[time-nuts] DCF77 Analyzer/Clock

Poul-Henning Kamp phk at phk.freebsd.dk
Thu Mar 23 04:18:29 EDT 2017

In message <20170323041013.GA4678 at panix.com>, Ron Bean writes:


He/They are missing out on a very big S/N improvement for DCF and
similar signals.

In reality only a very small subset of possible 2**59 timegrams are
potentially valid.  For instance day of month cannot be > 31.

Even fewer timegrams can follow each other, for instance the day of
month cannot change unless a lot of other fields also change,
or conversely, a lot of fields can only change if the hour
field changes to one specicfic value.

I used this in my old NTPNS dcf77 decode, with the result that it
can lock on to DCF77 in a matter of minutes, even when most of the
pulses are mangled.

Source code:


Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the time-nuts mailing list