Re: [gpsd-dev] PPS glitch - Venus chip

[Gary E. Miller]

Yo Hal!

On Sun, 31 Jan 2016 00:15:12 -0800
Hal Murray <address@hidden> wrote:

> I've got a scope on it.  The serial stuff takes about a 1/2 second
> starting at the PPS pulse.  Usually it starts 2-20 ms after the PPS,
> but occasionally it starts 10-20 ms before the PPS.  

Ouch. gpsd makes the assumption that the PPS is always first in the cycle.
This is somewhat mitigated by gpsd taking the time stamp of the first
time based sentence after getting the checksum.

But it throws a wrench into time measurement.  Either the PPS is getting
delayed 10-30mSec or the serial sentence is starting 10-20 mSec too
soon.  Either one is bad and would vilate causality in the univers if
true.

Can you see in your sccope if the first sentence completes befoe the 
PPS edge?

> Another worm for the can:
>   I think you are using the wrong edge of the PPS.  It's about 4 ms
> wide.

gpsd uses the leading edge.  It does not ccare about polarity.  For
a once per second PPS (there are many kinds), gpsd checks that the
previous edge of the same polarity was close to 1 second ago, and the
previous edge of the opposite polarity was more than 500 mSec ago.


RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

pgpe5Nwgq_4kB.pgp
Description: OpenPGP digital signature