> >> I note that Dick Smith are offering a GPS receiver that is M$
> >> compliant, and is around the NZD250 mark. Is there any support for it
> >> under linux

These things always have a serial port -> trivial. The protocol they speak
is NMEA-0183, http://www.nmea.org/pub/0183/ unfortunately an expensive
standard. It's based on the exchange of ASCII lines of 15 - 80 characters
(usually). If you get a manual which describes which control messages and
info messages are supported by this receiver (should be possible), you're
home and hosed.

The serial port will give you a time accuracy of around 1 second and a
resolution of 1 second.

> >I've been looking at these as well.  I'd be curious to see what the
> >clock accuracy is like on one of these devices.
> 
> Allegedly 0.1sec according to the label. I might nip in next time the
> boss isn't around (:

Ooouuuuuuch, man that's bad. That's 100ms. Not a good NTP source. NTP can
synchronise the time over the internet between Europe and America to within
5-20ms, and this box is hanging straight off the back of your computer.
Cheap though, and you never have to set your watch. (I had this in 1985 in
Germany for less money than you mentioned - nahnahnahnahnah... ;))) )

Note the accuracy below 1s comes from a pulse signal which the receiver
should also supply. You can expect 1ms accuracy here. Good receivers should
manage <1us.

Some other trivia: GPS runs on neither TAI (international atomic time) nor
UTC (which one could call civil time) (no comment). You'll have to
compensate yourself. NTP is probably supposed to use UTC, but leap seconds
are not implemented. Whenever there is a leap second in UTC, (in the words
of David Mills, inventor of NTP) "the behaviour of NTP can be described in
terms of a pinball machine". No comment either.

Volker

-- 
Volker Kuhlmann                 is possibly list0570 with the domain in header
http://volker.dnsalias.net/             Please do not CC list postings to me.

Reply via email to