[time-nuts] Histogram of T2 arrival times

Hal Murray hmurray at megapathdsl.net
Fri Jul 29 05:50:59 EDT 2016


> So, you have plotted offset versus temp?  Can we see that?

I don't have offset vs temperature.  This is what I work with:
  http://users.megapathdsl.net/~hmurray/time-nuts/LAN-clock.png
  http://users.megapathdsl.net/~hmurray/time-nuts/LAN-temp.png

Scan the wiki page for PID controllers:
  https://en.wikipedia.org/wiki/PID_controller

Consider the simple case (no I, no D).  If you have a constant rate of 
temperature change, there is a stable point with a constant offset.  The 
offset keeps adjusting the frequency (drift in ntp speak) to track the 
temperature.  The offset and temp derivative are related by the gain.

The graphs aren't great, but when the temperature is going up the offset is 
positive and the reverse.


> I would not be surprised if temp was dominant here.  But I can think of a
> few other things that may be dominant instead. 

Such as?


> I would only consider CPU temp a very rough proxy for XTAL temp.  Room
> ambient would be much closer to XTAL temp.

That box is fanless.  (There is actually a tiny fan under the disk.)  Mostly, 
it's not doing any work.

Room temp tracks CPU temp.

> Yup.  I have some USB thermometers that are good for such things. 

Model?  URL?


-- 
These are my opinions.  I hate spam.






More information about the time-nuts mailing list