[time-nuts] Sure Electronics GPS
michael.cook at sfr.fr
Mon Jan 17 08:45:46 UTC 2011
> The SureElectronics unit is general within 10 feet with no observed
> off the wall excursions.
> With that in mind, I wonder how accurate is their 1pps output?
There was some discussion on the MG1613S earlier. I ordered one off
the bay and did a few test.
I tried to post some pics of the PPS trace and my wiring mods and stats
for those who might be interested but they didn't make it through the
filters. Some details.
The board is configured as an NMEA device and functions fine on all
output interfaces available. You only get output at 9600 bps as standard
and according to the manual it can only be configured otherwise through
the serial interface.
I didn't try changing it though. There is a nice GUI to monitor it.
Works fine, but I haven't delved. Seems to allow sending NMEA commands
so there may be some useful config changes possible.
NMEA data output fixed at 9600bps, 8 bit, 1stop, no parity
It is very sensitive, pulling in double the birds that I can regularly
see from my unprivileged window sill , and also seems to be less
sensitive to reflections than my motorola timing GPSs. That was good
news as I have a big problem with the motorola receivers in that
respect. The VP/UT+ are not able to filter out the birds that are
detected out of line of site. I posted some time ago on that and still
have no solution (WIP).
When the unit is powered on but no satellites detected, output
sentences in order of reception are
GGA, GSA, GSV, RMC
no PPS output
When satellites are visible but no fix is available
GGA, GSA, GSV, GSV, RMC number of GSV may vary due to those in
view, but I always got 2 in this state
no PPS output
When a fix is available
GGA, GSA, GSV, ...., GSV, RMC
PPS is output
Although the manual indicates that GLL and VTG are supported, I did not
see either of these messages. I haven't tried manually configuring the chip.
I do not know whether it goes to position hold, but suspect not.
I scoped the PPS/NMEA stream output. PPS and. NMEA signal taken from
test pads on the edge of the board.
NMEA stream start is always at 300ms from the PPS . I guess that the
end is variable depending on the number of birds in view, but for me,
the stream ended at +720ms all through my test. So at least on this unit
the data does not leak into the following second. That delay can be
configured out in NTP. I selected just the GGA sentences for my test
with the generic NMEA driver and it works ok. Though as I have better
sources available I wouldn't use it to serve time.
PPS measurement - the PPS is positive going and 100ms long . The doc
I have been able to get at says that it is 200ms long. Anyway, what we
get is exploitable without pulse stretching over the serial interface,
once it is hooked up.
The signal at the test pad is not a very good square wave with respect
to the motorola receivers. It has a rise time of nearly 100ns and does
not settle to 3v (not 3.3) until 500ns or more. There is no jitter that
I can see (<5ns ?), but would I see any?. Still for my purpose , NTP ,
it was a possible candidate.
The PPS is not available on the DB9 connector. It is however a pretty
simple operation to get it as there are unused pins on the SP232ECN
driver chip. I connected the test pad to pin 11 and pin 14 out to the
DB9 pin 1 leg for a DCD input to the PC rs232 port. The RS232
transition is +/- 5V in 750ns.
However, although the PPS looks nice and steady on the scope , for some
reason I did not get offsets from UTC measured by NTP using the ATOM
driver that were as good as those from the motorola boards . I haven't
investigated why, as the major issue for me is that the board is too big
to fit in a Soekris case and can only be powered (standard hook up) by
the USB interface, and although the Soekris has a USB interface, the
size problem makes it a nono. I will probably hot glue it to a PC. I can
then use both the NMEA and PPS together quite reasonably.
More information about the time-nuts