[ntp:questions] Re: Designing an NTP network for a research ships at sea...

Marc Brett marc at fordson.demon.co.uk
Mon Aug 25 00:06:48 UTC 2003


On Wed, 23 Jul 2003 13:03:20 -0400, "Val Schmidt"
<vschmidt at ldeo.columbia.edu> wrote:

>Thanks Everyone for the help.
>
>I have a few more questions.
>
>If I look at the timing errors involved in logging and time stamping NMEA
>position strings I might make an error diagram something like this:
>
>
>*UTC*
>   GPS Receiver/NMEA time error->     (~10ms)
>                      time to write serial data->  (~ns - maybe 1 sec)
>                            time to log serial data->  (~1ms)
>                                       UTC to CPU offset-> (~100 ms with NTP
>NMEA driver - no PPS)
>                                                            *CPU TIMESTAMP*
>
>Do the estimates regarding nominal errors from each of these sources seem
>reasonable to everyone?
>Are there other sources of error that I've missed?

One gotcha that you might want to keep in mind -- make sure the GPS
receiver is not assuming a static installation.  Some NTP drivers (eg.
Motorolla Oncore and MX4200) will set the receiver to an averaged
fixed position so it no longer has to navigate and can supply a
marginally more accurate time solution.  No good, of course, on a boat
that's moving around.

For really accurate timing of your instruments, IRIG-B might fit the
bill.

Marc



More information about the questions mailing list