Was this tube failure a broken grid?

[nhbassguitar]

Built-from-scratch preamp, single dual-triode, designed very conservatively with a view toward long tube life, tested extensively for proper voltages (including spot-on 6.3 volts at the heaters and a 22% safety margin for heater-to-cathode voltage in the CF half), running fine for months, then one morning about 5 seconds after I power it up I hear this faint “pop” from the tube. The gain drops to maybe half of what it was, but the preamp still sounds great.

Long story short, bad tube, and it was a relatively new JJ with less than 100 hours on it. Replacement tube worked fine. So did four more. Tried the original tube again, low gain again. Wiggling the tube in the socket caused no changes. Tried one of the replacement tubes again, back to normal again. Socket pins are tight and clean and have never caused me problems. No internal wiring shorts.

I’ve heard of broken grids, i.e. separated somewhere along their lengths, so I’m thinking that’s what happened here.

Anyone else experience something like this? Sort of an interesting problem, actually. I’ve never seen a tube “fail” this way.

[g1]

Which side failed, the CF? What are the plate and cathode voltages when working right, and when failed tube is installed?

[Enzo]

OK, so you had a bad tube, and what do you know, all the good tubes you tried work OK. You can swap them around all you like, the bad tube won't work well.
What is the mystery?

ANY tube can fail at ANY time...Brand new, 30 years old, doesn't matter. Tubes fail. it doesn't matter if you are "not stressing it". Most tubes are reliable, but when they fail it is no mystery.


What exact little piece of what inside specifically failed? We'll never know. A broken piece of grid wire could be touching a support wire or another element, A flake of cathode material could be shorting two points. Or who knows.

[nhbassguitar]

Which side failed, the CF? What are the plate and cathode voltages when working right, and when failed tube is installed?

That's a great question. I don't know which side failed. At the time, I had Enzo's mindset and basically just wanted to get things moving again. If I get time in the next few days I can open it back up and check signal levels and DC voltages with good & bad tubes. I wish there were more than 24 hours in a day...

[nhbassguitar]

Enzo, the reason I want to know more is that I'm inquisitive. I like to know why things work and why they stop working. We're all different.

But anyway, yes, it does look like a short.

http://icango.net/music-electronics-...-good-tube.jpg

(To state the obvious, the "B" triode is a CF so naturally Vbb = Vb.)

Thanks for the help, guys.

[nhbassguitar]

Turns out I could've saved myself 15 minutes by just measuring the resistance among the 3 electrodes in each tube half, with the tube out of the socket of course.

Good half showed infinite resistance among all 3 electrodes.

Bad half showed 2.4Ω grid to plate and 25Ω cathode to plate.

[g1]

So from those good voltages, it looks like it was the CF side that went bad? They (CF) do seem to be more susceptible to problems, and can be picky about which 'brand' tubes will be most reliable in those positions.

[nhbassguitar]

So from those good voltages, it looks like it was the CF side that went bad? They (CF) do seem to be more susceptible to problems, and can be picky about which 'brand' tubes will be most reliable in those positions.

Correct. The CF. What bugs me is that I took extra precautions around internal voltages under CF operation, even taking into account the AC voltages present under full output. I heard the Rusky tubes at one time were the ones to watch out for in CF circuits but that they're ok now. Never heard the JJs were a problem. But then again, who's to say this tube half wouldn't have failed in a common cathode circuit. I mean, I can think of a half dozen different ways a tube can get buggered between the time it's boxed at a factory 10 Kmiles away and the time it's unboxed by me. I'm going to replace it with an identical tube and see if the failure duplicates. I'm betting it won't.