In a message dated 2007/04/21, Paulc@........ writes:

> But will somebody please tell me why the clock accuracy is so important?
> It appears that events last multiples of minutes and longer.  Why is clock
> accuracy in milliseconds so important.
> Unless one is trying to "triangulate" or compare phase of signals.. I must
> be missing something.

Hi Paul,

       You use the accurate quake time to determine the epicentre, using the 
P wave records of several stations.
       You don't generally need millisecond accuracy unless you are 
monitoring a local volcano or a similar feature, when the time delays are less than a 
second.
       The timing problem is due to cumulative clock errors. 1 ppm = 1 sec 
every 11.57 days. It is relatively difficult to hold even a quartz clock to this 
accuracy over the ambient temperature range. If you have a 10 ppm drift, this 
translates into about a second a day. Most computers use the cheap 32 kHz 
watch crystals. These can have quite large temperature drifts.
       The reason that I suggested using abouttime, is that your computer 
reads the time signal and it then sends a 'ping' signal which is immediatly 
reflected by abouttime. This enables you to measure and correct for transmission 
delays. Before digital circuits were used on phone networks, the delays were 
small. Now the digital information packet can be delayed by other traffic. I have 
seen delays of NIST of over 3 seconds in Europe.

       Regards,

       Chris Chapman   
In a me=
ssage dated 2007/04/21, Paulc@........ writes:



But will somebody please tell m=
e why the clock accuracy is so important?

It appears that events last multiples of minutes and longer.  Why is cl=
ock

accuracy in milliseconds so important.

Unless one is trying to "triangulate" or compare phase of signals.. I must
be missing something.




Hi Paul,



       You use the accurate quake time to dete=
rmine the epicentre, using the P wave records of several stations.

       You don't generally need millisecond a=
ccuracy unless you are monitoring a local volcano or a similar feature, when=
 the time delays are less than a second.

       The timing problem is due to cumulative=
 clock errors. 1 ppm =3D 1 sec every 11.57 days. It is relatively difficult=20=
to hold even a quartz clock to this accuracy over the ambient temperature ra=
nge. If you have a 10 ppm drift, this translates into about a second a day.=20=
Most computers use the cheap 32 kHz watch crystals. These can have quite lar=
ge temperature drifts.

       The reason that I suggested using about=
time, is that your computer reads the time signal and it then sends a 'ping'=
 signal which is immediatly reflected by abouttime. This enables you to meas=
ure and correct for transmission delays. Before digital circuits were used o=
n phone networks, the delays were small. Now the digital information packet=20=
can be delayed by other traffic. I have seen delays of NIST of over 3 second=
s in Europe.



       Regards,



       Chris Chapman