[time-nuts] Loran - any good for timekeeping?

M. Warner Losh imp at bsdimp.com
Fri Apr 21 20:01:58 UTC 2006


In message: <1616.1145648661 at critter.freebsd.dk>
            "Poul-Henning Kamp" <phk at phk.freebsd.dk> writes:
: In message <1565.1145648380 at critter.freebsd.dk>, "Poul-Henning Kamp" writes:
: >In message <20060421.133311.35013386.imp at bsdimp.com>, "M. Warner Losh" writes:
: >
: >>Can't you recover the time by tracking multiple chains and using the
: >>relationship between them to come up with at least small number of
: >>possible seconds?  Hasn't the US started broadcasting data that can be
: >>used to know which second you are at?
: >
: >This is called "Time Of Coincidence" and it works generally OK.
: 
: I should add that "TOC" is an overloaded term here: it's used
: both about when a chain coincides with a UTC second and when
: it does with another chain.

I almost corrected you here, since I've only seen TOC used as the
former term, rather than the latter term...

: All chains started at the top of the second 1958-01-01 00:00:00 UTC
: so you can just calculate forward using the gri.

Right.  Each chain gives you a set of possible answers for any given
phase relationship to a PPS.  Multiple chains give you a set of
overlapping answers, which is a subset of a single chain.  With enough
chains, or if the chains have a very long time between their time of
coincidence.  With the introduction of the loran data channel, you can
get this information with only one chain to a sub-microsecond level,
and often better....

I should say that in my day job, I work for Tom Celano who coauthored
the paper posted here recently, and helped to develop the timing
system that we delivered to USCG.  I'm not in the loop directly to
know when this equiptment is put into service...

: Remember leapseconds.

Also recall that there's no leap seconds before 1972 in the loran
timescale.  It followed the rubber second practice up until that
point.  After 1972, however, all seconds were the same, and the 10s of
accumulated difference between TAI and UTC were just ignored.    Since
1972, LORAN-C's timescale has been TAI + 10.

Warner




More information about the Time-nuts_lists.febo.com mailing list