Unstable PXFmini flight - drone doesn't hover in place

[dsposito]

I've built a DIY Pi0 drone using Erle Robotics' PXFmini, power module, GPS module and a Raspberry Pi 3.

The good news is that after a few months of learning the drone is up and flying! Thank you everyone that has helped me get this far!

The bad news is that the drone does not hover well in place - constantly drifting around and up and down.

Here is a video of my drone: https://youtu.be/pqvtXJlrwCk

Notice that the drone drifts around both forward/back/left/right and also up and down altitude wise which is particularly concerning to me. After a few seconds of drifting, the drone begins to fall slowly towards the ground - then goes back up to repeat the process. During the video, it is not windy and I'm not touching the RC controls except right before the drone lands back on the ground to avoid a hard landing.

I have already done these things:

• calibrated the ESCs
• calibrated the accelerometer
• calibrated the compass (I think)
• ensured the components on the drone are centered to distribute weight as evenly as possible on the drone frame
• replaced the propellers

I'm now in the process of trying to calibrate the power module. I thought the ER GPS module I bought would help but the flight doesn't seem any more stable.

What else should I do to stabilize the flight? The simple goal for my drone is to rise straight up ~50 feet and hover in place for a few minutes to record some aerial video.

Thank you for your feedback!

[dsposito]

Would PID tuning be something that would help stabilize the flight during hover seen in the video above?

[ibaiape]

Good day @dsposito, thanks for your patience.

Have you tried autotune? Indeed, PID tuning should help in stabilizing your copter.

So very sorry for taking this long.

[Gregory]

Hi Daniel, I had a similar problem with my DIY drone. The problem was much worse in the early afternoon. I discovered through an obscure reference in one of the ardupilot docs that barometers are sensitive to sunlight. I placed a piece of felt over the barometer so as to not block the holes in it, and covered that with black tape to keep the light off of it. The difference was night and day.

[Bernie]

Dear Daniel

I have the same Problem with my copter. DJI F450 + PXFMini + Rasperry Zero. The Copter hovers +/- 2m in Alt Hold Mode.

The Barometer is covered by a black foam (10mm). Vibrations are in an acceptable range. I notice in the Dataflash Log, that the CTUN-BALT Signal is different (smaller) to the CTUN-ALT Signal, which I think is used by the controller for the altiude controller Loop, but the CTUN-BALT Signal shows what I see during the flight.

Have you done the autotune calibration? I will do it in the next days.

Regards
Bernhard

[dsposito]

Interesting. Most of my drone brains are contained within a Raspberry Pi case though with the PXFmini it's a little taller so I just have the black blast lid taped on top. So it should be blocking out 80%+ of any light. To be sure, I'll try flying at different times of day to see if that makes any difference. Thanks for the tip!

[dsposito]

Hey Bernie,

1. Does your copter also go up/down in altitude when it "hovers" as seen in my video?
2. Are you controlling your copter via a remote control or software / ROS? When using RC, I'm actually not sure what mode my quad copter is in. I'll have to research how to configure that via RC - let me know if you have any tips.

I haven't tried the autotune calibration yet - I'm not sure I'll be able to successfully since the process takes several minutes and my copter's altitude can vary dramatically during flight (I'll probably have to try running the test a few dozen feet in the air to ensure it doesn't hit the ground during autotune). I'll give it a shot in the next week or so - let us know if you have any luck!