[time-nuts] OSA 8600 BVA repair

Magnus Danielson magnus at rubidium.dyndns.org
Wed Oct 17 20:13:01 EDT 2007


The effort of me and Björn to get a pair of BVAs had a bitter taste to it as
one of the BVAs had a crushed devar flask due to totally inadequate packaging.
It was however running but due to the devar flask it was no longer getting up
to temperature to experience reduced heating of the oven, it kept running at
heatup current which is about 430 mA (at 24 V supply).

Thanks to connection I managed to dig up a replacement devar flask which fits.
Yesterday I did the operation of replacement, which involves disassembly of the
outer shell, removing the end-piece, removing the outer PCB and pulling out the
oven from the broken devar and then reassembly of the whole BVA. The end result
was very encouraging as it finally settled at about 92,5 mA and got fairly
close to expected frequency. After som manual calibration it is less than
100 uHz of the 5 MHz frequency. It seems fairly stable but I would still let it
be heated for quite some time before I do serious judgement of it.

For those of you that don't know the OSA 8600 BVA intimitly, it's an AT-cut
BVA crystal oscillator of either 5 or 10 MHz. I have a 5 MHz one. It has two
SMA outputs with separate output/isolation amplifiers delivering 7 dBm into
50 Ohm. I have used the built-in thermistor for probing oven temperature.
I have also verified that I have operational EFC (called VC).

I was kind of supprised that it wasn't more of-frequency as the shock it had
experienced was obviously great, great enought to brake the inner surface of
the devar flask from the neck and throughout. The oven assembly is a metal
cylinder with two rubber rings against the inside of the devar flask, so it
should have experienced the shock too.

The main obsticle to get it on frequency was to wait out the main heatup.

Using a Pendulum CNT-90 in plot mode was very nice since I get real time
updates where as my good old and trusty HP 5372A needs to complete a
measurement window before it makes the update. For long measurements that is
definitly annoying. No fancy internal reference in the CNT-90, but considering
that the buffer-amp is directly below it and it's reference (my Cs-beam) is
directly below the buffer-amp, it was fair enought.

If somebody twist my arm, I could toss the pictures on the web. I fear some are
shakey thought. Need to get a better camera.

So, looking back at it, this sour story just turned sunny and happy after all.
Naturally greatly in depth to Björn and a few friends at OSA.

Now I only need a good SC-cut BVA, better and more Cs-beams, H-maser, dual
freq GPS and... and... :)

Magnus - happy today

More information about the time-nuts mailing list