Re: RTK Base and its accuracy

[Florian Kiera]

Hey Gary,

I ran gpspipe for 24h and created a plot with it following https://gpsd.gitlab.io/gpsd/gpsprof.html.

Here are the results:

Once we have tested the rover outside of the building I will give an update to that too.

Regards
Florian

Am 16.08.22 um 21:41 schrieb Gary E. Miller:

> Yo Florian!
>
> On Tue, 16 Aug 2022 09:25:48 +0200
> Florian Kiera <florian.kiera@logicway.de> wrote:
>
> > Hello Gary!
> >
> > Am 11.08.22 um 21:42 schrieb Gary E. Miller:
> > > Yo Florian!
> > >
> > > On Wed, 10 Aug 2022 10:49:24 +0200
> > > Florian Kiera <florian.kiera@logicway.de> wrote:
> > >  
> > > > > Which are not precise enough for what you are doing.  Just look at
> > > > > what cgps is telling you.  That is much more "precIse" than some
> > > > > random wiebdite.  
> > > > >             
> > > > That pretty much seems to be the issue already... the internet
> > > > translators seem to have some issues with calculating the latitude
> > > > correctly. cgps worked fine and gave the expected position.  
> > > >           
> > > Then you missed my point.  They can all be correct, and different.  
> > >         
> > Yet they calculated Latitude somehow "wrongly". (wrong by multiple 
> > kilometers) I understood that it can have some deviations. Still I
> > would expect them to be in a meter scale and not kilometer. I would
> > like to stick with the fact that cgps does it right and I am happy to
> > use it.
> >       
> There are USGS grid scales that differ by up to 3 meters.  My
> expectations are not as optimistic as yours.  If gpsd, by way of cgps,
> works for you, great.  cgps does not cumpte anything, it just passes on
> what gpsd tells it.
>
> > >  
> > > > > > Latitude: 53.59931 11.41833  
> > > > > >               
> > > > > Determined how?  
> > > > >             
> > > > With the position of the rover.  
> > > >           
> > > I don't think you understood my question.  How do you know the
> > > "accurate" position of the rover?  
> > >         
> > Using a map. It can either be OpenStreetMaps or Google Maps. Both end
> > up in the same spot (roughly where our "repeater source antenna" is 
> > positioned) with the Lat/Long I provided. Also the base has nearly
> > the same coordinates as the base-antenna is close to the
> > rover-repeater antenna.
> >       
> There you go.  That is what "repeaters" do, they repeat what the repeater
> sees.  Not good for precision work.
>
>
> > > > Considering the base was not that wrong after all we can go back to
> > > > the RTK base-rover scenario.  
> > > >           
> > > How wrong?  How do you know how wrong?  
> > >         
> > cgps gave the right position and therefore hardly can produce
> > gibberish RTMC3.
> >       
> Ah, lost me.  cgps has no effect on the RTCM3 from your base.
>
> RTCM3 is gibberish by design...
>
> > A right position is determined in my opinion by
> > taking the Lat/Long to a map and actually end up in the position
> > where I actually stand at.
> >       
> Google maps are not known for their accuracy.  Go find a benchmrk.
> And the accuracy varies minute by minute, so you need to run gpsprof
> for many hours to determine your real performance.
>
> > In this scenario I expect Lat/Long values
> > which when used on a map end up close to where the antennas (base
> > antenna and gps repeater antenna for the rover) are positioned on the
> > roof.
> >       
> We already agree your repeater is messing up your measurements, and yet
> you still use it?
>
> > > > I start the survey-in as described in
> > > > the first mail and than start str2str from the RTKLib to push the
> > > > RTCM3 messages to the running ntripcaster.  
> > > >           
> > > Which can work, but there are easier wasys to get the same retuls.  
> > >         
> > Is it documented? I would always prefer easier ways when they produce 
> > the same outcome. We do want to use a NTRIP caster tho.
> >       
> Not docuemnted by gpsd, but it should be in the u-blox manuals.  Just
> read the raw RTCM3 from the base, and nectcat (nc) it to the rover.
> That is what most do for short range stuff.
>
> > > > > The latitude  
> > > > >             
> > > > on the rover was accurate but the longitude was a bit off (~20m).  
> > > >           
> > > Once again, you do not use "accurate" propertly.  I'm guessing your
> > > repeater antenna is 20 m away?  
> > >         
> > I understand that the position of the repeater is "repeated". So the 
> > rover should think it is at the position of the gps repeater antenna. 
> >       
> Sort of.  There are other effects in play.  Too many to correct for.
>
> > The antenna is like 4m above and 5m in a horizontal direction away.
> > The offset probably comes from the fact that the rover antenna
> > obtains values not only from the repeater but from outside the
> > building as well.
> >       
> Yeah, that will royally confuse your rover.  I have played with
> feeding 2 antennas into one GPS,  The results are not pretty.
>
> > The offset actually goes in the direction of the
> > windows. (shortest way inside the building) The rover thinks it is
> > 20m outside the building in that direction while it *should* think
> > its on the roof of the building.
> >       
> Then maybe you need to rethink "should" considering your data does not
> match your expectations.
>
> > We are going to test the rover outside the building soon and
> > hopefully receive what we expected all along.
> >       
> Hopefully.
>
> > > > I
> > > > used the latitude and longitude that gpsd gave out in google maps
> > > > which gave me the visual offset on the map.  
> > > >           
> > > Not exactly a high precision method.  
> > >         
> > If the Lat/Long isnt precise/accurate enough to be used on a map what 
> > are they for then?
> >       
> Depends on what you need.  If you are +/- 3 meters off 10% of the time,
> then that is perfectly good o navigate a ship or fly an airplane.  It is
> not good enough to driave a car in a lane.
>
> You need to quantify what you actually need.  Then  run tests, like with
> gpsprof, to see what you are actually getting.  Very few initially
> expect how bad a GPS position can be 2% of the time.
>
> > Simply imagining where I am? This is not about a 
> > single meter.
> >       
> Getting under a meter with 8-series will be very hard, and certainly not
> with a repeater invovled.  Your own data tells you this.
>
> > > > I will add the output.  
> > > >           
> > > Looking forward to it.  Be sure to run gpsprof for at least 12
> > > hours.  
> > >         
> > I will run gpsprof today. I already added the output of the "simple", 
> > probably wrong, gpsprof run to my last email tho.
> >       
> Sorry, if it was there, I missed it.  Dunno what you mean by "simple".
> Unless it runs modulo 12 hours, the data will be misleading.
>
> RGDS
> GARY
> ---------------------------------------------------------------------------
> Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
> 	gem@rellim.com  Tel:+1 541 382 8588
>
> 	    Veritas liberabit vos. -- Quid est veritas?
>     "If you can't measure it, you can't improve it." - Lord Kelvin
>     
-- 
###############################################################
Florian Kiera               Tel:    (0385) 201 000 58
Logic Way GmbH              Geschäftsführer: Andreas Loerzer 
Mettenheimer Straße 73      HTTP:   http://www.logicway.de
19061 Schwerin              email:  mailto:florian.kiera@logicway.de
###############################################################