Antique car load dump

I said "at ratings".

The average tube was designed for, and process-optimized, around about 5khr operation at ratings.

Small tubes often went for much longer (I have tons of 12AX7s and whatnot which measure as good as new, despite unknown random histories). Simply because it's hard to run such a device at ratings, especially if you're doing something piss easy like an audio amplifier (where 1mA is a lot of plate current, and resistors over 100k are standard).

TV tubes were the most notorious, all types failing frequently as they ran hot to squeeze every last micromho out of them, shaving every thin cent out of the IF strip, sync and oscillator, amplifier and chroma (for color) sections as possible.

This continued into the hybrid era, where transistors were obviously cheap enough to replace the low level tubes (first, IF, sync and audio; later, everything but the high power and high voltage stages; then everything but the CRT; and finally these days, even the CRT is gone).

Tubes designed and optimized for longer lifetime, did, in fact, achieve longer lifetimes. These were industrial and military versions (e.g., 7025 vs. 12AX7), or "premium quality" versions (ECC88 vs. E88CC). Often, the industrial versions /were/ identical, and simply rated less -- less cathode current and less electrode dissipation gives longer life.

It's not completely true, but it's practically quite close: a thermionic cathode can only deliver a certain total amount of charge in its lifetime. There's an electrophoretic effect, drawing impurities to the surface and poisoning emission.

Typical commercial transmitter finals (ceramic and metal construction) were optimized more for the 10khr range, and could be pushed to 20-30khr with reduced ratings, tight regulation of parameters (especially heater power), and a bit of luck.

Luck being what it is, there will of course be many survivors. See the new thread down below: Survivorship bias in engineering.

The longest lived tubes were made with absolutely stringent chemical purity, reducing the most common failure modes (cathode interface formation, poisoning). They were made with ceramic instead of mica (less dust production, lower leakage), frames instead of rods (greater strength and precision), and often not just gold-plated, but gold-dipped (indeed, gold being used as the filler for furnace-brazing the grids together). They were also burned in for a minimum 5khr, which eliminated early failures (in which time, the average commercial grade part would've expired!).

All in all, a terrifically long, drawn out, and exceedingly expensive process; but what could possibly justify all of this? Undersea cable repeater amplifiers. Indeed, the record stands: among all vacuum tube amplifiers, there were exactly zero failures*. Tens of millions of hours of total amplifier-hours. No need to dredge up an undersea cable and service it (can you imagine?!).

*Or maybe it was one, I forget. Ma Bell was involved in all of this in the 60s and 70s, so you can read all about it in the BSTJ. And I recommend you do; it's great stuff!

So it goes to show you, if you do something with extremely high purity, you can get excellent life. Semiconductors just happen to require this level of purity in order to simply exist at all. Still, early, low purity (and poorly sealed / encapsulated / passivated) devices suffered poor lifetimes, just as low purity tubes did.

Tim