Re: A puzzle with internet time and NIST time
On 20121016_102703, Helmut Wollmersdorfer wrote:
>
> Am 16.10.2012 um 04:35 schrieb Paul E Condon:
>
> >I've been running Debian for many years. During most of those years I
> >have had a SkyScan(tm) 'Atomic Clock' on the wall near my Debian
> >desktop computer. The physical computer has changed over the years,
> >but not my using Debian, or my Atomic Clock. Until a few weeks ago,
> >they always displayed the same time, once I had correctly installed
> >the chrony or the ntp package. But some time recently they started
> >disagreeing by about 18 seconds with Debian/Squeeze running ahead, yes
> >ahead, of the Atomic Clock which is supposedly getting its signal by
> >radio direclty from the NIST transmitter. How can this be????
>
> I also like precise watches and have radio clocks in every room. Here
> in continental Europe the radio clocks usually receive the DCF77
> signal.
>
> As written by others in this thread the clocks only synchronize once
> a day to the signal. The rest of the day the clock is controlled by
> crystal quartz which has a precision of +/- 1 minute/year.
>
> My experience with radio clocks is that they have cheap electronic
> components failing to work after some time. In the last 15 years I
> had to throw away ~8 of them. Shortest lifetime was below 6 months,
> longest 12 years. In your case I assume that the synchronization of
> the clock isn't working any more since some months and it runs only
> under control of the crystal quartz.
>
> ntpd in the default debian configuration is usually accurate at +/- 2
> ms.
>
> You can use as root on the console terminal
>
> # ntpq -p
> remote refid st t when poll reach delay
> offset jitter
> ==============================================================================
> +dns2.teleport-i 73.120.242.92 2 u 562 1024 377 21.244
> -3.468 1.467
> -lswb-de-01.serv 95.211.148.1 3 u 746 1024 377 22.068
> 1.015 12.176
> +78-159-107-102. 193.171.23.163 2 u 753 1024 377 34.084
> 2.664 47.216
> *ntp3.rrze.uni-e .PPS. 1 u 403 1024 375 25.044
> -1.860 1.541
>
> Helmut Wollmersdorfer
I think you are mistaken about the sync only once a day. The signals
giving the year month and day are given once a day, but the precise
time-tics are given throughout the day. My clock has an indication
when it is receiving the radio signal which goes out when I move to a
place where the signal is weak, or when I change the battery and it
has not yet locked-in on the rhythm of the signal. Then, for a time,
the hour,minute,and second are good, but the date is wrong. Then after
a while, the correct date and day-of-week appear. I think the date
signal is actually more frequent than once a day, but definitely less
frequent than once a second. This is also the way that it is claimed
to work in the marketing materials.
The reason for my puzzlement is that the clock has always behaved in a
way consistent with the above --- for about twelve years starting when
I lived in California and continuing here in Colorado, where I now
live --- until recently when I noticed this 16sec offset. I have no
knowledge of international time signals. I'm sure there is world-wide
coordination, or perhaps the word is harmonization. Until quite
recently, it was very difficult to get information about the rest of
the world here in USA. Now more information appears to be available,
but I discount most of what I read because it appears to be false.
--
Paul E Condon
pecondon@mesanetworks.net
Reply to: