What would cause this?

[dorrisant]

I have an amp here (BJr) that is giving me fits. It is eating one of the power tubes. I replaced both power tube sockets, installed larger screen grid resistors, replaced all filter caps and modded for adjustable bias with a balance pot. It had crappy filters with ESR through the roof and crusty stuff at the leads.

When I set the bias all was good for a few minutes... Then I noticed something odd but only on one tube. When using the OT shunt method to derive bias, I noticed that after a few minutes the voltage I was watching (plate to CT) changes polarity. I clip my negative meter lead to the CT and the positive lead to either plate. Normally I'm looking for a measurement of -2.6v from CT to each plate. One tube measures as expected the other one will start out fine... but will rise to about +1.8v! I have never seen this happen (with a BJr).

I replaced the power tube sockets because the traces were starting to lift. The board looked a little scorched but not too scorched. Maybe it is after all.

The only conclusion that I can come up with is that the board has become conductive for that one tube and the voltage on either the plate or screen is pulling the bias voltage positive...

Has anyone seen an amp do this before? I have repaired numerous Marshall DSL/TSL amps by drilling out the PCB for pin 5 (EL34) to isolate the bias voltage from the screen voltage on pin 4. This has worked every time. Just wondering if that may be the case here.

[JazzGuitarGimp]

Are you seeing any strange behavior with the bias voltage at the grid pins with the EL84's removed?

[dorrisant]

No... Looks stable... then again, it isn't heating up nearly as much.

Forgot to mention: I had another OT here. I swapped it for no difference.

[dorrisant]

Here are some pics of the bias mod... I lift the legs of the 220ks, put the ends of the pot in those holes. Tack the 220ks to the ends of the pot, the wiper gets a new hole drilled close enough to tie to the bias supply. The 50k is the bias adjustment.

Swapped out the PI caps... no dice.

[Stevem]

Does ungrounding the protection diodes on each plate make any difference?

[dorrisant]

Good call Steve... I will check those diodes and report back later... 😊

[JazzGuitarGimp]

Question: After some "warm up time" one of the output tube plates is rising 1.8V _above_ B+: Does the idle B+ voltage change when this happens? It took looking at the schematic for it to click in my head: the output circuit is somehow creating a voltage _larger_ than the highest voltage it sees from the power supply. This is strange.

EDIT: I suppose an oscillation might fool your meter into thinking its seeing a DC voltage higher than the highest DCV in the amp. Have you looked at that point with a 'scope?

[dorrisant]

Measuring DC at the OT CT to chassis, I see it drop from about 334 down to 322 when the CT to plate goes positive... I still can't see any oscillation on the scope tough. Just to check, I pivoted the clip lead on the plate so that it was in closer proximity to the PI ribbon cable... That made an oscillation show up on the scope. Just not seeing it when the plate was positive in relation to the CT. See pics below.

Fluke 1587 said MAX +4.58vdc MIN -2.6vdc measuring from CT to V4 plate.

I lifted the protection diodes and now it seems to be ok for the moment. I will let it cook for a while and see.

The bias balance pot is not working as I would like... I'm thinking I will remove the 10k trimmer I have in there and replace the two 220k resistors with a 500k pot. Any better ideas? Of course, I'm not touching anything until I confirm that it was the protection diodes first.

[JazzGuitarGimp]

Hey Tony,

Did the protection diodes turn out to be the culprit?

Cheers,
Lou

[dorrisant]

I think so... good call Steve.

Several hours of running normal and I think I'll call it good.