[time-nuts] Cold Rubidium?

[Attila Kinali]

Moin,

On Fri, 25 Oct 2019 09:32:59 -0600
"AC0XU (Jim)" <James.Schatzman at ac0xu.com> wrote:

> Does anyone have any experience/first hand knowledge of this Cold Rubidium standard?
> 
> <https://spectradynamics.com/products/crb-clock/>https://spectradynamics.com/products/crb-clock/
> 
> The specs look very good. The mfr claims that, unlike traditional rubidium 
> oscillators, it has no long-term drift.

I have not used it myself (I don't have that kind of budget, unfortunately),
but I talked with Franklin Ascarrunz quite a bit about the design and
what kind of decisions went into it. I will refrain from going into
too much detail, as Franklin shared with me a lot of information that
he probably does not want to be publicly known. If you want to know
those, just go to IFCS or PTTI and talk to the nice people from SDI
yourself.

First of all, yes, this is a cold atom clock. Yes, the principle is
similar to the one of µQuans. Both are turn-key solutions, i.e.
you buy them, you install them, you switch them on, they work.
Both SDI and µQuans have been running one of their modules at
NIST and SYRTE respectively for long term measurements. Both report
very good performance in this regard. Both peform about the same,
which suggest that they are both close to the technical limit of
this type of atomic clocks. It also means you get something that
is in the order of the stability of an active hydrogen maser.
Though, the cost is lower. Low enough that I would choose one of
these instead of buying a 5071.

Size wise, the SDI clock is about the size of a large PC tower,
while µQuans' is about the size of a small cabinet (probably 
three times in volume). I have been told that the µQuans is
mostly air and that most of the space is taken up by the laser
system, but this is second hand information, so take it with
a grain of salt.

The working principle is that they use a magneto-optical trap(MOT) to
gather neutral Rb atoms. This gathering cools the atoms down
(µQuans speaks of ~10µK, I have no numbers for SDI, but should be <100µK
for sure). Using the MOT the atoms are also prepared (ie pumped) to
the correct state, just like a normal vapor cell standard does. Then
the atoms are released by switching the lasers off and fall through the
cavity, which is fed with the interogation signal (the 6.8GHz). When
they reach the end of the fall, they are interrogated with one of the
lasers by means of absorption/fluorescence detection.

There is no cooling of the cavity itself. These clocks are ment
as a replacement for Cs Beam standards and AHMs, as such they
don't have to be as accurate as a fountain standard. Thus the
slight frequency shift due to black-body radation does not matter.
Yes, this means that there will be some slight temperature dependency
of the output frequency. But rest assured, this drift is so small
it is completely drowned in the system noise and other uncertainties.
As for the cavity, the µQuans uses a polished copper cavity, as
they are using the cavity as reflector for the lasers as well. I didn't
ask what SDI uses as cavity material. But guessing that it is a cavity
in high vacuum, I would guess copper as well (aluminium does not behave
quite as well due to the imediatly forming porous oxide layer).

You do not need to replenish the Rb supply, as the Rb atoms do not get
used up. They are free floating in vaccuum and get captured by the MOT.
Thus, unlike the Cs beam's oven, which just spits the Cs atoms out,
the MOT does not need a storage for Rb atoms. Of course, some of
the Rb will get lost due to wall absorption, but that should be only
a minor fraction. I would also suspect that the laser system will
fail before the clocks run out of Rb atoms.


The µQuans clock drops the atoms only for 40ms (IIRC), which means they
don't fall very far and thus there is little change in the field in
which they see. I don't know how far the fall in the SDI clock is.
For clock stability, it's less important that the atoms stay at the
same place rather than they see always the same field over time.
Ie for this application having two interogation spots (due to Ramsey
interogation) does not matter.


I hope this answers all the questions.

			Attila Kinali

-- 
<JaberWorky>	The bad part of Zurich is where the degenerates
                throw DARK chocolate at you.