Re: [gpsd-users] GPSD, PPS, Garmin 18x LVC and 93 meters of serial cable

[Sander Smeenk]

Quoting address@hidden (address@hidden):

> On 25/11/2014 8:47 PM, Sander Smeenk wrote:
> Hi Sander,

Funny. My original message must have been stuck somewhere en route
back to me. I thought it was never delivered to the list:
| Date: Tue, 25 Nov 2014 11:47:02 +0100
| Delivery-date: Sat, 13 Dec 2014 14:45:26 +0100


> > The receiver is connected to the server by 93 meters of CAT6 UTP
> > over which the GPS' serial outputs are patched/connected.
> Whoa! A cable length of 93m is WAY outside the spec for conventional
> RS232 (30 feet or 10m). The 18LVC doesn't use RS232 signalling levels
> (-30 to -3; +3 to +30) either - it outputs 'TTL' level 232 (0-5V) - so
> I'm surprised that you're getting any sane data at all, and the skew
> on your PPS pulse must be awful.

In the end it turned out to be a GND-wire which i thought was
connected/correct. The Garmin has three black wires (GND) but only two
are connected internally it seems. After resoldering the GND to another
wire i got stable serial data and a valid 1PPS signal.

The setup worked rather nice with GPSd through the 93m UTP, see these
graphs for the GPSd peers in NTP:  http://8n1.org/10173/997f
(if you're an astronomer or scientist these offsets/jitter might be
horrendous, but for the average joe it's quite nice)

Though, i haven't been able to get a proper sync on the NMEA data,
even though i tried to fudge the time offset as close as possible to the
average offset. According to the good folk on ntp-questions this might
be caused by the Garmin 18x, which aparently tries to push the NMEA data
precisely between the PPS signal pulse...

Anyways, i've now switched to NTP's GPS_NMEA (20) and ATOM 1PPS (22)
clocks and getting about the same accuracy with that.

Thanks!

-Sndr.
-- 
| Class trip to the Cola factory...  I hope there's no pop quiz...
| 4096R/20CC6CD2 - 6D40 1A20 B9AA 87D4 84C7  FBD6 F3A9 9442 20CC 6CD2