[time-nuts] H-Maser drift (was: Why discipline Rubidium oscillator?)

[Attila Kinali]

On Tue, 21 Nov 2017 04:24:51 -0600
Dana Whitlow <k8yumdoober at gmail.com> wrote:

> This is an active maser, meaning a self-sustaining oscillator whose gain
> medium
> was a volume of hydrogen atoms (at low pressure) maintained in a population-
> inverted state by squirting a thin stream of state-selected H atoms into a
> glass bulb,

They are not maintained in this state. Rather the Hydrogen atoms are
"used up" by emiting a photon at 1.4GHz. They are then "removed"
by leaking out of the opening of the bulb and pumped away.
The source for the Hydrogen atoms is usually a heated platinum
valve (a heated plate of platinum that is thin enough that the
Hydrogen will leak through). The advantage of the platinum valve
system is that it "generates" single atom Hydrogen, as required
by the maser. These atoms go through a specially formed magnet
that deflects the atoms that are in the wrong state (c.f. Stern-Gerlach
experiment). Those in the right state make it into the cavity.
Within the cavity there is a small glass bulb that keeps the atoms
in the right position of the cavity field. In order not to perturbe
the atoms too much, the bulb walls are coated with Teflon.

> The inside of the bulb was treated so that the collisions with the surface did
> not usually cause a quantum state change of the H atom involved.  I've read that
> the average excited atom typically "survived" a large number of such wall
> collisions before being "consumed" by contributing a quantum of energy to
> the oscillating mode; this has always amazed me.

Yes, IIRC normal numbers are several 10s to 100s of wall collisions
before the atom loses its state due to wall colisions and without
contributing to the signal.

> So the primary frequency-determining mechanism is the collision-broadened
> line width of the gain mechanism. However, the cavity resonance exhibits a
> noticeable frequency-pulling effect, and our maser has a feedback loop that
> strives to keep the cavity tuned to the center of the medium's gain
> profile.
> But I think this loop is not a tight loop, ergo not completely successful.

The loop is quite tight. But there are multiple effects that prevent
perfect operation. Major problems are the low signal levels and the
shifts due wall colisions and cavity pulling.

> I've long wondered what causes the slow frequency drift, typically amounting
> to about 3E-14 over a time span of several months.

Mostly changes in the wall coating leading to a different wall collision
shift and mechanical changes of the cavity dimension (think air pressure
and creep) leading to a different cavity pulling. To a lesser extend
it's the changes in the quality of the vacuum and number of Hydrogen atoms
in the cavity.


				Attila Kinali

-- 
It is upon moral qualities that a society is ultimately founded. All 
the prosperity and technological sophistication in the world is of no 
use without that foundation.
                 -- Miss Matheson, The Diamond Age, Neil Stephenson