[time-nuts] Timing Distribution in Mountainous Terrain

jimlux jimlux at earthlink.net
Fri Sep 10 13:53:48 UTC 2010


David C. Partridge wrote:
> Jim is it possible you just gave a workable solution: Equip each site with a small radio telescope and watch pulsars. 
> 
> Aren't pulsars a reliable accurate time source or do they not provide the 30nS over ten days accuracy? 
> 

Pulsars *are* pretty stable.. the problem is whether propagation is 
stable enough. Maybe with a flywheel..

GIMF (pulsar time stability) and it turns up a paper by Petit on DTIC 
(the research was done in 1995 at BIPM in France)
http://www.google.com/url?sa=t&source=web&cd=1&sqi=2&ved=0CBcQFjAA&url=http%3A%2F%2Fwww.dtic.mil%2Fcgi-bin%2FGetTRDoc%3FLocation%3DU2%26doc%3DGetTRDoc.pdf%26AD%3DADA502306&rct=j&q=pulsar%20time%20stability&ei=-DOKTLXwKouqsAO2vNHZBA&usg=AFQjCNEH_epMW_tthuGiX50n93HH4xPxfg&cad=rja

Covers performance and all the perturbations (gravitation effects from 
stars alongside the propagation path, propagation through atmosphere, etc.)


take home value:   for tau of few years, ADEV of few parts in 1E15.
OTOH this paper also says the measurement uncertainty is 1 microsecond..


Also, one might look at:
tf.nist.gov/general/pdf/2166.pdf
Relativity and Timing in X-ray Pulsar Navigation
This paper is about XNAV (think GPS in deep space, using X-ray pulsars 
as the S/Vs)... it seems to show that ModifiedADEV is 1E-13 (an order of 
magnitude worse than "atomic clock"; presumbably Cs) at tau of 0.1 year.

ANother chart shows the relative performance of a whole raft of clocks 
(and you can see why everyone is excited about JPLs Hg-ion clock...)  In 
any event PSR 1937+21 is about 1E-12 at tau of 500,000 seconds.. 
probably not good enough for the OP's need. (PSR 1937+21 has a period of 
about 1.5 msec, by the way.. if pulsars are really a spinning star, 
that's mighty impressive)



thesis.library.caltech.edu/3590/
describes work on another millisecond pulsar.. the abstract says 100ns 
uncertainties, and long term (3yr) residuals of 500 ns.






More information about the time-nuts mailing list