[ntp:questions] DCF77 Problem

Rob van der Putten rob at sput.nl
Wed Feb 25 18:26:25 UTC 2009

Hi there

Rob wrote:

> No you should not touch the pulse length, it conveys the time information.
> 50 baud is the correct setting for the port.

If missing a stopbit on each '1' is OK.

> There have been various problems in the parse driver that cause things
> like trash being written to the logfile, ntpd exiting on bad input, etc.
> It will depend on the version of ntpd that you use,


> if you have any
> problems like that.  For some time, I had to run a "watchdog" process
> that restarts ntpd when it has crashed.

It doesn't crash. It keeps processing the other clocks.

> But lately this has not been
> a problem.
> It is normal to get things like this in the logfile:
> 25 Feb 18:50:00 ntpd[4540]: parse: convert_rawdcf: parity check FAILED for "-#-#---##----#----M-S----1-4p---81-p12-8--124--4---2--1---p_"
> 25 Feb 18:50:00 ntpd[4540]: PARSE receiver #0: FAILED TIMECODE: "-#-#---##----#----M-S----1-4p---81-p12-8--124--4---2--1---p" (check receiver configuration / wiring)

That's not the problem.
Once I have a line like above, it doesn't go back to normal DCF 
processing any more. Unless I restart NTPD.

> When it happens too often or all the time, you will need to find a better
> place for the receiver, or align its direction.

If have a scope and a rs232 tester connected. The signal looks fine. The 
pulse lengths are OK. It can count from 0 to 58 with the pulses and get 
a 2 s interval. The signal is so clean that I get a 1 or 2 µs jitter on DCF.


More information about the questions mailing list