My Marshall JCM900 blow tube.

[Alan0354]

When you pulled the two tubes to lower the output power, did you also change the output impedance/tap on the amplifier? Also, did you verify the amp was indeed putting out 30W into the speaker?

No I did not change the tap. It was mismatched and the tubes see half the recommended impedance. That does not help. I edit the post already.

Yes, I measure about 15.5Vrms across the speaker using a meter. V^2/R= 30W. I mean roughly, say between 27 to 32W.

[jazbo8]

No I did not change the tap. It was mismatched and the tubes see half the recommended impedance. That does not help. I edit the post already.

Right, please take a look at the load line below, I think you simply exceeded the tubes' maximum Pda for prolong period of time and if you were using new production 6L6s, they probably can not even handle the 30W per the datasheet to begin with...

Operating conditions: Eg1=-45V, RL=1.7k.

[g1]

Both those tubes have lost their vacuum. I've never seen or even heard of this happening from tubes overheating. The glass would melt before that happened.
As these were both used tubes I think it may just be a coincidence. It is possible the seal broke somewhere in the base due to wiggling the tube around in the socket when installing. That is most likely what happened. Even tubes running hot enough to red-plate don't break their vacuum seal.

[Jazz P Bass]

The tube on the left, that has the whitish stuff at the bottom of the tube, appears exactly like a screen failure.

I have actually seen it happen (Fender Bassman into an attenuator, @ Vol 10, with 470 ohm screen R's)
Twice.
And the results look the same.

[Alan0354]

Screen resistors is not burn, it's not discolored. Looked new.

[pdf64]

I wouldn't fit a power tube in a fixed bias amp and hope that the plate dissipation would be ok, rather my experience is that it needs to be verified.
There's just too much variation between tubes within a type to rely on bogey voltages.
Pete

[Dave H]

I wouldn't fit a power tube in a fixed bias amp and hope that the plate dissipation would be ok, rather my experience is that it needs to be verified.
There's just too much variation between tubes within a type to rely on bogey voltages.
Pete

Exactly! Measure the bias current not the bias voltage. You don't need a bias probe. Use the transformer shunt method. If you only need 60Hz to break in speakers use a mains transformer. Save your tubes for music.

[Jazz P Bass]

It's not the screen resistor that goes bad.

It's the screen grid inside the tube that melts.
Literally.

[g1]

Anyone want to speculate on the mechanism by which a tube biased too hot or with screen failure would lose vacuum?
There have been posts on this forum with pictures of tubes that got so hot the glass melted and warped, yet still kept their vacuum seal.
Is this just one more result of modern tube QC ?

[The Dude]

I'll speculate, and it is purely speculation. My best bet is that there are other factors at play, i.e. vibration, rate of temperature change, tubes at idle vs. tubes with significant audio applied, dirty tubes vs clean tubes, etc. I have also noticed that there seems to be a variance in glass thickness from tube to tube.

It's an interesting conversation and subject of wonder, but who's gonna use a tube that's been hot enough to melt any way? It might as well be cracked.

[Alan0354]

It's not the screen resistor that goes bad.

It's the screen grid inside the tube that melts.
Literally.

But doesn't the screen resistor supposed to take the blunt first?

[Alan0354]

I am pretty comfortable to say it's me screwing up, not the Marshall. I did not think and pumped high power. I bet if I just play guitar, I would have got away with it, just not 60Hz.

So far, the Marshall is working just fine.

[Jazz P Bass]

Screen resistors usually fail when the output tube fails.

Now there are a number of failure modes that can make the resistor fail.

An overdriven amp that melts the screen itself is not one of them.

[Alan0354]

But the amp is not over driven, I did not hear any clipping sound.