Phil Kane wrote:
Klystron wrote:

Wouldn't it make more sense to include WWV and WWVH along with WWVB?
Are you familiar with the Internet-based ntp system? Then, there is the
matter of GPS, which has a time capability that is incidental to its
navigation function.


Want some fun? Compare the time ticks received from WWVB, WWV,
NIST-on-line, and GPS. What, they are not all simultaneous? Welcome
to the real world. GPS time does not correlate with UTC by any means
(several seconds difference).

In one of the first digital military command and control system that I
was involved in during the early 1960s, we used rubidium standards at
our switching centers to get accurate time synchronization, and even
then it was rather crude because the line delays varied so much. HF
propagation (WWV/WWVH) is even worse in that regard.


My understanding is that ntpd can handle that problem quite well. An
OPTIMAL setup would involve 1 computer per radio, each acting as a radio
controller (also called a strata 0 server). You could have a radio for
WWVB or WWVH, a second radio that is set to scan the WWV frequencies and
a third "radio" for GPS. Those 3 computers would connect to a fourth
computer that would act as a strata 1 server. The result would be a time
server that is as accurate as if it were connected to other ntp servers
via the Internet. Such an arrangement is sometimes used by firms that
need metrology-grade time service on a secured, internal LAN.
By the way, do not be put off by the expense of the four (or more)
computers described above. According the ntp documentation that I have
read, they need to have at least 100 MHz processor speeds for optimum
accuracy, but there is no benefit in going much above 100 MHz. Thus, a
pile of old, junkyard computers will do the job quite well and at an
aggregate cost of $20 to $100 in total.

--
Klystron