This is all backwards.

The main reason the typical Rubidium box needs a serious heat sink is that
there is an active heater inside it heating up the lamp to get it up to
operating temperature.  That part of the system better be "tolerant" of high
(enough) temperature.

Maybe things would be a lot better/simpler if the heating/cooling we have
been discussing were split into two sections.  One for the lamp assembly, and
a second for the electronics.

Anybody know what the thermal coefficient of the lamp is relative to the
electronics?

--
These are my opinions, not necessarily my employer's.  I hate spam.

Hi

I looked at it that way for quite a while. More or less: A rubidium is like an OCXO, and running it at the upper end of the specified range is just fine.

Then I cooked a few rubidiums ....

The real answer appears to be that the "rest" of the circuitry in there drops MTBF quickly as it gets hotter. The cell runs just fine, but the circuits that drive it die. There are tables in the data sheets that pretty well document this.

Since these gizmos already have been out in the field for a while, I need to get all the life I can out of them.

Bob

On Dec 24, 2009, at 4:59 PM, Hal Murray wrote:

Hal Murray wrote:

... or a less heat-producing alternative could be used. The
Rubidium-lamp produces two wavelengths of which one is filtered by a
Rubidium-filter which leaves the final pumping wavelength. This is what
a laser diode could supply instead.

Most of the discussion has been on thermal isolation of the entier
units. Not what needs generates temperature and what requires
temperature stability etc.

I am not sure I know what you mean by this...

Cheers,
Magnus

Magnus Danielson wrote:

Probably the temperature fluctuations of the absorption cell is more
significant than that of the lamp itself.
The effects to consider are:

1. The effect of temperature fluctuations of the electronics.
   Probably dominated by the short term temperature fluctuations of the
   internal crystal oscillator.

2. The effect of temperature fluctuations of the Rubidium lamp and
   associated optical filters.

3. The effect of temperature fluctuations on the Rubidium absorption
   cell hyperfine transition frequency.

Bruce

Hi

The original intent was to simply take an existing "cheap" rubidium and do simple things to it. Tearing it into pieces and redesigning parts of it was not anything I originally contemplated. The tight integration of the physics package to the electronics would make this a fairly involved process.

Bob

On Dec 24, 2009, at 5:42 PM, Magnus Danielson wrote:

Bruce Griffiths wrote:

The lamp itself shifts intensity sligthly, but the three most sensitive
points to temperature and temperature shifts is the resonator cavity
(pulls the frequency as it detunes), buffert gas shift change with
temperature and this balance the wall-shift so shift in temperature
change the buffer-gas/wall shift balance and shift in temperature will
cause the OCXO to shift and that needs to be canceled in the loop.

The main effect of the lamp temperature is to shift S/N.

All according to my limited knowledge in the field.

The magnetic field applied could also vary. To the best of my knowledge,
I do not know of a way to servo the magnetic field in the fashion it is
done for cesium beams. This servo is part of the modernisation that made
cesiums much more stable.

Mainly shift in intensity and also the temperature widening of the
relevant wavelength. Likewise with the rubidium filter cell absorbing
the unwanted wavelength.

Buffert-gas mix vs. wall-shift balance depend on temperature. At some
temperature the buffert-gas completely cancels the wall-shift. We can
expect the gas-mixture to shift over time and thus the
shift-compensation for a certain temperature, so there is a wear
mechanism in play.

4. The effect of temperature fluctionations on the Rubidium frequency
   resonator.

Resonator cavity frequency shifts with temperature as material expand
(on rising temperature). The Q of the resonator plays a role in the
amount of frequency pulling.

These effects is well covered in the literature.

Cheers,
Magnus

That's why I've been suggesting active control with TE devices.

You can buy a small TE cooler at Walgreens for about $20. It's big enough
for a 6-pack of Coke cans and already comes in an insulated box. Add a
simple temperature control in series w/ the DC supply and you should be
well on the way.

-John

=================

Hi Bob,

I am beginning to wonder if perhaps you may be fooling yourself as to the expected end quality of your efforts regarding all this Rb reconstruction.  These "telecom" units are not in the same class as a hp 5065 unit.

That said, if you study the pictures, previously published, of the insides it is obvious that the physics package, itself, is separate from the electronic boards.  Thus it is possible to separate them such that the electronic boards can be outside of the heating mechanism of the physics package.
Careful repackaging should allow for better control of the physics package so its impact on the electronics is not such a problem.

You have to keep in mind that the packaging was done to fit into small spaces such as card racks.  There is no requirement to keep it that way for your purposes.

I will send, in direct email, the pictures that I am referring to for your perusal.

Bill....WB6BNQ

Bob Camp wrote:

Bob Camp wrote:

Well, the main point with that was that while passive temperature
stability craze have been raving high here, and into more and more
expensive and elaborate propositions, relative simple changes (not
without its challenges) would change the equation (amount of heat to
cool of) quite noticeably. If money was no object, building
no-compromise/prisoners temperature stabilization scehemes around used
commercial rubidiums should not be the optimum way to go. Building a
Rubidum or Cesium fointain would probably be way better use of the
money. Quite a different project thought.

Maybe we need to get back to doable levels, and also consider what
changes Rb frequency, why and what can we do to avoid it.

I have been dipping my nose into the literature, to refresh myself on
the complex interactions. Lamp intensity in itself is a fashinating
topic, while the filtering cells temperature to intensity dependence is
another little complex field of its own and that (as I suspected)
intensity too pulls the frequency. Oh, and after a quick glaze, I found
that the necessary side-peaks needed for servo of C-field exists for
Rb-87, so it can be done similar to that of Cesium.

Cheers,
Magnus

Hi

I certainly agree that, say potting the circuit board, would be a lot easier than some of the stuff we have been talking about.

My main concern about tearing up the unit is impacting the magnetic shielding. I assume that the outer enclosure forms part of the magnetic shield (at least that's what the data sheets say ...).

Bob

On Dec 24, 2009, at 7:51 PM, Magnus Danielson wrote: