair V9 & V10 show a noticeable bit of blue flash inside the Groove Tubes 6L6GC's switching out of S/B, while very little flash seen on the right pair V7 & V8. I checked my service records, finding this had been in the shop for the same reason back in May, while never having found the cause back then.

I'm presently out of fresh new power tubes, though that's my inclination....swap out the power tubes. I did leave it running with burst pink noise driving the amp, loud enough that you can't talk over it, but not so much that there's substantial increase in mains current when the signal bursts on for 3 seconds.

The ambient light level in the shop is bright enough that I don't normally notice the change in color inside the tube when it's powered up.

After the amp being on for an hour, the power xfmr is hot enough to not want to touch it long. And, having a Twin Reverb back on the bench, I tried 52 Bill's findings on Turn-On Thump NOT making any difference by turning up the Normal Ch's volume control at any setting, as well as that for the Vibrato Ch. Not that I'm surprised about the thump.

When I switch back into S/B, I do notice a slight shift in apparent color on V7.

I haven't yet changed the test set-up to drive the amp (burst conditions) into a dummy load, and get substantial mains current increase during drive, and see if I can get the mains fuse to blow. Suspect it's a tube thing, but, still could be early failure symptoms of the Illinois filter caps?

[nevetslab]

I've driven the amp hard for several days, pulling up to 200W/2.6A @ 120VAC, using burst pink noise, amp driving 4 ohm dummy load (as the SPL level with speakers would be insanely loud). The power xfmr has stabilized at around 85 deg C towards the end of the day, and I've yet to have the amp blow it's 4A Slo Blo fuse.. The power tubes all sit at 31mA ea at idle. Power supplies look normal. I've requested a fresh matched quad set of tubes for it (letting CenterStaging fund it). If I don't change the tubes, something tells me it will be back again with a blown mains fuse.

I haven't had it apart for any length of time, to stress-test it and monitor the O/T primary with the current probe, or the same on the screen supplies under drive conditions. Amp is still here in the shop.

[Chuck H]

It seems silly to even mention this to you... Did you swap the existing tubes left to right to see if the flash follows the tubes or stays with the sockets?

[nevetslab]

Silly is as silly does. NO, I'm embarrassed to say. But, I do still have that amp set up on the check-out bench. Lemme go look.

It stays on the same side...V9 & V10. Interesting.

[Chuck H]

Ok. So whatever is causing it is exclusive to that side of the push/pull circuit. The bias is steady Eddy and the same on all the tubes. I'd say that rules out the coupling cap. There are no diodes or anything else hanging off the plates in that amp. If all the components spec out that only leaves the OT as something that could be exclusive to that side relative to voltage switching. I might try disconnecting the NFB and swapping the OT ends to see if the problem follows the OT leads.

[pdf64]

How about if Merlin's trickle charge 47k bypass for the standby is used? I'm thinking that it may be contact bounce combining with back emf from the choke doing something wacky.
Are the power tube grid stoppers mounted at the tube socket terminals on these RIs?

[nevetslab]

How about if Merlin's trickle charge 47k bypass for the standby is used? I'm thinking that it may be contact bounce combining with back emf from the choke doing something wacky.
Are the power tube grid stoppers mounted at the tube socket terminals on these RIs?

They ARE still on the tube socket. I had to stop to look.

It may be that I've just never noticed the slight blue flash when HT fires them up. There's still a couple Florescent lights 12 ft up on the ceiling that I can't get at the power switch for, so I haven't seen this in total darkness. I had read where it's not uncommon for there to be some 'blue light' within the power tubes under power. As there are around 20 Fender Twin Reverbs in our rental inventory, I'm a little reluctant to modify one for this. There are a new matched quad set of tubes on order, so I'll probably wait for those to show up.

I will take a whack at Chuch H's suggestion of removing the FB and swapping the O/T primary leads, to see what that does....see if it moves the flash to the other two power tubes. Since swapping the two pairs of tubes symmetrically (V10>V7, V9>V8) and the symptom remained with V9 & V10, ya has to go hmmmmm??? Interesting.

[pdf64]

It may be quick, non too invasive and easily reversible to tack a 47k >=3W bypass resistor across the standby?

[nevetslab]

It may be quick, non too invasive and easily reversible to tack a 47k >=3W bypass resistor across the standby?

That's certainly worth a try. Thanks.

[nickb]

Can you scope the grid voltage of the 6L6's in question when you come out of standby? If it's bouncing high, take a look at what the PI is doing at that time. Also, try just pulling the PI to see what effect that has. I'm suspecting that on energizing the HT the PI is outputting a big differential signal. If it is then look down the chain to see where it is coming from.

[pdf64]

Yes, it may be that the 100 fold disparity in time constants at the ltp grids could be contributing to this.

[nevetslab]

V9 & V10 Grids hitting 0VDC @ turn-on

Can you scope the grid voltage of the 6L6's in question when you come out of standby? If it's bouncing high, take a look at what the PI is doing at that time. Also, try just pulling the PI to see what effect that has. I'm suspecting that on energizing the HT the PI is outputting a big differential signal. If it is then look down the chain to see where it is coming from.

Nick, you nailed it! At least the results of what's going on. V7 & V8 briefly pulse less negative by about 20V coming out of S/B, but V9 & V10 are going all the way north to 0VDC, before restoring bias voltage. Almost like that with the driver tube removed. I've attached the photos showing V7 thru V10 at the grid Pin 5, then on V7 with the driver tube removed:



I haven't looked at the PI yet, but did briefly look with signal applied, and am seeing the same signal (opposite phase) at the grids of V7, V8, V9 & V10. I'll look at PI stages and see what it's doing.

Before I looked with the scope, I removed the S/B switch and attached a 47k 5W resistor across it. This amp, being a 65 RI Twin Reverb, the S/B switch connects the neg side of the discrete bridge rectifier to Ground, unlike the earlier Twins where it interrupts the power feed to the O/T primary following the output of the first supply filter stage.

Also interesting in the posted images is the direction of the turn-on transient when I close the S/B Switch. On V9 & V10, that transient goes negative. Or, is this just because of the PI phase on lower side?

[Helmholtz]

Contact resistance at CP16/17?

[nevetslab]

Turn-on Transients @ V6 stage

In looking at the V6 stage, I find the neg driver is emitting a large positive-going pulse, up to 400VDC before settling down to it's nominal plate voltage, while the pos driver is smoothly ramping up to it's nominal plate voltage. Differences also seen on the grids of the two pos/neg driver stages, with a ramp-up at the cathodes and the junction of the cathode resistor, the grid resistors and the tail of the cathode resistor where it joins with the F/B circuit & the resistor to ground.



Now I'll exercise the Fast-on connections at CP16 & CP17

[nevetslab]

FB open, Reversed O/T Pri Leads

Ok. So whatever is causing it is exclusive to that side of the push/pull circuit. The bias is steady Eddy and the same on all the tubes. I'd say that rules out the coupling cap. There are no diodes or anything else hanging off the plates in that amp. If all the components spec out that only leaves the OT as something that could be exclusive to that side relative to voltage switching. I might try disconnecting the NFB and swapping the OT ends to see if the problem follows the OT leads.

I first checked the results with just the FB open (removed grey wire from output spkr jack), and the results remain unchanged. Then, I swapped the O/T Primary leads, to see if that makes any difference. I kind of expected no, as pdf64 had suggested it may just be the 100X difference in time constants on the input caps to the PI grids C20 & C21. Here's the photos from this change:



I also cleaned the contacts of CP16 & CP17 Ground connections, which also didn't change anything.

I thought I'd clip on a 100nF cap across C20, to see if that makes the turn-on @ V7/V8 pulse similar to those of V9/V10.



That increase in capacitance at C20 now causes the turn-on transient to hit 0VDC, so I think pdf64 is right.

Now....does all of this have to do with the turn-on pops we hear switching out of Stand-by? Remember, some amps DON'T exhibit that.

At any rate, now looking to see if there is Blue Flash on all four tubes (after I restore the O/T wiring and the FB connection. No, the Blue Flash is still only showing up on V9/V10. Though the burst envelope is longer on those tubes, even though I've got the 100nF clipped onto C20. It's a narrower pulse.

Now, does any of this point to why this amp has had periodic AC mains fuse blowing? After driving it hard for days, pulling 200W from the wall during all that, jacking the power xfmr core temp up close to boiling water temp, I'd have thought at some point, I might blow a fuse, but not yet.

Love a mystery.