[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: Online atomic clock

William B Magnusson wrote:

>This recent thread caused me to wonder about the dynamic nature of the
>internet and variable propagation delays. Does anyone know the relative
>accuracy - at your computer - of "internet" timing services as compared to
>"hard-wired" timing services as are available in many cities in many
I believe with NTP you can get 1 to 50 ms accuracy.  The best NTP 
clients attempt to estimate long term drift of the system clock as well 
as propagation delays. They figure out by how much the system clock 
should be slowed down or sped up to keep in sync with the time server.  
The NTP client needs to be running for several days to stabilize and get 
within 50 ms range.    For a very high short term clock accuracy you 
would need a locally attached time reference, for example a gps 
disciplined quartz oscillator and an operating system with a kernel that 
supports a nanosecond time resolution.  Then you could probably get 
stability of about 1 us over 24 hours.

>Is it possible to estimate the order of the longest propagation delays to be
>expected via internet circuits? Would such delays be insignificant against -
>say - the short-term accuracy of the average computer clock? On balance,
>would the 'hard-wired' services be preferable in critical applications?
 The round trip time of an IP packet can be measured by sending the ICMP 
echo packets (aka ping).  NTP assumes that propagation delay is half the 
round trip time, this can be anywhere from a millisecond to hundreds of 
milliseconds.  But if the uplink and downlink paths are not symmetrical, 
the client has no way of figuring out the exact delay and not able to 
compensate correctly.  You can read more about this if you google for 
Network Time Protocol.   You might also find interesting info on this 
site about a relatively cheap (from ebay) HP time standard:  
http://www.realhamradio.com/GPS_Frequency_Standard.htm.  And if you want 
to see the ultimate precision time nut check out www.leapsecond.com
Sent via amsat-bb@amsat.org. Opinions expressed are those of the author.
Not an AMSAT member? Join now to support the amateur satellite program!
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org