[FOSS-GPS] NSV08C

[Danny Miller]

I've been playing with the new NVS08C GPS, which Michele outlined in a blog:
http://michelebavaro.blogspot.com/
Who wrote an RTKLib branch to support it.

They're pretty remarkable.  I have 2 units with identical patch antennas 
(also have Sarantel quad-helicals if I need 'em).  These are cheap 
antenna I bought on eBay like a year ago, but they do seem to receive 
GLONASS sats nicely (the frequency is close).  One thing to note, the 
spec sheet seems to say it provides "2.65 to 2.8V" of antenna power, 
which may not be enough.  The reference schematic contains a resistor 
and cap pair needed to use voltages supplies from another rail (like 
3.3v power for a 3.3v ant).

Since they receive GLONASS as well as GPS, the constellation view is 
instantly quite huge, 20 sats isn't uncommon.  In fact, on RAW, the 
serial bandwidth is "significant" and from the look of it 115200 may not 
be sufficient.

The units' output is amazingly consistent.  With a pair of static patch 
antennae on a groundplane, after about 30 sec they stop moving and 
rarely move more than a few cm.  Given a couple of hours, you may see a 
track drift 1-2M, but it's a very smooth, noise-free track.  If you move 
it in a 1M square you'll see a ~1M square on the track.  That's without 
PPP or RTK, just watching the hardware solution.  But, they're off by 
about 1-2M from my "best result" from a PPP static fix, and 1-2M from 
each other's solutions, too.  But, they're HIGHLY consistent inside of a 
unit.  If I move the antenna under a metal shield to kill the signal and 
bring it back out, the solution will restart within about 0.5M of its 
last lock.

However, strangely, the RTK and PPP has so far been "off".  That is, 
when RTK DOES get to "Fixed", sometimes it's Fixed within a few CM, but 
early on I get a lot of Fixed off my like 2-4M, and most of the time I 
only have Float.  The Lea-6T worked more stable in RTK IIRC. It's odd 
because the simple non-RTK hardware solution can get better results, at 
least early on.

I don't know if there might be a problem in the pseudorange/carrier 
phase/doppler resulting in what seems like relatively poor RTKLib 
performance, or maybe Michele's config or application code has something 
strange, or maybe RTKLib has trouble with an unusually high number of 
sats?  (I guess I could retry that with GLONASS support disabled to go 
back to a more modest number of sats).

Danny