Blue Tube Red Plating = Help!

[wushuliu]

I have an Onix SP3 5881 P-P tube amp. Worked fine, then I decided to add some film caps to output caps on PS, and replace the coupling caps. Nothing major. When I powered it up one of the power tubes started glowing blue and then 15 seconds later started red plating. I checked the bias with the pots provided on the amp: the other 3 tubes were .30-.4v, the red plater was 10v(!) (the recommended is 1.15v setting for this amp). Swapping tubes didn't help. I took off the bypass caps on the PS and issues remains. The preamp, driver, phase and 3 other power tubes seem fine. Any ideas on troubleshooting?

I made no rearranging of components, just popped bypass caps on and then off so I'm at a loss.

Here's the schematics from another thread:

[wushuliu]

Pics

[Yvesm]

Since this happened after your intervention, the most probable thing is that YOU make a mistake !

Are output tube still red when you remove the 12AU7 ?
If yes, hunt for a short that killed the -B supply.
If not, suspect your new link caps.

BTW they don't need to be so big

Yves.

[wrenchone]

Those coupling caps really need some sleeving on their leads. With all those leads snaking around, they're just aching to short against something.

[Mr. Zenith]

I hate to be the one to tell you this, but I'm afraid that what you've witnessed is what Tubelab calls the "Red Glow of Death".

According to this site, you appear to have an outdated version of the schematic:

This circuit appears to be of an earlier version of the SP3 since it only shows one bias adjustment pot per channel, but we know that there are 2 pots/channel--one per output tube. So, they must have reconfigured the biasing circuit. to allow for biasing each output tube individually. The circuit was further modified so that the cathode of each output tube is now grounded via a 30 ohm resistor (in the diagram, the cathode is connected directly to ground).

Given this information I can only assume that the 30 Ω resistor is used for bias measurement. If this is indeed the case then the 0.3 to 0.4 V measurements on the other three tubes correspond to a cathode current of 10 to 13 mA, meaning that they're operating at near cutoff (total dissipation of around 3 to 4 Watts each).

The tube measuring 10 V is drawing 330 mA or so cathode current - causing a combined grid and plate dissipation of around 117 Watts!!! Depending on how long the tube was in that state, you'll be extremely lucky indeed if something hasn't melted inside, like grid wires and such. Even worse, drawing that kind of current through circuits not designed for it can easily blow expensive parts - like output transformers.

It's tough for me to go much further without knowing exactly what parts were replaced or bypassed, or without a revised schematic showing the actual bias circuit used.

[wushuliu]

Originally Posted by Yvesm

Since this happened after your intervention, the most probable thing is that YOU make a mistake !

Are output tube still red when you remove the 12AU7 ?
If yes, hunt for a short that killed the -B supply.
If not, suspect your new link caps.

BTW they don't need to be so big

Yves.

I reflowed the solder on the PS joints where I had the bypass caps and also tightened down that socket - it was somewhat loose. No red plating so far, but some of the tubes do have some blue glow now which I don't recall previously. I will re-test tomorrow and proceed from there.

[wushuliu]

Originally Posted by Mr. Zenith

I hate to be the one to tell you this, but I'm afraid that what you've witnessed is what Tubelab calls the "Red Glow of Death".

According to this site, you appear to have an outdated version of the schematic:

Given this information I can only assume that the 30 Ω resistor is used for bias measurement. If this is indeed the case then the 0.3 to 0.4 V measurements on the other three tubes correspond to a cathode current of 10 to 13 mA, meaning that they're operating at near cutoff (total dissipation of around 3 to 4 Watts each).

The tube measuring 10 V is drawing 330 mA or so cathode current - causing a combined grid and plate dissipation of around 117 Watts!!! Depending on how long the tube was in that state, you'll be extremely lucky indeed if something hasn't melted inside, like grid wires and such. Even worse, drawing that kind of current through circuits not designed for it can easily blow expensive parts - like output transformers.

It's tough for me to go much further without knowing exactly what parts were replaced or bypassed, or without a revised schematic showing the actual bias circuit used.

I only had the tube on for 15- 20 seconds at a time. Yes it was damn hot and i figured that tube was drawing a ridiculous amount of wattage, but luckily there seems to be no damage to the rest of the amp so far. Bias is now 2v on the problem socket and 1.3 on the other I tested. Off to bed before I do more.

[wushuliu]

Originally Posted by wrenchone

Those coupling caps really need some sleeving on their leads. With all those leads snaking around, they're just aching to short against something.

Yeah, I intend to do that though there's more room around them than it seems in the picture.

[Rundmaus]

Hi there,

the inner cap on the right side of the picture seems to have its lead in contact with the metal can - possibly shorting it (temporarily?). As written before, the leads should be insulated!

Greetings,
Andreas

[DF96]

I assume the new coupling capacitors are much larger than the old ones. They will have much more stray capacitance to nearby circuit nodes. The result could be ultrasonic/RF oscillation around the phase splitter/driver stages, due to capacitive positive feedback. This may explain the blue - RF can do this to a valve. One option is simply to use normal good quality caps instead of giant boutique items. Another option is to see if you can insert a grounded insulated screen around the caps.

I had similar trouble with a 5-20 phase splitter, due to large coupling caps to the outputs. A screen made of a zigzag of stiff copper wire solved the problem. It stopped the input to the phase splitter from seeing the output.