[ntp:questions] Re: Good GPS for attic?

[David Schwartz]

"Chris Malcolm" <cam at holyrood.ed.ac.uk> wrote in message 
news:cum4vr$mn3$1 at scotsman.ed.ac.uk...

>>    Which of these works everyplace you have a view of the sky and can 
>> keep
>>time to an accuracy of better than one microsecond?

> While GPRRs have to keep time internally to within several
> nanoseconds, few of them bother trying to report it to the user with
> that degree of precision. For a start it's impossible over an RS232
> serial link. Up to a second out is not uncommon. You'd get more a more
> accurate time fix with a program that listened to radio time pips over
> a microphone!

    If you are purchasing a GPS receiver for timekeeping purposes, you 
should always buy one that's specifically designed (or at least rated) for 
this purpose. You are correct, RS232 is not accurate enough, you need at 
least a PPS (pulse per second) output. Many GPS receivers are available that 
have a PPS output that is guaranteed to always be +/- 1 uS of UTC. The best 
one suppress the PPS output if that level of accuracy is not available.

    I'm not suggesting doing this half-assed. I'm suggesting a well-designed 
solution. You can get a GPS receiver that has guaranteed timing 
specifications for less than $300. A low-end FreeBSD or Linux server to do 
the rest can be had for about $350 as well. This solution will give you +/- 
3uS of UTC accuracy 99% of the time.

    DS