[time-nuts] Pressure sensitivity of Rb vapor cell standards

[Attila Kinali]

On Tue, 18 Feb 2020 13:42:20 +0100
Magnus Danielson <magnus at rubidium.se> wrote:

> > I still think, replacing the 5065 lamp with a 780nm LED would be worthwhile
> > to try. Yes, the LEDs have a broader spectrum than a lamp, but it is a lot
> > easier to keep it stable than a fluerescent lamp. It would also allow to
> > lower the temperature of the Rb cell to get to the zero tempco point (which
> > seems to be around 60°C for most mixtures).
> 
> Well, as I recall it one has to have a quarter-wave window to remove
> polarization issues, as that too creates light-shift issues. Also, one
> wants a beam-expander to optically cover more of the cavity than the
> direct laser do. I have a bunch of 780 nm LED-lasers for this purpose,
> but it would be fun to try even as quick-hack.

LED, not laser diode. A LED has a bandwidth in the range of 10-40nm,
enough to cover the necessary frequency/wavelength even with production
variation. A laser diode is somewhere in the 100MHz range (unless ECDL),
which means the wavelength has to be tuned to the right one in order
for the system to work. Advantage of the laser: can be very narrow band
(order of 100kHz is readily achievable) and thus have less AC Stark shift.
But you need a lot of electronics and control loops to keep it where it
should be. LEDs on the other hand are drop-in replacements which need a
very small bit of electronics (bascially just a constant current source
is enough). But they need the 85Rb filter to filter out the "wrong"
wavelength, like the Rb lamps do.

				Attila Kinali

-- 
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