In order to minimise the temperature dependency of my system clock I made a thermostate with an uA723, and soldered it on the 14.31818 MHz reference crystal of the system clock PLL.
The temperature of the thermostate converges to approx. 50 degr C and the results are really good for such a simple solution. My jitter decreased with an order (approx 10 times). With FreeBSD I have 'pps stabilities' of 0.001 - 0.003 ppm, whereas 0.020 - 0.050 ppm was observed without the thermostate. The PLL offset remains within +/- 1 usec most of the time for my ntp server ntp.remco.org. For graphs: see http://remco.org/difference.php Remco _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
