[ntp:questions] dcf77 time synchronization problem

Wolfgang.Hennerbichler at ooe.gv.at Wolfgang.Hennerbichler at ooe.gv.at
Mon Aug 16 09:39:27 UTC 2010


> Von: 
> questions-bounces+wolfgang.hennerbichler=ooe.gv.at at lists.ntp.o
> rg 
> [mailto:questions-bounces+wolfgang.hennerbichler=ooe.gv.at at lis
ts.ntp.org] Im Auftrag von Ulrich Windl
> Gesendet: Freitag, 13. August 2010 13:21
> An: questions at lists.ntp.org
> Betreff: Re: [ntp:questions] dcf77 time synchronization problem
> Hi!

Hi and thanks for the reply; 
 
> A few basics:
> 
> Whenever DCF-77 receives a bad spike, it needs at least
> 120 - "time of the spike" to synchronize again. Maybe even one extra
> minute to be sure.

allright. thanks for that. 

> The DCF-77 antenna is directional, so you need to have it 
> turned in the
> right direction. Unfortnuately it's not easy to find the optimal
> direction.

hm. given that it only has three led's I guess it's gonna be a hard job
to check on the antenna. 

> For the clock hopping: Use "ntpq -p" to find out whether your 
> DCF-77 has
> a significant offset against the other servers. You'll ahve 
> to minimize
> that.

I don't know how how to interpret significant. Currently all the remote
timeservers have an offset of 36 or 37, wheres my dcf timesource has an
offset of 0.0something. Is that one significant? :) 

> I'd suggest to capture long-term statistics of the clock status. With
> the parse driver "ntpq -c cl" should display the statistics.

Thanks, I've been doing that already, but not with nptq -c cl. 

> Regards,
> Ulrich

I'll keep you posted. 
Wolfgang



More information about the questions mailing list