Re: Which clock to trust?

Frank Reed (f-reed@metabien.com)
Sat, 19 Sep 1998 13:26:15 -0700

On Thu, 17 Sep 1998 09:25:00 +0200 Marco_Hahn@sl.maus.de (Marco Hahn) wrote:

>If there is a software that automatically adjusts the PC clock and evaluates
>the round-trip dealy, please let me know.

There are many computer programs that do that - follow the links from:


http://www.eecis.udel.edu/~ntp/software.html

I use SocketWatch on my PC and it seems to do a good job.  It's shareware
with a $10 registration fee.  I have it set to update the computer clock
every hour.  On request, it will report the status of the latest update,
which include the amount by which the computer clock was corrected.  These
magnitude of the corrections typically are a few tenths of a second.
Presumably, this means that the error of the updated local clock is of a
similar magnitude.

Further information about SocketWatch can be found at:


http://www.locutuscodeware.com/

Following is a copy of the status screen from Socketwatch:

-----------------------------------------------

Time Server Status:
[norad.arc.nasa.gov] - Synchronized
 - Stratum:1
 - Reference Clock:WWVB
 - Roundtrip Delay:0.295 seconds
 - Server Score:295
 - Times Used:1

Last time correction:
+0.45 seconds @ 9/19/98 12:49:49 PM.

-----------------------------------------------------------------

The only gripe I have about SocketWatch is that it doesn't keep a log of
the clock corrections it has made.  Such a log would be useful in analyzing
the accuracy/consistency of the clock updates.

Does anyone know how these programs handle the round trip delay?  An
obvious choice would be to assume that the delay in each direction is 1/2
the round trip delay - is this a valid assumption?

Note that there is an additional potential error in the local PC clock
update arising from the (somewhat unpredictable) delay of the operating
system (Windows 95, or ??).

- Frank

-- Frank Reed
Scottsdale, Arizona, USA  111.898W  33.484 N  1227 feet