[time-nuts] More HP5065 experiments

[Magnus Danielson]

Poul-Henning,

On 09/27/2015 09:46 AM, Poul-Henning Kamp wrote:
> Discovery of the day:  The voltage supplied to the Rb87 lamp changes
> the frequency on the order of 1.5e-11 per volt.
>
> I have no idea why...
>
>      http://phk.freebsd.dk/hacks/HP5065A/index.html
>

I can think of two mechanisms in play:

1) The obvious one is the light-shift as such, which comes from the 
intensity of the lamp. This is due to the Stark DC shift mechanism.
If you dig in literature, stabilizing the intensity of pump-lamp/laser 
is important in reducing the effect. The same goes for the RF-field 
strength.

2) A little less obvious is the fact that the intensity and hence heat 
shift the spectrum slightly, such that the filter-cell does not do as 
good job in filtering the D-line. If one has replaced the filter-cell 
with a modern filter, this effect can still be seen. The miss-match 
between the filter cell and the lamp will cause leakage of the other 
D-line and that will counter-act the pumping and well, reduces S/N and 
such. An even more subtle shift is that you drift away between the pump 
lamp and the rubidium in the reference cell, but most of that is 
compensated by the fact that the lamp is so wideband sources that you 
hardly notice.

Expect not only the feed but also temperature to shift intensity and 
frequency of the pump oscillator.

You should be able to servo the intensity using the DC intensity level 
detected by the photodetector.

Cheers,
Magnus