I recently installed a Pixracer in a F450 frame, installed it in a P360 (Painless 360 3D printed Enclosure), I have installed a foam on the barometer, etc. I even used a lot of foam around the enclosure to avoid pressure variations, etc.
So far it got better with PX4 and Pixracer but unfortunately, the baro variations are still there and I can not obtain a perfect altitude control when switching from MANUAL to POSITION, etc.
This afternoon I have tried Arducopter 3.3.4-rc1 and when switching from Stabilized mode to LOITER it is perfect. Absolutely no altitude variation, etc.
The question is the following: what am I doing wrong? is there a parameter I can tune to make it better? I have probably tried every possible gain settings in QGroundControl and I can’t get it perfect.
I think this problem comes from some relatively recent changes to the position controller.
I will open an issue on the firmware side in order to discuss this issue with the corresponding
developers.
Just to let you know that today I have installed the actual BETA version in QGC and by using a different Airframe (Generic 450mm QUAD I think) I got much much better results. It is not perfect, but usable. I still got altitude variation over time meaning that after take off the same 0m became ex: -10m. I think I got this after a Geofence altitude violation. So far, I am very pleased with its behaviour in flight.
Which position estimator are you using (INAV, LPE, or EKF)?
What kind of GPS is installed?
Are you operating in an open area or near trees and/or buildings?
I find that PX4 ALTCTL mode is generally unacceptable when GPS altitude quality is poor, and that is usually the case when operating near trees and buildings. If I unplug the GPS, ALTCTL performs quite well.
I am operating in an OPEN area with trees on my right. What I can tell you is that yesterday, weather was very hot, humid, good wind and air pressure changing rapidly (maybe this is how I could explain the pressure variation).
There’s a parameter in the “system” group which selects the position estimator. I believe INAV is the default for multirotors, but that may have changed to LPE recently, depending on which version of the firmware you’re flying.
I may have to revise my opinion of the potential quality of GPS in trees; I flew a Blade Chroma yesterday in what I thought was a bad signal area and both altitude and position control were rock solid.
Did some testing yesterday, was very windy and I tried both inav and lpe. LPE seems more “regulated” in terms of altitude, only thing is that it is not constant, altitude varies when switching from stabilized to position.
So far I love PX4, QGC and Pixracer but I would only need this problem solved to use this at large with my students.