[time-nuts] My Casio G-Shock watch and other fun stuff
shoppa at trailing-edge.com
Thu Jun 21 05:57:37 EDT 2007
Neon John <jgd at johngsbbq.com> wrote:
> I don't quite understand why they'd all run so close during the day after a
> sync but drift so fast if a sync was missed.
In a typical disciplined oscillator timeclock, the predicted error goes
like the square of the time since the last lock. "Next day" is probably
6 to 12 hours after the 2AM time sync. Two days later is 30 to 36 hours.
36 squared is 9 times bigger than 12 squared, so the predicted error
would be 9 times bigger for the day after a night where sync was missed.
Now, I cannot actually say that the Casios or any other brands actually
discipline their oscillator. I know that the cheap WWVB-locked wall clocks
are not disciplined - if they're running fast 2 seconds a day, then every
night they just jump back two seconds.
I've had $60-$100 (Pulsar, Bulova, etc.) quartz watches keep time
to within a second a month without any disciplining/tweaking since
they left the factory. Not all of them are that good, some run as
much as a second a day too fast or too slow. A second a day is 11ppm,
a second a week is 1.6ppm, a second a month is 0.4ppm. If a watch
is off by just a second a month, I don't set it except a couple
of times a year. If it's off by a second a day, then it gets set
several times a month.
More information about the time-nuts