vacuum tube failure mode and repair?

[coppercone2]

I have a Keithley power supply that I restored by replacing capacitors, carbon composition resistors and general cleaning/lubrication. After about a week I got it to what I consider nice, and one of the things I found that needed replacing is a 8068  vacuum tube.

The 240A supply works with the bad tube, but it has a high offset voltage of 0.3 volts, where with the other tube I can get it less then 1mV with the trimmer. When I turn the unit on, the offset voltage is small at first (~10mV, which is 10x the spec, but its small compared to 0.3V), and as time passes the drift goes up until it gets to 0.3V and seems to stabilize. This also means that there is a high noise on the lower ranges (1-2V RMS), but in the high ranges the noise goes to ~10mV levels, but the spec is again <1mV rms.

I am curious about what failure mode of the tube this is, if anyone has a hunch. I think I hear a 'clicking' noise from it too after like 30 seconds, like a thermal expansion strain noise.When I put the new tube in (GE brand), it works perfectly now, and I don't hear nothing. So maybe its not totally failed, but like degraded, since it looks like at least it does get a OK regulation from it, way out of spec, but it could theoretically be useful for something still. That tube that I took out has really lost the label it had (like rubbed off), I put it into the unit ~2013, without doing any other repairs, and it did work kinda ok, but it broke, possibly because when I replaced the caps, I soldered one of the leads to a floating terminal so the rail voltage of the supply was reduced to 200V from 1800V and it had some floating parts near the rectifier, which also caused a neon regulator tube to break (cold tube).. but I am not sure if this did the damage, or if I just got a dodgy ass tube on ebay in 2013, or maybe it damaged it while cleaning the sockets n stuff.


And also, did anyone ever manage to repair a vacuum tube with the help of a glass blower or something? I have the bad tube here, and it seems like a shame to throw it away, because besides the glass, it seems very repairable compared to say a processor. Just curious

[trobbins]

Valves degrade with age, especially those that dissipate significant heat and conduct significant current.  The manual says to fault find by replacing tubes - you need to get a NOS 8068 to confirm that is your issue.  That's a high power (38W) octal, so it's going to run hot, and grid conduction will likely degrade over time and cause the grid voltage to drift as internal gas conduction varies.  You could halve the grid leak (100k) as a way of alleviating the affect of grid conduction on grid voltage, but you are then starting to play with fire unless you have lots of valve experience, and can work through fault current levels and what could turn out to be collateral damage.

You may be able identify an alternative valve, but again you need to have valve experience.

[coppercone2]

yeah, the fault was fixed by replacing the 8068 tube with a new one from GE that looks to be in good condition, claims NOS, with original box. (the previous tubeI bought for the tubeless supply (it was sold without tubes) had no markings, making me think it was dodgy to begin with, not to mention I bought it for like $8).

So the problem is something with the grid? I replace all the composition resistors with metal oxide ones too, because when I was checking stuff inside, I saw sparks fly from under a resistor, which later turned out to be cracked (100 ohm).

[trobbins]

I was working off your earlier post - good to hear about the new 8068 working.

It's nigh on impossible to recover an aged power tube.  If you're keen on gaining some awareness on failure modes and how tubes operate, then there are very technical reference books you can immerse yourself in - given valves reached maturity over decades of the best science and technology during the mid 1940-50's - eg. RCA 1962 Electron Tube Design will keep you busy even if you have a physics degree.

To me it sounds like changing grid current - a sign of what is commonly referred to as a tube going 'gassy'.

[coppercone2]

Yeah I was not sure if I wanted to study it in depth, that at least narrows down the reading of interest.

That kind of sounds like heating the getter material could help maybe? I compared the working and not working tube, that splotch looks pretty much the same. I guess its not a reliable indicator to just look at them?

[trobbins]

Not always a good indicator.  You can read up on what others have tried, and you can try and measure the change in leakage with time and anode temp if you're keen, but unless the 8068 is unobtanium then its like a degraded lead-acid battery - it's not coming back to full rated capacity and you're going to have to let it go.

[coppercone2]

Yeah I don't even have spare sockets for trouble shooting.

If I carefully remove the glass and disassemble it, do you think the fault will be detectable ? I.e. cut it apart with snips and look at the parts under microscope. But based on your description if its just a slight gas problem, I bet I won't find anything.

[vk6zgo]

Something easy to try is to check the integrity of the soldering of the tube element leads to the pins on the octal base.

If there is a poor solder joint, the connection may fail after it heats up for a while.
The failure may be physical, hence the "click" sound.
Heating the pins & applying a small amount of solder & flux to their ends would prove the point one way or another.

[coppercone2]

oh wow, they don't spot weld that?

[james_s]

Some types of tubes were rebuilt, CRTs in particular, and I think it was once common with high powered transmitting tubes too. The most difficult aspect is that you need to replace the cathode, it is the part that degrades in normal use and it will be poisoned when the tube is up to air for rebuilding. By the time you disassemble it enough to replace the cathode, you're pretty close to the effort of building a whole new tube.

[vk6zgo]

oh wow, they don't spot weld that?

Maybe some, but most of them were just soldered.