[time-nuts] Fine frequency adjustment in 5370B

[Dr. David Kirkby]

Tom Van Baak wrote:

> A PRS10, for example, may have 100x or better
> accuracy than a typical 10811. Yet a 10811 can
> easily have 2x to 10x better short-term stability
> than a PRS10. You can indirectly measure this
> by seeing which reference, internal or external,
> gives the smallest standard deviation for the
> particular measurement you are making.


Tom,

as you once said before, the phase noise on the PRS10

http://www.thinksrs.com/products/PRS10.htm

is poorer than the 10811A too.

Also interesting, is that the phase noise spec on the PRS10 (< -130dBc/Hz at 10 Hz) is sligthly 
misleading, when it is about 100x worst than that at *larger* offsets - see graph on that SRS page. 
However, the worst case which is -111dBc/Hz looks to be *about* 60 Hz (best I can tell from the log 
graph), so perhaps this is a measurement problem, rather than the oscillator itself. Although at about 
200Hz it is poorer than -130dBc/Hz too and I can't think of an explanation for that.

That behaviour is not what "the books" shows, although having never measured the phase noise of an 
oscillator, I don't know if this behaviour is typical or not.

I've been thinking about implementing:

GPS -> PRS10 -> Shera board -> 10811A

where the 1pps output from the PRS10 is used to lock the 10811A to 10MHz using Brooks Shera board. 
That *might* (and I'd be interested in opinions) give better short term stabilty and phase noise than 
the PRS10 itself, while retaining the long term stability of GPS.


-- 
Dr. David Kirkby PhD CEng MIEE,
Senior Research Fellow,
Department of Medical Physics,
Mallet Place Engineering Building,
Gower St,
University College London,
London WC1E 6BT.