When flying our DEXI drone, we’ve encountered multiple situations where the DEXI begins to drift and ignore commands. Then, it flies into a wall, tilting and producing full throttle. In our most recent flight test, one or more of the cells in the battery burned up (pictured left). When trying with a different but similar battery (pictured right) we have not encountered this issue.
The picture doesn’t show it very clearly, but near the bottom left and bottom right of the battery you can see that the battery is burnt.
Can you use your cell checker and share the output of the cell voltages? You are welcome to use a lipo battery, but you won’t get as much flight time. We have specifically tuned DEXI to run for 15 minutes on the lithium ion. If DEXI is not responding to your transmitter inputs then something else is at play. I’m assuming there’s an issue with the receiver. If you could remove the SD card (the all black one) and grab the PX4 logs we could do an analysis. There is a log folder on that card you could zip up and share with us.
We ran another test flight and it looks like the controller disconnected mid-flight. The controller displayed no connection and the drone continued flying in the direction it was already flying in for approximately 8 seconds, then landed on its own.
Any chance you can get us a flight log? I believe this could help us more quickly get to the bottom of the issue. Also, what flight mode do you have engaged?
This behavior was mostly exhibited in the stabilized flight mode (white LED). I submitted the flight logs to support@droneblocks.io under DEXI Flight Logs - 1350A
Here is a video of what its doing. It seems to happen in all modes except stabalized.
Thanks for the video. Any idea if this was a more recent flight or an older one? I ask because I’m looking through about 50 different log files and trying to find the one that matches up with the video above. I’m also trying to locate the one where it goes into the wall with full throttle. Here is the successful position flight mode (a quick 7s flight):
https://review.px4.io/plot_app?log=52a9a831-f71f-445c-9186-2ee9220bbabc
and here is one I discovered where manual control was lost but it doesn’t look like PX4 ever let DEXI takeoff. It quickly switched to land mode and disarmed:
https://review.px4.io/plot_app?log=690063e6-2e3b-44bc-8123-b275e74e4a8f
Will report back when we learn more.
I forgot to mention two things:
- I see in several of the flights that there are critical battery levels. See this log:
https://review.px4.io/plot_app?log=24b77f3c-e18d-4f27-955b-8e12575b9d12
Please make sure to keep an eye on your battery levels before, during, and after flight.
- I’m not sure if your lab is where all the test flights are happening, but there are a lot of metal objects (tables, chairs, filing cabinet) that can lead to magentic interference. If you have a more open area to test that would be ideal.
That flight was last night. It wouldnt be in the log that was sent. They flew it again today in the hallway and it did a little better. It has occured in both the lab and in the hall. Ill have them keep a better eye on battery levels.
Ah got it. One thing I’m seeing in the following flight log:
https://review.px4.io/plot_app?log=c699a187-2abc-456f-9ec8-5f4352ef04b6
If you look at the norm of the magnetic field it’s pretty inconsistent:
You want something more like this with less variance:
From the PX4 docs:
If the norm is uncorrelated but not constant, most likely it is not properly calibrated. However it could also be due to external disturbances (for example when flying close to metal constructs).
I would recommend doing another compass calibration away from metal objects.
I will have them do this in the morning and see if this gekps. Thanks for taking a look at the logs.
1 Like