Re: [gpsd-dev] A detail in your ntpoffset script

[Andy Walls]

Yup,  something like that.

Just do what any good tracking filter does with multiple independent measurements of the same thing: fudge them all towards a (weighted) midpoint and call that the answer.

You know the answer is wrong, but it has a minimum distance (in some sense) to all the possible candidates for the correct answer. :)

Regards,
Andy

Beat Bolli <address@hidden> wrote:

On 23.10.13 21:22, Gary E. Miller wrote:
Yo Beat!

On Wed, 23 Oct 2013 21:15:03 +0200
Beat Bolli <address@hidden> wrote:

Of course, if one wanted to fudge the PPS too, a similar procedure
would have to be done for unit 1 (one more requirement for ntptune!)

On one host I have two GPS/PPS.  They differ by a consistent 1.5 uSec.

It annoys me, so I fudge one of them so they agree.  Which is the 'wrong'
one and which is the 'right' one?  Flip a coin.

I'm starting to see what all this "time-nuttery" is about ;-)

The one that is further from a third reference clock?

Beat

-- 
Sent from my Android phone with K-9 Mail. Please excuse my brevity.