[time-nuts] NTP Temperature Compensation

Chris Albertson albertson.chris at gmail.com
Thu May 16 18:28:16 EDT 2013

You are fixing a problem that doesn't exist.  The paper that showed error
plotted with temperature was from a system where the PPS was turned off.
 If you connect PPS the error vs. temperature goes away.  In other words
the pulse per second pretty much fixes the problem.

Now, for status 2 and higher servers that lack an attached GPS temperature
sensing might help a lot.   I think the best temp sensor is the fan RPM.
 But this is different for every computer.  So you'd need some sort of
configuration file.

On Thu, May 16, 2013 at 3:18 PM, Hal Murray <hmurray at megapathdsl.net> wrote:

> jason at extremeoverclocking.com said:
> > Since most modern PCs have at least one temperature sensor on the
> > motherboard, I wonder if it would be possible to monitor that to help
> > compensate for temperature changes. I realize the locations of said
> sensor
> > can be anywhere on the board, but you would think it could help out
> *some*
> > amount.
> Newer CPU chips have a temperature sensor on chip.  That's usually a long
> way
> from the crystal used for timekeeping.
> It might be possible to model the temperature of the crystal depending on
> the
> heat flow from the CPU and the fan speed and ambient air temperature.  Is
> the
> heat flow just a low pass filter? (probably not a sharp cutoff)
> --
> These are my opinions.  I hate spam.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.


Chris Albertson
Redondo Beach, California

More information about the time-nuts mailing list