[time-nuts] Re: 5071B - and next a 5081A?

[Attila Kinali]

Salut!

Collecting a few replies in a single mail to not clutter the mailinglist
unnecessarily.

On Tue, 13 Jun 2023 16:13:13 +0000
Tom Knox via time-nuts <time-nuts at lists.febo.com> wrote:


> My question to all of you, is if you were leading development of the next 
> generation 5081A what design changes would you implement and what new 
> features would you employ?

Microchip has been working on a trapped ytterbium microwave ion clock
for some time now. It's basically the same principle as the DSAC
that has been mentioned here several times, but using ytterbium instead
of mercury, which gives a longer ion life time (Yb+ is much less reactive
than Hg+). Because there is also no easy way to make a ytterbium lamp,
they have to use a laser for interogation instead, which complicates
things a little bit in terms of production. But the biggest advantage
of ytterbium ion clocks over mercury ion clocks is that ytterbium has
a cycling transition. Which means that one can probe the same atom
multiple times with a laser to see in which state it is, without
destroying the state. Which gives ytterbium a much higher photon
efficiency and thus lower short term noise. 

Current prototypes have a short term stability of 3e-12 and a flicker
frequency floor of 5e-15. Because they contain buffer gas for cooling
of the ions, there is a slow drift, but they haven't published any
data on that so far.

It will still take a few years until we see ytterbium ion microwave
clocks on the market, though. They still struggle with quite a few
technical problems, one of which is laser reliability.

On Tue, 13 Jun 2023 11:24:17 -0700
Richard Karlquist via time-nuts <time-nuts at lists.febo.com> wrote:

> About 5 years ago,
> at the funeral for Lou Mueller, I met a manager from Microchip (or maybe
> it was Microsemi then) who told me that lasers were still not
> sufficiently reliable to use for a Cs beam unit.  It is not clear to me
> that optical pumping would improve the accuracy/stability either.  It
> would supposedly improve the Allan deviation.

Oscilloquartz did that. Appartently, according to the chief engineer,
everyone in the industry told them that it wouldn't be feasible to
do an optically pumped Cs beam standard, due to longgevity of lasers.
Yes, it took them a few years to get it done but it now works. The
imporvement in short term stability is quite significant. Long term
is a bit better than the 5071, but the data I saw was a few years ago
where it was still just a few prototypes, so I am not sure whether
this data is correct. I haven't seen any recent data yet.


On Tue, 13 Jun 2023 19:16:06 +0000
Ed Marciniak via time-nuts <time-nuts at lists.febo.com> wrote:

> If one made a physics package only small enough to fit into a 3U rack 
> chassis, what other types of clocks could approach 5E-13? A cubic 
> millimeter optical clock isn’t going to cut it, but what if the physics 
> package were a liter in volume?

Trapped ion microwave clocks are already on the horizon, as I wrote above.
It is also likely that we will see optical clocks in a small form factor
soon (next 10-15 years). There has been quite a bit of research in that
area, especially in terms of shrinking the physics package (e.g. Opticlocks
Yb ion trap is tiny [1]) and some of these systems are already very compact
(e.g. the university of Wuhan Ca+ ion clock [2,3])

> Or how about designing the physics package in such a way that they
>cesium can be moved between several locations in some way that allows
>the tube to regenerate without breaking it open? I’m thinking some sort
>of valves that could be opened and closed and wicking action not unlike
>a heat pipe cooler uses as transport.

This is more difficult than it seems. Most of the spent Cs ends up
either coating the hot wire detector or the vacuum pump. For the Cs
to migrate back one would need to heat up the whole tube, including
the vacuum pump to the point the Cs boils off and cool the oven down
enough that everything condenses there reasonably fast. This would be
an increadibly slow process (stochastics is not nice when you want
to get things done fast) and would also release all the stuff you don't
want in the vacuum tube that the vacuum pump gathered. Sending the
whole clock in for a tube replacement would be still much quicker and
less problematic.

On Tue, 13 Jun 2023 08:23:00 -0700
Tom Van Baak via time-nuts <time-nuts at lists.febo.com> wrote:

> They kept the yellow-green LCD, which looks more out of place now both 
> in color and tech era. Maybe OLED isn't reliable enough yet. They didn't 
> mess with the red LED display.

OLED, due to it being based on exciting organic molecules to the point
they radiate light, still does not have the lifetime that we would expect
of an atomic clock. OLED barely works for smart phones, where the expected
product life time is less than 5 years and with an on-time of less than 20% 
(probably closer tro 5%). And we still notice a visible degradation of the 
OLED during that short period. LCD on the other hand has easily a life time
of 20 years with little degradation, most of which can be compensated with
increasing the applied voltage. LEDs on the other hand, have almost infinite
lifetime if operated at less than 20% or rated current, which still makes
them blindingly bright, these days.

				Attila Kinali

[1] "Die optische Atomuhr geht in die Anwendung", by Biethahn and Berger,
2019
https://doi.org/10.1002/vipr.201900730

[2] "A compact, transportable single-ion optical clock with 7.8e-17
systematic uncertainty", by Cao et al. 2017
https://doi.org/10.1007/s00340-017-6671-5

[3] "A compact, transportable optical clock with uncertainty and its
absolute frequency measurement", by Cao et al.
https://doi.org/10.1063/5.0079432
(see suplementarry material for size of the whole system)

-- 
In science if you know what you are doing you should not be doing it.
In engineering if you do not know what you are doing you should not be doing it.
        -- Richard W. Hamming, The Art of Doing Science and Engineering