Without buying a second RTK, is it possible to get the 300mm accuracy? Everything I do, I end up 600-850 mm off which is about 2-3x worse than the claim. I’ve put my process, my calculations, and screenshots in the attached PDF file.
Tessar.pdf (361 KB)
Hi,
Many thanks for reporting this. Just a heads-up that Sparky - our RTK expert - is away for a few days. He will reply once he’s back from the wild…
Best wishes,
Paul
Nice documentation, thanks for providing all the details. Looks like you’re getting a great view of the constellations.
Getting sub-meter (e.g. 850mm) accuracy with a DGPS (SBAS not RTK Fix) solution is pretty good in my experience. I’d be happy with the accuracy you’re getting. Sparky will need to speak to the 300mm vs. 850mm question. Unless you’re getting RTK Fixed solutions, in my experience accuracy claims (both in manufacturers’ published literature and provided by the GNSS receiver) are best taken with a grain of salt.
You could use an internet source of RTCM and feed it to your rover to get RTK Fixes with one GNSS receiver. (SparkFun has a writeup on doing this.) This is like using someone else’s receiver as a second unit instead of buying your own. This could get you 1-2 cm accuracy.
You could occupy the same point a number of times over a few days and average your results.
You could turn the UBX raw logging on, gather raw data for a few hours (SD card), and post process your results (SparkFun has a writeup on doing this.)
As a side note, did you include offset of the TOP106 in the HI (Height of Instrument, or Instrument Height) you entered into SW Maps? I don’t have that antenna, but I believe you need to enter the height of the Antenna Phase Center (APC) above the survey mark you are occupying. The APC is somewhere inside the antenna that is probably hard to measure. You might need to measure from the survey mark to the bottom of the antenna screw base (the antenna reference point, or ARP), add the ARP-APC offset, and enter that sum into SW Maps as the HI. The docs for the TOP106 antenna give the L1 ARP-APC offset to be 43.60 mm. This probably won’t make a huge difference, but best to eliminate every error source possible.
I don’t believe SW Maps includes the ARP-APC ffset when you select “SparkFun RTK” as “SparkFun RTK” applies to multiple SparkFun products with different antenna mounting. To test this, using my Facet, I change the instrument model in SW Maps from Generic NMEA to SparkFun RTK (and back and forth). The elevation didn’t change. One of my other field data collection software applications does know the antenna ARP-APC offset and add it automatically.
Good writeup - thanks! Looks like you followed most of the steps from the verification page: https://docs.sparkfun.com/SparkFun_RTK_ … rification
I can confirm most of your numbers, however I get accuracy of ~823mm. I don’t know or trust the google earth tools and at this level of accuracy, rounding makes a difference. In the future, consider converting your lat/long/height from SW Maps to ECEF then computing the sqrt(a^2+b^2+c^2) result (all this is in the example spreadsheet).
The ZED-F9P receiver is reporting HPA of 120mm (a, b )and a VPA of 260mm (c) which leads to a 310mm variance sqrt(a^2+b^2+c^2) from an ECEF coordinate. I cannot speak to how u-blox is calculating the reported HPA/VPA but that (plus the datasheet) are what we use for our accuracy claims. As Tony says, if you want better performance, RTK is really the way to go.
Follow-up: I went back and re-did everything, and realized that I had made a stupid typo when copying the coordinates into Google Earth. Fixing that, the average of the last 7 positions is 29 cm away from the actual point. Yeeha!
Tessar.pdf (359 KB)