I believe it is less noisy in the UM982. All the charts are for 18,000 readings. The X-axis values have a lower total span, 1.15 degrees for heading and 1.62 degrees for roll, than the F9P. The number of bars are less indicating fewer changes in heading and roll values for the UM982.
I noticed that as well. Should the AOG config files for the F9P dual setup set the F9P to 8Hz (125ms) instead of 10Hz (100ms)?
The UM982 firmware only allows the discreet values 1 Hz, 2 Hz, 5 Hz, 10 Hz, 20 Hz and 50 Hz for output commands that take a time interval as an argument
I am no expert here as I do not have Dual, just found the correct PDF files. But I think you can easily test the real output Hz by doing the test recommended in the PDF: chapter 2.2.3 Testing and debugging RF link application
The NMEA GGA sentences don’t have GPS TOW in them. They have the UTC of the position fix with 2 decimal places of precision. I don’t see any skipped timestamps in the data.
Then we can believe 100 ms apart, that 10 Hz does work with dual.
That is what the F9P config files in the AOG repo set the F9P to. I didn’t develop those configs so I do not know what testing the developer of the configs did when creating them.
I am assuming (bad idea) that most people using AOG are using those configs and we don’t see reports of tractors going crazy and driving all over fields. One would think if those configs did not work there would be lots of reports of issues on this Discord. There are a number of videos where AOG autosteer is working perfectly.
Excellent test, is your base 3 band? What is baseline between antennas as roll is dependent on base line distance.
Using same antennas for testing would eliminate antenna quality questions, you can use same 3band antennas for both tests.
Interesting roll spread for um982, assuming nominal baseline distance of 1.5m. If you only got 1.82deg spread that is very close to theoretical maximum and surpass F9P max of 1.89 deg. (Receivers see master antenna at 0 so inaccuracy in data is only from slave antenna, so its ±0.94deg. But in reality master antenna vertical position can be wrong also by 25 or 15mm respectively. So we and up with ±1.89deg and ±1.51deg accuracy.) This seams high as 1.5deg would be around 8 cm away from line at 3m height. Is filter used with dual roll?
Thanks radmuffins. The base is a dual band F9P. I did update the setup description above to note that fact. I think the antenna spread is 1.0m but it is too dam cold out to go measure it now. 
If the Ublox ANN-MB is of lesser quality than the Aliexpress special I have, then would not be a good thing.
I did not test a Budget board. The Pro version of that form factor board shows the -002B as does the Micro. I tested he simpleRTK2B Lite with the F9P-002B-00 modules. FWIW, the website shows the Lite boards with a -02B module on them.
I don’t see difference between Ublox ANN-MB and bt560 that is half the price. But its not about which antenna is better, its about eliminating antenna quality influence on data. By using same antenna for both test you eliminated that variable.
But test comparing aliexpress antenna to ANN-MB would be interesting on its own.
How much filtering is used by AOG for dual? As even with ideal accuracy (±1.52deg) its possible to be ±8cm from line on 3m height.
Using the same physical antennas removes one variable and introduces another. I took the position readings close as possible to the same time. I wanted to eliminate different satellites being in view and different angles to the satellites. Not sure which, if either, would make a difference for AOG use.
I do plan to do more tests one variable at a time. First up would be using the v1.13 F9P firmware and configs. There is some evidence the older firmware works better at 10Hz. Don’t want to change antennas before testing the firmware with the current setup.
Other tangentially related thought that comes to mind is how good is good enough for AOG use. Given the output from AOG drives other electronic, hydraulic and mechanical systems. Those systems have their own response times and a degree mechanical slop. A what point is additional GPS precision even actionable by these systems.
1 Like
Good point, ideal would be to do test on 4 antennas (2 rovers simultaneously). Then switch rovers antennas for next test. You get both antennas performance and can compare roll.
There is certainly a point where more accuracy is not beneficial. This is heavily dependent on exact machine, quality of parts of aog and steering system. Now we maybe reached point of satisfactory accuracy with <cm horizontal, heading is also good with 0.1deg. This leaves just roll as main problem for accuracy.
I don’t plan on buying another set of antennas. I’ll just switch as fast as I can. Readings should have about a 35 minute difference.
Regarding roll, I am not sure how much roll influences the pure pursuit algorithm. @BrianTee_Admin Any input on how much GPS heading and / or roll influences the AOG algorithm?
On the safety side, I would not rely on the AOG roll gauge and the manufacturer specification for maximum roll angle. The manufacturer spec is usually uses best case conditions. Real world conditions such as a dip or bump can cause the tractor to roll. Also, implements attached to the tractor and even the load / position of the bucket affect roll tendency.
1 Like
What command can you disable unnecessary messages in UM982. So that they are not broadcast to the server.
See the “unlog” command on page 279 of the Command Manual linked in the first post of this thread. It will guide you on the appropriate syntax for what you wish to accomplish.
Thank you
is there some kind of “init” command? I’ve got one of those small ones UM982, and if i hook it up and connect with putty via com port (with correct baudrate) there is no nnmea output, and I cannot seem to be able to send any commands (like MODE).
But if I launch Uprecise, connect, close it, and then use putty, I see the NMEA output.
I need some kind of init command? But how do I send it?
never mind doing SAVECONFIG persisted anything Uprecise was doing, all good now, can see everything on the serial
Some interesting things happened on the way to v1.13 firmware for the F9P. The firmware and configs appear to have helped the F9P. It also put out 2 additional decimal points in the GGA messages.
Here are the updated setups.
SimpleRTK2B Lite boards with the F9P-02B-00 module - Setup in dual configuration as used by the AOG firmware. F9P firmware version v1.13. Configs; 1.13 DualAntennaHeading_RelPos.txt and 1.13 DualAntennaPosition_GGA%20VTG%20RTCM.txt from the AOG boards repo. Antennas: ANN-MB-00 and the Aliexpress 3-Band antenna linked earlier in this thread.
UM982 - Setup dual configuration mode. Antenna: Aliexpress 3-band antenna.
The platform was the metal roof of one of my outbuildings. The indoor area is climate controlled. The RTK base station is an F9P based (dual band) on my property about 40 meters from the GNSS test platform. Both systems were run continuously for at least 24 hours before taking the readings. These charts are from the PAOGI readings directly from the Teensy USB output. The UM982 LAT / LON readings have been truncated to 7 decimal places to make them the same level of precision as the F9P readings. All readings are at 10Hz for 30 minutes for a total of 18,000 samples. The vertical bars are the count of the number of readings with the value on the X-axis.
IMHO, both GNSS receivers are suitable for AOG.
The F9P Aliexpress 3-Band heading readings are 180 degrees off and the roll is negative instead of positive due to me reversing the antenna connections when moving them from the UM982 to the F9P.
Zip file of the raw data.
GNSS_Test_Data.zip (848.3 KB)
4 Likes