[time-nuts] Tracor 600A Rb works then slowly fails

[Stewart Cobb]

I recently purchased a Tracor 600A rubidium for a good price from the usual
site, expecting to get basically a non-working collection of interesting
parts. However, when I received it, the unit seemed to be in good enough
shape that I tried powering it up. No smoke, so I let it keep going. The
lamp starter relay clicked for a few minutes and then stopped, indicating
that the lamp had lit. The OCXO warmed up normally and stabilized way off
frequency, so I adjusted it back close to 5 MHz. After an hour or so, the
control loop came alive and started steering the OCXO precisely to 5 MHz.
After another hour or so, the green lock light came on, indicating that the
instrument was happy with the level of "second harmonic" it was seeing.

Everything seemed to be working, so I moved it off the workbench to a rack
and started taking data. I tweaked the OCXO frequency adjustment a couple
of times to keep it centered on 5 MHz as the crystal drifted.

Unfortunately, after about three days, the green lock light went off and
the red unlock light came on. The actual control loop stayed locked for
another couple of days but finally unlocked, and the output frequency began
to drift with the OCXO. At that point, I switched it off to let it rest
while considering next steps.

I found a 600A manual online which shows a switched metering setup similar
to a 5065A for monitoring internal conditions. Unfortunately, my unit seems
to be a cost-reduced successor. There is no big switch, and the meter only
shows the control loop voltage. For anything else, I need to manually probe
inside. I wasn't doing this while the unit was running, so I don't have
(for example) data on photocurrent or heater currents or anything helpful
like that.

I know that the lamps are the weak point on these units. However, the fact
that the lamp lit quickly, and stayed lit even as the unit unlocked again,
gives me hope that this particular lamp may still be in good shape. I don't
understand how a "bad" lamp could have recovered enough to run well for
several days before dying again.

Does the group have any suggestions for troubleshooting? Ideally, which
steps to take, in what order?

Cheers!
--Stu