Re: A puzzle with internet time and NIST time
On 10/16/2012 02:07 AM, Paul E Condon wrote:
On 20121015_214840, John Hasler wrote:
Paul E Condon writes:
Ideas?
Run cronyc and post the results of the "tracking" and "sources"
commands.
--
John Hasler
I've now switched to chrony. The offset between 'atomic clock' and
Gnome clock display remains greater than 15sec. Its hard to be more
precise because I can't get the clock and the computer in my field of
vision simultaneously. But, no way do they agree to anything like
under a second. I do have an always on connection to the internet, and
I did take care to remove the 'offline's from the chrony.conf.
root@big:/var/log/chrony# chronyc tracking
Reference ID : 204.235.61.9 (name1.glorb.com)
Stratum : 3
Ref time (UTC) : Tue Oct 16 05:47:52 2012
System time : 0.000000121 seconds fast of NTP time
Frequency : 190.723 ppm fast
Residual freq : -6.591 ppm
Skew : 2.035 ppm
Root delay : 0.099491 seconds
Root dispersion : 0.186323 seconds
root@big:/var/log/chrony# chronyc sources
210 Number of sources = 4
MS Name/IP address Stratum Poll LastRx Last sample
============================================================================
^* name1.glorb.com 2 8 191 +1029us[ +950us] +/- 90ms
^+ d7.hotfile.com 2 8 184 +13ms[ +13ms] +/- 62ms
^? lttleman.deekayen.net 0 10 10y +0ns[ +0ns] +/- 0ns
^+ vpn.cumquat.nl 2 8 177 +17ms[ +17ms] +/- 103ms
root@big:/var/log/chrony#
If you need very accurate time on the computer, you might want to
get a GPS receiver and download time from the satellites. (There must
be software somewhere on the 'net that will decode the GPS output.)
You'll need a serial port for the GPS data. The GPS antenna must be
outdoors.
A way to tell which source you're looking at is correct would be to
listen to WWV at 5, 10, or 15MHz on a short-wave receiver. You should
be able to receive one of these frequencies most any time of the
day or night. You'd need at least a moderate antenna, altho depending on
where you are, a whip on a short-wave receiver might work for one
of the frequencies, some time of the day.
At one time, you could get a time signal over the phone from the
Bell Telephone co.--now Verizon. Don't know if that's still possible.
I don't know if any of the time beeps on the broadcast radio stations
are accurate.
A receiver for WWVB at 60KHz that would decode the signals would be
as accurate as anyone could want. That's the signal that your "atomic"
clock receives, so the clock *should* be accurate. You might only be
able to receive the signal in the nighttime hours, like the clock.
Typically, the clock will sync up at about 2AM local time, and a WWVB
receiver would basically do the same thing, with an accurate crystal
oscillator as an internal reference to keep time when no signal is
being received. Such a receiver would probably cost over $1000,
at a guess, but that's the difference between a scientific reference
and a $20 "atomic" clock!
--doug
--
Blessed are the peacekeepers...for they shall be shot at from both sides. --A.M. Greeley
Reply to: