[time-nuts] Aging 5065A ?

[Ulf Kylenfall]

Having been shut of for several months I fired up the
refurbished 5065A a few weeks ago. 

It has since been equipped with a new AC-amplifier
(A7) and a new Rubidium cell with the repaired lamp
(thanks to Corby Dawson) and a receiver with a 
TED that was connected according to HP instructions.

Also the 5 MHz original crystal oscillator has been replaced
with an 10811 upgrade kit.

Almost immedeately, the FET's in the 60 MHz Multiplier A3
not replaced during the first repair all failed. 
They were replaced with metal can 2N4416 and all was well. 

The next thing that happened was that the amber 
pilot light "Continuous Operation"
went out. From just a single event, the lamp started
going out just every time I turned my back on the 
instrument. Since there were no sign of out-off lock
conditions and I had no clue other that removing the
logic assy A14 and checking the lamp driver Q18
(1854-0003, selected from 2N1711) I found the transistor
to be leaky. After replacing it with a 2N1893, I have
only had one single such event in a week. Looking at
the 1854-0003 from the bottom, the TO-5 package is most
likely sealed with brown epoxy. And the A14 board
is littered with them.

I could of course replace all these
small signal transistors, be it NPN or PNP with modern
types  but contemplating on this, I wonder how long a 5065
can be operational if the heater windings in the Rb cavity
remains OK and the synthesizer with its unobtanium
logics are OK. From the records on the inside
of the meter and switch parts on the front panel, I can
see that the photo current of the original unit
slowly decreased over the years. I have one more
Rb TX unit but the Rb bulb in that one is really "brownish"
and not in any way clear as the one presently used.

Standard "non-critical" semiconductors can of course
be replaced, even the integrator IC. But what about
transistors in the phase detector assy A8? There
are some oddly looking packages and trying to find
data sheets only returns bad photocopies from
years ago, not really detailing what criteria 
I should be looking for. 

Many web-pages from time nuts displays modern
versions of the 5065 (digital clock). The one
in my lab dates back from 1968 so it is over
50 years old.

Any research made on possible replacement
types for those exotic semiconductors used
in the critical parts as eg. the phase detector?
Best Regards
Ulf KylenfallSM6GXV