[time-nuts] Thinking outside the box a super reference
attila at kinali.ch
Sat Nov 5 05:59:51 EDT 2016
On Fri, 04 Nov 2016 23:04:22 +0000
"Poul-Henning Kamp" <phk at phk.freebsd.dk> wrote:
> First: Yes, but if you pick a sensible vibration mode for your
> microwave resonance, that can be done with an screw-in endcap.
> Second: No, I would actually not need to tune it.
> Historically resonance cavities were used so that step/avalance
> diode multipliers had enough power to excite them. Today we have
> semiconductors which work at those frequencies.
Yes, it is easier to create the required frequencies with high power
(getting 7GHz with 10dBm is trivial and 30dBm can be readily achieved
with single chip amplifiers) and the signal will be cleaner than what
an SRD setup can achieve (no spurs, comparable close-in phase noise
characteristic). Thus we can use cavities with lower Q which are easier
to build and tune. But we still need to ensure that the field is properly
oriented and homogenous over the whole vapor cell. For this you need
a cavity that is properly designed and most likely will be resonant at 6.9GHz.
(I don't know whether it is possible to design a non-resonant cavity with
the above properties)
> But the resonanance leads to all sorts of trouble, including frequency
> pulling, temperature sensitivities etc.
Frequency pulling is not so much of an issue for a vapor cell standard
as it is for hydrogen masers or the primary standards. The shifts due
to buffer gas and wall collisions are so large that a calibration is
needed anyways. Frequency pulling just adds another term. Temperature
sensitivity is IIRC lower for frequency pulling than for buffer gas shift,
but I could be mistaken. For high stability applications, temperature
stabilization of the cavity and the cell are advised anyways.
> Third: A lot of the "everybody knows" about which atoms can be
> used for active vs. passive atomic standards comes from the
> state of the art electronics about 30 years ago.
A lot of the "everybody knows" has been challenged and rewritten
in the past 10-20 years. Although, it looks like the vapor cell
standards have not changed in 50 years, there has been a lot of
research going on and people optimized old ideas and came up with
new ones. Yes, a lot of the electronics design is rather crude,
but most of the people working in the field a physicists or electrical
engineers that just graduated, one cannot expect the level of skill
and expertise of someone who has been doing RF designs for 30 years.
Beside, these people are there to do research, not to design electronics.
The circuits are just a tool for the research, not their main topic.
Hence they rather spend 100k€ on a synthesiser from keysight instead
of spending 6 months for designing their own to save money.
> Using laser-pumping and modern semiconductors, it might actually
> be possible to detect the 6.8GHz photons from the Rb.
You mean an optically pumped active maser? (our vapor cell standards
are passive masers allready) This has been already done in the 60's
and studied later again (eg ). I guess this isn't popular because
the short term stability of Rb vapor cell standards is already quite
good and the long term stability does not get improved. So it does not
justify the additional complexity.
> They won't be coherent photons, like in a Hydrogen maser, but we
> don't need them to be, in fact that just causes the same exact
> problems as the tuned cavity anyway, as long as we can measure
> the frequency well enough.
If you mean to detect the decay from the 5S hyperfine splitting,
then i have to disapoint you. There will not be much radiation to detect.
On one hand, the lifetime of the hyperfine splitting is quite long, thus
the electrons will just get stuck on the upper level after they fell
down from the 5P state (it's a forbidden transition after all).
A photon to stimulate the fall to the lower state is required, either
provided by spontaneous emission from other Rb atoms or from an external
source. On the other hand, if the photons are not coherent, they will
not build up a signal that one could detect. They will just be spikes
that get burried in noise. Yes, one could build a spectrometer which would
average over a lot of these photons, but that would require a stable
frequency source to be able to avearge for a long time. And I don't think
a simple OCXO is up to this task.
 "The Optically Pumped Rubidium Maser", by Davidovits and Novik, 1966
 "Experimental Study of the Laser Diode Pumped Rubidium Maser",
by Michaud Tremblay and Tetu, 1991
Any simple idea will be worded in the most complicated way.
More information about the time-nuts