[time-nuts] Thoughts on Cs tube failure modes

[John Miles]

I recently found a tube from a 5062C on eBay in unknown condition for not
*too* much money, and thought it would be interesting to power it up on the
bench.  Once I got it and saw the 19xx-prefix serial number, I wasn't too
optimistic, since it could potentially be 25 years old or more.  Things went
well for the first few steps of the process, but then the experiment failed
big-time.

1) I first applied +2600V to the ion pump with nothing else connected.  Spec
is < 10 uA.  There was a brief spike to ~100 uA, but within a few seconds,
the current began to drop rapidly, ending up at about 1 uA after a few
minutes.  So far, so good.

2) I then brought the Cs oven up to temperature slowly with a variable
supply.  The 5062C runs its oven in a thermostatic loop, but it was easy
enough to warm the oven up slowly over 10 minutes or so, watching the
thermistor resistance to achieve the 200-ohm reading indicated on the tube
label.  The ion pump current rose to about 2.5 uA during the Cs oven warmup
process.

3) I then attempted to bring up the hot-wire ionizer, which takes 1 volt at
about 1.6 amps (when hot).  Simultaneously, the 22-mA C-field current and
13.9-volt mass-spec supply was applied.  As with the Cs oven, I brought the
ionizer voltage up slowly.

4) At that point the ion pump supply went into full current limiting at
circa 300 uA.

I killed the power quickly, removed the oven and hot-wire ionizer supplies,
and tried powering the ion pump up by itself once again.  Although a DMM
check indicated infinite resistance across the ion pump, the HV supply still
went into current limiting.

I'm guessing that the hot-wire ionizer element had enough crud on it to kill
the vacuum when it vaporized.  The tube envelope is probably OK, because the
ionizer wire itself didn't burn out.  Unfortunately I was watching only the
hot-wire ionizer current during that part of the process, so I don't know if
there was a point where I could have observed a rise in ion pump current and
backed off in time to avoid permanent damage.

The last step would have been to connect the -1900V electron multiplier
supply, feed in a 9.192632 GHz signal from an HP 8672A which would be
frequency-modulated with a slow sawtooth, and watch for an output signal on
a scope with a high-Z opamp buffer.  Unless there is some kind of sequencing
taboo that says "bring up the electron multiplier before the ionizer", I
don't immediately see what I might have done wrong.  Anyone see any obvious
newbie mistakes in the account above?  Or was it just a matter of expecting
too much from a Cs tube that might have been 20 years old?

For what it's worth, the electron multiplier also shorts out its (negative)
supply now.  I don't know if that would've happened earlier, since I never
tried to energize it during the pre-test checkout.

-- john, KE5FX