Re: Date/Time accuracy (was Re: First TOM1 image under Rob's Software)

[Chris Albertson <chrisalbertson90278@yahoo.com>]

The best write up is here.
  http://www.eecis.udel.edu/%7emills/exec.html

The quick answer is that adjusting the rate can be much more
sensitive then simply syncing the time.  For example if I can
synce to within 1/100th of a second then wait 100,000 seconds
and do it again then I can determine an error in clockspeed
to better then 1ppm.  NTP does the above as a continous process
using multiple souces of standard time.

Once you have to local clcok's _rate_ controlled you can
adjust the absolute time based on a running weighted average of
many thousands of synchronization attampts

NTP also continously monitors the process and keeps "tuning"
the varous parametrs so it can adapt to chances in the network
and the varing  quality of and connectivily to the available
sources of time.



--- "Robert J. Bradbury" <bradbury@aeiveos.com> wrote:

> 
> Chris,
> 
> Re:
> > NTP is about three orders of magnitude better solution then the
> > above.  The problem with "syncing" cocks is that periodically
> > time "jumps". (It might even "jump" backwards.) This is seriously
> > not good if you happen to be running an interval timmer off the
> > system time.
> 
> Can you explain why NTP is so much better than nistime?
> I understand the problem with the internet delays but I don't
> see how you can work around them unless you have a GPS rcvr
> connected to the time server.
> 
> Thanks,
> Robert
> 
> 


=====
Chris Albertson
  Home:   310-376-1029  chrisalbertson90278@yahoo.com
  Cell:   310-990-7550
  Office: 310-336-5189  Christopher.J.Albertson@aero.org
  KG6OMK


		
__________________________________ 
Do you Yahoo!? 
The all-new My Yahoo! - What will yours do?
http://my.yahoo.com