[time-nuts] Frequency processing scheme of HP5065 vapour rubidium standard

[Magnus Danielson]

From: "Rick Karlquist" <richard at karlquist.com>
Subject: Re: [time-nuts] Frequency processing scheme of HP5065 vapour rubidium standard
Date: Tue, 7 Nov 2006 10:29:49 -0800 (PST)
Message-ID: <9313.192.25.240.225.1162924189.squirrel at webmail.sonic.net>

> Magnus Danielson wrote:
> > There is frequency pulling even in Cesium beams. Only a few beams handles
> > the
> > phase error pulling by reverting the beam direction. Fountains have this
> > together with doppler reduction.
> 
> In the 5071, nearly every error source has been beaten into
> submission.  However, CBT phase shift is the last hold out.
> Extraordinary proprietary measures are taken to make the microwave
> cavity mechanically symmetrical to with tenths of thousands of
> an inch, and with no systematic error.  I can't get into detail,
> but you can believe no stone was left unturned.  The remaining random
> error is typically a few parts in 10^13.  The systematic error
> seems to be below 10^-14.

I beleive you that no stone was left unturned, and then there is a few to turn.

What I have been thinking about is the possibility to electrically detune the
microwave cavity to introduce the necessary phase shift control. The key issue
with that would probably be how to detect the zero-phase shift in a separate
loop not involving external comparision. I haven't looked into the issue in
detail thought, it's not my field really.

However, I nevery fully understood what the issues with the RF path that was
solved/improved with the 5071 generation clocks. Maybe you could elaborate or
even point to some suitable document.

> As Magnus said, the solution to this is beam reversal, however that
> would be too complicated in a compact clock like the 5071.

I agree fully. It would be far too complicated for such a clock. I know PTB did
this for some of their clocks, and I beleive NIST also did it. But those are
laboratory clocks with the emphasis on laboratory, i.e. far from commercial
clocks.

Cheers,
Magnus