Hi Richard,

I think, you made an error in your timer calculation. Since an error of 1 / 1513 (= 660ppm) seems very big for a crystal.
Typical crystal frequency errors are given in the data sheet with only 15...50ppm.

E.g. I use a 12MHz crystal and made the calculation for 12.000000MHz.
After 1 month the deviation was 30 seconds. This was corresponding to a real crystal frequency of 12.000139MHz. After repeating my timer calculation with this value, the clock works now right.


Peter