RTK Facet off in all directions

GPS/GNSS Development Boards
1

For a while, I’ve been trying to figure out why measurements taken with my RTK Facet don’t match The Norwegian Mapping Authority’s survey points. At first, I thought the problem was with QGIS and the conversion from coordinates given in the Norwegian SOSI format, then I thought it might be how SW Maps handled UTM 32 or perhaps some internal some rounding errors. The RTK Facet is running version 2.0 and the ZED-F9P is updated to version 1.32.

For one of the survey points in Viken county (near Lillestrøm), the X axis was off by 1,337 m, the Y axis by 2,651 and the Z axis by 2,778 meters. (The X-Y vector is 2,97 m.)

For another point in Telemark county, the X axis for two measurements was off by 1,405 m and 1,158 m, the Y axis by 0,924 m and 0,815 m and the Z axis by 3,141 m and 2,735 m. (The X-Y vector is 1,68 m and 1,42 m.)

Today, I got the chance to play with a Septentrio Altus NR3 survey unit. With both units lying side by side, I could see the UTM 32 (and WGS 84) coordinates of the two units were different. I decided to connect to the Altus NR3, take some measurements, move the RTK Facet to the same spot (well, give or take a cm or so as I didn’t mount the units on a 5/8" bolt but eyeballed the position), take some measurements and then switch back to the Altus NR3 to take yet more measurements. This way, differences in satellite positions should be averaged out. This ended in 39 measurements by the Altus NR3 and 61 measurements by the RTK Facet. Both units had RTK Fix during the measurements. I have access to the Altus NR3 a few more days and could check again if needed.

The standard deviation of the RTK Facet measurements (using Excel STDEV.P) is for the X axis 0,005647297 m, for the Y axis 0,015817795 m and for the Z axis 0,017680326 m. For the Altus NR3, the deviation is for the X axis 0,003643553 m, for the Y axis 0,003870691 and for the Z axis 0,006698641 m.

I used the median of the measurements for both units, and they differ in the X axis by 0,45 m, in the Y axis by 0,604 m and in the Z axis by 0,259 m. (The X-Y vector is 0,75 m off.)

(*Maths was not my best subject, so I might have messed up. That’s why I’m attaching the zipped xlsx file…)

My main concern is that while the RTK Facet claims 14 mm accuracy and the Altus NR3 claims 11 mm accuracy, they differ by 75 cm and that’s well beyond both 11 and 14 mm. Also, in this case given my $700 RTK Facet’s history of being off by quite a lot from the government survey points, I lean towards trusting the €8750 Altus NR3 over the RTK Facet…

Had the error been a systematic “off by 30 cm in the X axis and 45 cm in the Y axis”, a software flaw could be assumed to be the culprit. Random difference in any direction? Meh.

So, to the question: How can I get my RTK Facet to perform as expected? Are there any ZED-F9P settings that should be (re)set, should I reset the RTK Facet, (re)load some settings or send the unit back as faulty?

Best regards,

Ola

Altus NR3 vs RTK Facet.zip (18 KB)

2

what software did you use for measurements, did you use rtk corrections from a local station? try to uncheck the sbas constellation from the settings. it matters and the measurement software try to use a demo from field genius, or surpad 4.2 are much better software than the free ones, set the location for the country and don’t forget its indices and the L1 height of the antenna if you use altimetry for geodetic measurements, give a sign if you still have questions, and if you leave the settings on 4 hz yes and at the first 4 nmea settings everywhere, it will go more smoothly, it took a while until I got caught

3

Hello Gabriel, thank you for answering.

I’ve used the Android app SW Maps for the measurements as recommended by SparkFun. I’m not sure how you mean paid software are “better than the free ones” - we’re talking about recording the coordinates the GNSS receiver output, and I’ve been using the same app for all measurements. The antenna height has been set to 0 m for both receivers as I’m more interested in their relative accuracy than the absolute (in other words, the test was set up to see how the two receivers compared to each other, not to get the exact coordinates of the randomly placed pole.) I also re-did the measurements the following day where I fixed the survey pole level and used the 5/8"-11 threads to make sure the two receivers were in the exact same spot.

I’ve factory reset the RTK Facet several times so it is set to 4 Hz, as well as reset the logging settings so they should be good.

The Altus NR3 got its RTK corrections from a preconfigured NTRIP server, which I’m confident gave the correct corrections as my employer paid a lot of money for that unit to be survey-grade accurate.

I ran a series of 200 static measurements for each of the receivers during the morning (first at 8:52, the last at 14:12) where standard deviation (STDEV.P) for the Altus NR3 were for the X axis 0.004521148 m, for the Y axis 0.010269166 m and for the Z axis 0.008926185 m. For the RTK Facet, the numbers were for the X axis 0.016423456 m, for the Y axis 0.07353906 m and for the Z axis 0.096365606 m. In other words, quite consistent for each receiver.

However, when compared to each other, the difference in the X axis were 0.471825 m, for the Y axis 0.613725002 m and for the Z axis 0.25264 m. Not exactly 14 mm accuracy in any dimension.

To rule out misconfiguration of the RTK base I’ve used for the RTK Facet I also ran a series of 118 DGPS measurements with the RTK Facet showing 160 mm accuracy. The standard deviation for these were for the X axis 0.113735467 m, for the Y axis 0.206843938 m and for the Z axis 0.395467193 m. However, when I compare these DGPS readings to the 200 Altus NR3 readings they are 0.574637034 m off in the X axis and 0.947590421 m off in the Y axis, a vector of 1.108212672 m difference. This tells me the claimed DGPS accuracy for the RTK Facet of 160 mm cannot be true.

I’m a hobbyist user - I planned to use the RTK Facet to map the drainage pipes I’m going to lay this summer, to survey a farm road I’m going to build and to map buried cables, drainage manhole covers etc. I’m a believer in open source software and hardware. I could justify buying a RTK Facet as the schematics and software is available to modify (perhaps not by me, but by the community - or by someone paid to fix problems) long after commercial units are deemed “EOL” by the producer wanting to sell a new model. I could afford a commercial model, but not really justify spending money on a device I’m not making any money off.

Still, at this point, I’m considering what to do. The RTK Facet I have in hand claims to be accurate but is not. All the measurements I’ve made since I got it is potentially wrong. Not by a lot, but up to a few meters. I could accept that from a consumer-grade GPS ten years ago, when 2.5 m was great - but SparkFun says “the RTK Facet is your one-stop shop for high precision geolocation and surveying needs” and my unit is not.

Sorry for the long rant, I rather like the RTK Facet, but I’d just wish mine was doing its job like it’s supposed to.

4

To answer you in fewer words I am a geodesic, cadastral, and topograph engineer, I am not the Open Surce enemy, but when you try comparison 2, equipment, you do them in the same conditions, soft, NTRIP connection, settings are different from equipment To another, it matters a lot as you do the settings, the commercial software have imlemented for each equipment the specifications of Antenna L1, L2 ARP OFSET, etc. And I was circumspect and tested with other more expensive equipment, Topcon, Sokkia, South, Emlid, Hi Terget, which are professional equipment, which are usually delivered with dedicated software , it is a market economy after all, I used And equipment from another company in Europe Tot Open Source, which also uses ZED F9 P from Ublox, Because I wanted to something like Emlid RS2, the manual you can find on the site at Ublox, the settings for Facet make them safer from the Window terminal point 2 GNSs massage and from here you change the settings and the rate of receipt, delete the Sbas, first one gives errors with the ublox and you may have points where you should not : viewtopic.php?f=116&t=57714, and if you use NTRIP you will not go with more stations 35 km long, although I have tried with stations from 60 km with errors in RTK up to 1, 2 cm acceptable for extravilan urban lands, if you do not know how to set your tool, without upset this, and without it. Use a geoid specialized in your country that commercial programs have , yes it is not a brand equipment, but it works acceptable if you know how to go. success

and my English pond excuse is not a native speaker

5

Hi gabriel_r - I am very appreciative of your help. Your english is quite good! Please continue to help where ever possible.

try to uncheck the sbas constellation from the settings

Why do you recommend turning off SBAS? This is easy to do from the GNSS configuration menu (https://sparkfun.github.io/SparkFun_RTK … gure_gnss/).

Hi olas - we are watching this thread closely. I appreciate your candor and your technical abilities. I’ve also brought up your issue with my contacts at u-blox who are somewhat mystified. u-blox does a tremendous amount of very complex testing against competitor units (as you mention Topcon, Emlid, etc). While they can’t share the data with me, they are quite serious the unit should be performing as stated. So let’s get to it.

You mention ‘Both units had RTK Fix during the measurements.’ How is the Facet obtaining its corrections?

Altus NR3 got its RTK corrections from a preconfigured NTRIP server

I will assume both the NR3 and the Facet are obtaining RTCM corrections from the same NTRIP caster you mention. Correct me if I’m wrong. How long is the baseline to the correction source?

I have not used the Septentrio Altus NR3 before. Are you confident it’s outputting NMEA in WGS84?

6

Clearly u-blox would not be selling high precision receivers in millions, if their position is off by tens of cm. I am very confident that this is a issue of your setup or the calculations you make. u-blox is also used in academic reseach as a reference in many publications

My guesses are:

  1. poor data from correction source. What are you using, is it a commercial service, is the base station close (<10km) and was the base accurately surveyed in.

  2. Missmatch in datum format between the receivers you compare

  3. Inaccurate datum or position conversion formula (e.g. you use an approximation)

  4. math/excel skills or poorly parsing the data, to be honest it is a bit unclear to me what you do with the data until you compare it. Maybe start comparision tge position with a known WGS84 lat lon coordinate.

7

Hi mazghc,

You are very likely correct, I’m an amateur and probably do something (or perhaps all of the things) wrong.

  1. The correction sources I’ve used might be the problem. That’s why I also made some measurements using DGPS without corrections to compare the difference of a trusted RTK device. I’m in the process of setting up my own base station which will be surveyed in using PPP.

  2. Both receivers were set to UTM 32.

  3. I’ve used the conversion formula in QGIS as well as for the state surveyed-in points.

  4. Probably both. That’s why I attached the spreadsheet so others could help me figure out what I’m doing wrong.

I have one of the Norwegian Mapping Authority’s survey point a few km away, and one day with a few spare hours I might use the RTK Facet to record some data for PPP-processing.

Other than that? Well, the feeling of “I must be doing something wrong but don’t know what or how” is a bit disappointing.

8

Hi olas,

Don’t worry - we’ll help you get it figured out. I welcome comparisons to much more expensive equipment.

using DGPS without corrections

Let’s leave comparisons between units without corrections aside for now. The goal is to get to RTK fix.

Both receivers were set to UTM 32.

I don’t believe UTM 32 is possible with the Facet or the ZED-F9P receiver. How did you achieve this?

For now, I would highly recommend either setting up a fixed base or establishing your correction source is good.

9

Fra det jeg har søkt i Norge og riktig hvis jeg tar feil, har du som offisielle koordinater flere geodesiske anslag er sannsynligvis avhengig av interesseområdet, NTM Sonne fra 5 til 30 på WGS84 eller NGO ellipsoid/NGO -soner og Lavii på Geoid Bessel endret, jeg brukte ikke qgis ærlig, men fra det jeg så har du muligheten til å sette og konvertere den fungerende geoiden, under hensyntagen til din lokale projeksjon. Overfor å ha en Zed F9 P -arbeid innfødt med WGS 84, gjør du innstillingene dine i henhold til ditt område, slik at du ikke kan gjøre målingene, jeg anbefaler at du ikke bruker QGIS for komplekse arbeider, prøv dette emnet: Dette emnet:viewtopic.php?f=116&t=57911

10

oversatt fra rumensk

11

Please use English wherever possible.

Translated:

From what I have searched in Norway and correctly if I am wrong, you have as official coordinates several geodesic estimates are probably dependent on the area of ​​interest, NTM Sonne from 5 to 30 on WGS84 or NGO ellipsoid / NGO zones and Lavii on Geoid Bessel changed, I did not use qgis honestly, but from what I saw you have the ability to set and convert the working geoid, taking into account your local projection. To have a Zed F9 P work native to WGS 84, make your settings according to your range so that you can not make the measurements, I recommend that you do not use QGIS for complex works, try this topic: This topic: viewtopic.php? f = 116 & t = 57911

12

Jeg beklager at SurveyMaster ikke har NoVegia i datalisten, men i FieldGenius hvis du ikke har noe imot å gjøre 30 dekkprosjekter på arbeidet og deretter gjøre et annet punkt for ditt sted i Norge. En hyggelig kveld og å vite at ingen ble født klokt, vi lærer alle noe nytt hvis vi er åpne for det. https://helpdesk.microsurvey.com/index. … /view/1665

13

I apologize the message was for Olas maybe in the native language I will understand easier and I find it hard from Romanian in English I wanted to give a tip for him in his native language

14

I would suggest to try to isolate the probkem somehow:

In a first test can you try to make a conparision of the coordinates in WGS84 with both the F9P/RTK Facet and your reference geodetic receiver. You can do this in the Lat/Lon/Alt or X/Y/Z coordinate system but make sure both are RTK fixed. If this difference is small few cm/mm then you already know that the hardware, your correction source and antenna work well.

15

Hi mazgch,

Unfortunately, I just visited the project where they have the Altus NR3, but I’ll try to get access to a known good correction source and check if the problem has been a bad RTK base all along.

But first, some vacation. :slight_smile:

Best regards,

Ola

16

sparky:
Hi gabriel_r - I am very appreciative of your help. Your english is quite good! Please continue to help where ever possible.

try to uncheck the sbas constellation from the settings

Why do you recommend turning off SBAS? This is easy to do from the GNSS configuration menu (https://sparkfun.github.io/SparkFun_RTK … gure_gnss/).

Hi olas - we are watching this thread closely. I appreciate your candor and your technical abilities. I’ve also brought up your issue with my contacts at u-blox who are somewhat mystified. u-blox does a tremendous amount of very complex testing against competitor units (as you mention Topcon, Emlid, etc). While they can’t share the data with me, they are quite serious the unit should be performing as stated. So let’s get to it.

You mention ‘Both units had RTK Fix during the measurements.’ How is the Facet obtaining its corrections?

Altus NR3 got its RTK corrections from a preconfigured NTRIP server

I will assume both the NR3 and the Facet are obtaining RTCM corrections from the same NTRIP caster you mention. Correct me if I’m wrong. How long is the baseline to the correction source?

I have not used the Septentrio Altus NR3 before. Are you confident it’s outputting NMEA in WGS84?

Disable SBAS, to avoid position jumps during short RTK corrections interruptions.