Hello Time-Nuts,
Just wanted to throw in my two cents worth on the rubidium oscillator
heatsink discussion.

My philosophy has always been - keep the rubidium oscillator mounting
surface in the 40-45C range and that is about as good as you can do.

My justification is as follows:
The physics package runs in the 100C range.  With no cooling the
entire oscillator will rise in temperature (as much as it can) to this
temperature.  As commercial components are rated 0-70C this is where
the shortened life comes in from elevated temperatures.
When a heatsink is provided that can keep the baseplate at 40C I
roughly figure the electronics is now at about an average between the
physics package and the baseplate, so about 70C (longer life).
Going overkill and cooling too much doesn't buy you anything (or at
least not much) because now the physics heaters are working extra hard
to maintain the physics package temperature.  This puts added stress
on these components so then they tend to fail faster.

In the end, there is an optimum baseplate temperature that minimizes
failure rate (not eliminates it).  You just have to do the best that
you can to determine what it might be.

Regards,
Skip Withrow

In the NIST paper available at the URL below:

http://pdfs.semanticscholar.org/47ac/742de238c0ece5e91ff7d12c515b9173eb60.pdf

At the beginning of page 2 (4th line) the paper
states:

"Note that the shield permeability is a nonlinear function of the
magnetization and increases to a maximum value of umax =400,000 at
higher applied fields. “Shaking” the shields by continuously
applying an alternating magnetic field is a way to take advantage of umax."

Another paper "The effect of shaking on magnetic shields"
has this abstract:

"The increase of the shielding factor due to shaking was measured in a
scale model for a magnetically shielded room. The increase was found to
be 7 dB for a single-layer square cylinder biased by the Earth's
magnetic field. The shielding factor of a large-volume three-layer
Mumetal ® room was estimated to increase by a factor of 30, thus
confirming the feasibility of shaking in magnetic shields. The shaking
parameters, amplitude, and frequency are not critical according to the
experiments. Winding the shaking coils along the edges of the cubic
shield leads to minimum disturbances inside the cube, and the winding
can also be applied to demagnetize the shield by an alternating field of
25 A/m, 50 Hz. The relative incremental permeability of Mumetal was
studied as a function of the shaking and biasing fields. The
permeability was found to increase considerably by shaking and by
decreasing the biasing field. With zero biasing and with shaking field
of H s = 5 A/m root mean square (rms), 50 Hz, the permeability reached
its maximum value of 89 000, which is sevenfold the value without shaking."

In all my work on atomic standards, I never heard of this.  Has
anyone else heard of this?  I don't understand how a large
AC mag field can be applied to the shield without getting inside
and messing up the atoms by means of Zeeman effect.

BTW, the NIST paper has a nice exact formula for a spherical
magnetic shield (eqn 3 on page 1).  Good reading, as usual from
NIST.

Rick Karlquist
N6RK