Michael Karas wrote:
-------------------------------
"And we all wear watches that give reasonably accurate behavior."

Ah! But that's because the crystal inside a wristwatch is essentially 'ovenized' ;-)

A typical 'watch crystal' has an initial tolerance of ±20 ppm @ 25ºC and exhibits a parabolic temperature coefficient of -0.038ppm/(T-T0)², where T0 is 25ºC (or 298K).

The observed inaccuracy of 2 minutes/day equates to almost 1400ppm, which can hardly be caused by any normal crystal behaviour. A software problem, i.e. inadvertant write(s) to the device seems a much more plausible explanation.


Rob.