With dual antenna if you mounted one antenna at the back of the tractor and one at the front, you would get quite a lot of distance for heading. Yes, roll would not work, but we do have the MMA. Could you get the position of the 2 antenna’s? You could use the front antenna for steer distance off guidance line, and back antenna for pivot position, with an incredibly stable heading. Thoughts?
Would you feed rtcm corrections from a base into both or does the one act as a base for the other
Is it possible to place the antennas diagonally, e.g. one in the front left and the other one in the right back of roof in order to get highest distance and roll? Has anyone figuered out yet what the min. distance between antennas should be?
1 Like
One is the primary, the other slave and yes corrections to both
Hi, I’m writing the code for a dual Antenna system. It is working quite well. When putting the antennas left and right you can additionally calculate roll. I’m using it with a wineyard tractor, so I can get only 70cm antenna distance, but the failiure in heading is around 0.5° accordung to UBlox docs.
Yesterday I started with the webinterface to be able to config it later on. Maybe next weekend on GitHub. The actual one has no Webinterface and not all the math…
See:
Software for ESP32:
Commercial systems do left and right. Why choosing to front and back? Or are commercial systems doing it wrong? 
I like that idea too. I have an antenna already installed on the nose at a height of about 150 cm. It would be so easy to just add a second antenna on the back of the roof. All I’m thinking is that, on a sidehill, a difference in height of the antenna placement would result in slightly false heading values. Unless that’s accounted for via the roll values of e.g. the MMA.
Behind the roof and tractor, there is no real option for an antenna at 150 cm in my case.
Ships have 4 antennas, font back, left right. The advantage of spreading them out farther is more accurate heading - but the question is about placement and how guidance works, it takes the pivot position, and using sin/cos figures out where the front of the tractor is based on the heading. Then the distance from the guidance line and the delta of that part of the line and the heading of the tractor is used to know which way it needs to steer. Now you can see that if the heading is a bit off, it puts the steer position in the wrong place, and won’t work as well.
Having those two antennas in the 2 most critical places of knowledge may have tremendous advantage for guidance. But none for roll. OR like a ship, we use 4 antennas. I think arduSimple would like that plan!!
1 Like
— 
—
1 Like
3 antennas would be enough, 2 on the cabin for the Roll (largest place to slip side by side) and one on the front center then you get a triangle.
1 Like
Front of tractor mounting isn’t possible for me as the tractor spend 70% of its time with a loader fitted…the arms get in the way!
So probably 1.5m spread is the best either direction.
This would be the ideal situation I think… Do all the processing on the computer / AOG, save having another external device with potential updates / changes etc, would also be easier to diagnose having each UBLOX connected directly to the PC…
Would also be nice to be able to pass through the correction information etc so that AgOpenGPS could display its fix info / RTK status etc…
Possibility to build these calcs into software with config page to setup how many antennas and location of said antennas / connection to PC Brian?
1 Like
It would be an option to connect both to AOG. The thing is, that the system with moving baseline and the UBX RelPosNED sentence is Ublox specific.
It works like this: #1 sends RTCM to #2, so connect #1 TX2 to #2 RX2.
#2 calculates the relative position to #1: heading, 3 vectors of direction, and length of vector to #1 this contains the RelPosNED sentence,that is Ublox specific.
#1 delivers position as every other GPS chip. When feeded with NRTIP it gives an absolute position, if not the normal GPS dissolution, whereas the heading and roll is allways correct, as it is relative from the 2 receivers.
So what the ESP32 does:
Feed one unit with NTRIP
Get position
Get relative vectors
Calculate roll (trigonometry)
Correct the antenna position by using roll and virtual Antenna movement (trigonometry)
Checks the sentences, the flags (GPS valid…), and the accuracy (length of vector to other antenna (suggested in Ublox manual))
Generate NMEA sentence (PAOGI (one for all) or GGA (position) + VTG (speed) + HDT (heading))
Transmit via UDP/ Wi-Fi or USB
It is doable in AOG, but you go back to USB. I like UDP since you can connect/ disconnected and Windows won’t care.
With the ESP32 solution it’s similar to a commercial solution, you get corrected data. Building is easy: the Ardusimple boards fit to Arduino uno layout, the esp too. Build a stack, connect to Arduino IDE, flash, configure in Webinterface- done. No IMU needed, maybe Wi-Fi router (with GSM for RTK) or connect ESP32 via USB
2 Likes
a distance of 75cm will do, more is better see page 5 in UBlox reference:
2 Likes
2 Antennas are enough, as pitch does not affect the direction.
I can certainly see reliability benefits of reducing number of USB / com connections to the PC…always potential for issues this way I guess… Does current version of AOG allow for GPS data incoming via UDP? Have got my autosteer PCB connected via UDP / makes sence to use this for as much as possible to simplify things…
I’ve actually never seen a dual antenna system in a tractor despite there being loads of autosteer tractors in my area.
Me neither come to think of it… But it does make sense to me / has got to be the best and fastest way to determine heading one would think…