Re: ✘PX1172RH_DS performance

[Greg Troxel]

"Gary E. Miller" <gem@rellim.com> writes:

>> > Actual 2D error:
>> > Actual Alt error:
>> 
>> It is not really error, but a meaure of variance.
>
> It all depends on how you define things. "error", as in Circular Error
> Probability (CEP) was defined by other people before I got a chance to
> complain.  gpsd uses the terms and equations from RCC-261-00/

OK on RCC-261-00 but there is still an important difference between
spread relative to average and spread relative to the correct answer.

>> I am 99% guessing
>> that you don't have a mm-level position for the APC,
>
> Uh, what is APC?  

Antenna Phase Center.  The physical location whose position is actually
determined by a GNSS receiver in the most straightforward setup.  On
better antennas, this is specified relative to an Antenna Reference
Point (ARP), which is something you can physically access to measure.
Typical ARP is horizontally the center of the 5/8 thread and the
vertically "Bottom of Antenna Mount".

Getting more hard-core, there are antenna calibration files and the APC
varies with satellite elevation and sometimes also with azimuth.  The
antenna thus also has a North Reference Point (NRP) which should be
pointed to true north, so that the azimuth data can be used.  However,
some antennas have elevation calibration only.  An example is the
ANTCAL2B from Ardusimple.

>> One could, for a
>> fixed roof antenna, do
>> 
>>   raw observations for 24h and use PPP (both?)
>
> Done that, not with this part.

So you could plot that alongside these values and see if PPP and the
center of the RTK line up.

>>   raw observations for 24h (same bits) and use OPUS
>
> OPUS is not available where I live.

Really?   OPUS takes raw L1/L2 files and calculates positions based on
NOAA CORS Network stations.  I am 99% sure it works anywhere in CONUS.
Maybe they are too far away to get really good positions.

>>   Use F9P with ORGN baes
>
> No F9P in this setup.  I just use ORGN as the fixed base.  I'm not gonna
> worry about improving on 0.007 m for a while.

My point is that 0.007 m is relative to self average and you don't know
how accurate that average is.  If you have the antenan and cable you can
swap in the F9P and log raw for 24h, while also logging output from
RTK.q

>>   Use PX1172RH_DS with ORGN baes
>
> Which is what I did.
>
>> and compare them all.
>
> So much to do, so little time...

Sure, your time, your call.  I am just cautioning not to treat
RCC-261-00 "error" as error, when it is a measure of self consistency
instead.

> Thus the need for the roof mount.  Which makes it not very practical
> as a mobile application.

I haven't tried it on a car yet, but walking around in places without a
lot of trees works well, with MaCORS.

>> And,
>> one's base having more constellations helps make it work better.
>
> ORGN is just GPS and GLOBASS.  This part can use BeiDou, GALILEO, etc.

MaCORS has 4 constellations.  Having more satellites really helps.
I suppose you could run your own base with an F9P.


Anyway, very cool what you have managed to do and improve the code -
thanks for posting about it.

signature.asc
Description: PGP signature