[ntp:questions] Ntp-statistics to adjust a radioclock?

Michael L. Semon mlsemon33 at verizon.net
Sat Mar 31 01:43:31 UTC 2007

Rune Magnussen wrote:
> Hi
> Is it possible to use the statistics gathered by ntpd to find the optimal
> adjustment for a radioclock?

It depends on where your already-calibrated time sources are located...

...on your local PC: Yes.

...on your LAN: You can get very close.

...on the Internet: Probably not.  You will get a result that reflects 
the the calibrations of the external servers and their connections to 
your PC; but that result may or may not be equal to the various delays 
from the radio station to your radio clock and ultimately to ntpd.

> Based on an estimate of the distance to the transmitter and some trial 
> and error I have managed to get my clock in pretty good agreement with the
> external servers I use but is there a way to improve that?
> Regard Rune

Yes.  Try the scripts from the ntpd source distribution.  They are 
located in the ./scripts/stats directory.  peer.awk and loop.awk are 
particularly useful.

When calibrating against servers, it's helpful to clear the statistics 
files once the reach of every server is 377 and the time offset has 
reached zero at least once.  After that, the longer that you let ntpd 
run, the better the results will be when you run the scripts.


More information about the questions mailing list