[time-nuts] 1pps; what good is it?

[M. Warner Losh]

In message: <20060721172027.5065ABDF6 at ip-64-139-1-69.sjc.megapath.net>
            Hal Murray <hmurray at suespammers.org> writes:
: > It also has a 1 pulse per second output; what is it good
: > for, in industry, and for me?
: 
: That 1 PPS signal ticks on the UTC second boundary.  That lets you transfer 
: accurate time as well as frequency to other boxes.
: 
: Suppose you want to build a time server.  If you connect up the serial port, 
: you can ask the Z3801A what time it is.  The answer won't be very accurate 
: (few ms) because of delays in the serial port hardware and software.  With a 
: PPS signal (and widely available patches to the OS) you can get that down to 
: microseconds.

You can likely get closer to nanoseconds, if you know the cable delay
from the on time point of the system.  We use the PPS signal in a
system at work to steer a three clock cesium ensemble down to the
nanosecond or so level.  But we have specially cut cables to make sure
that we measure the PPS from the cesiums at a known offset.  Going
below nanosecond level requires compensation for many other factors,
and time is likely better transferred using a higher frequency than
1Hz.

Warner