Dear Team,
Some of our Agri Drones sent to customer from few days are doing Toilet Bowling after takeoff. They have done multiple times Compass Calibration, but the problem is still present. I couldn’t find out any problem from logs i checked compass offsets,Magnetic interference and mag X,Y,Z all are within the range. I am sharing the Calibration and Flight Logs, Please check if am missing something and any solution for this.
Thanks.
That is very old firmware.
Do you have a log of reasonable size where it’s flying?
Do you use Magfit for Compass Calibration?
For the long flight, there are two takeoffs, and from what I can tell they look fine. Can you point to the moment in the big log that is of concern?
Also, it doesn’t appear that you actually flew the aircraft for the mag cal and might be able to improve your cal with some power data. Still not clear what behavior you’re trying to correct however
Thanks Dave,
In the same folder Shared link I will upload more logs. As for the firmware we shipped the Drone with this firmware version and it is locked.
Thanks @Joshua_Kordani, after takeoff the drone moves in circle once the pilot gives pitch or roll,the drone is moving according to the stick inputs.
I see now. I ran your longest flight through magfit. Compass 3 seems to be the most affected during parts of flight, you might consider disabling it. After pushing this to your craft, also consider doing a more comprehensive magfit flight.
When you land, are you near some big piece of metal? I assume you’re refilling a tank with some ground equipment?
MAGFit(1).param (892 Bytes)
Hello. Retired electronic engineer here. I have no experience with Ardupilot but I do have some years of working with multi-axis line-of-sight stabilization servo systems.
Referring to your graph of flight path, I see the circling motion. I also see that the “straight” path before the circling motion is not straight but wanders left-right slightly.
If this were one of my old LOS projects, I would suspect some small cross-coupling of axes.
The circling motion is actually an oscillation occurring in two axes.
Some small part of X axis if feeding into Y axis and vice versa.
Your apparent wandering left-right during “straight” flight just continues when the command for straight flight becomes zero. With no forward flight, the wander becomes a somewhat circular path.
The cross coupling between axes could be mechanical, due to rate sensors, due to accelerometers, or even due to some variable intended for X-axis being wrongly used in Y-axis calculation. Or Y-axis variable being used in X-axis.
It might also be due to some mechanical mass moving loosely on the airframe thus causing a shift in center of mass.
So many possible causes…
Compass 3 is near motor power system, Will disable it. Right Now the Drone is at customer location, might take time to reach to the office. Customer uses the drone for spraying. Can find out tomorrow whether they are landing near any metal equipment? Any particular reason.
Thanks @tsmith2e7 for the explanation.
Will cross verify for any loss parts mechanically, Is there any way i can check/verify accelerometers and x/y axis variable being used in y/x axis.
While the system was landed both times, there seems to be a large mag field that goes away after takeoff, unless I’m misreading the graphs. I guess it’s neither here nor there, just trying to figure out what I’m looking at
This might be all you need to do and run Magfit. Post a link to a log after doing so.
You should really consider updating your firmware. It’s ~2yrs old.
I do not know how to check your specific machine, but I do have some additional comments based on my past experience.
I would like to see a picture of your drone. This would be in order to get some idea of the mechanical structure.
I may assume that your drone has a tank for agricultural chemicals.
Did you test your drone before you sold it/them ? Maybe with an empty tank ?
Did the drone operate OK with an empty tank ?
What happens if your customer flies the drone in the exact configuration in which you sold the drone ?
In years past, my procedure would be to use a drone that has the spiral motion feature then test it in various conditions. Test once to see the spiral motion then change something about the configuration and test again. Empty tank versus full tank, for instance. Half-full tank ?
When the spiral motion has the larger amplitude, does the amplitude die out with time, such as down to the amplitude of the cross-axis wander seen the commanded “straight” motion.
Try introducing an attitude disturbance during hover. Hover overhead and reach up and lightly hold the bottom of the structure. What happens to the spiral motion ? In the same hover, instead of gently holding the bottom of the drone, try hitting the bottom of the drone to introduce a step disturbance to the autopilot. What happens ? If you do not like to stand under the drone then tie a string to the bottom of the drone and jerk the drone bottom from a safe distance.
Does the drone have significant mass located below the autopilot ? Such as the tank for chemicals ? When the drone starts or stops straight-line motion does the drone have a pendulum motion ? Where is the center of rotation of the pendulum ? If the center of rotation is below the autopilot, the the autopilot will experience the opposite pendulum motion of the mass below the pendulum’s center of rotation.
If the autopilot is moving in space in a pendulum motion, then the autopilot will control the motors to stop the a/p’s motion. Stopping the pendulum motion will result in the airframe moving laterally, probably in a spiral pattern.
All for now. My poor brain may generate more ideas as hours go by.
The root cause appears to have been identified. You are introducing massive assumptions and variables. Suggest we allow the current train of thought play out before further naive explorations.
Thanks Everyone, one of our engineer is on the way to the location, I have informed him to do the magfit flight and disable the Compass 3. Do the test. He is taking extra modules to change if any issue persists. Will share more details.
Sorry, we cant change the Firmware version.
It’s pretty wild to send drones out without doing these basic things.
