I did an experiment with my quad held on the ground and then held in my hand while I boosted the throttle. The ground effect is really quite pronounced as can be seen here:
Midway you can see where the copter is in my hand, still pressure changes but not nearly so noticeable.
I’m hoping http://ardupilot.org/copter/docs/ground-effect-compensation.html will help as advised by @rmackay9, but this seems quite extreme!
That’s normal… It doesn’t take much wind to change the pressure on the barometer.
I needed a truly surprising amount of foam and protection on my little 3" copter to get it working.
A bit over the baro was not enough. I covered the entire board in foam, and then put tape on the edges to keep the wind out.
The baro is on the bottom of my board, so I basically sandwiched foam between the FC and the board under it, foam being the same 20x20ish size as the FC. Then put tape around the edges between the FC and ESC under it.
I can still make the altitude change if I hold the copter in my hand and blow air on it. But, it seems fine in flight now.
Nice test though! Txs for providing this data!
By the way, I plan to enable the GND_EFFECT_COMP by default in 3.7 (or maybe even sooner) 'cuz it only seems to help. It doesn’t remove the problem but it helps…
I’m not sure I agree that this is because I haven’t got enough foam - I think there genuinely is a change in static pressure because of the large increase in propwash, rather than just buffeting.
i think there is something in the code that still creates a consistent deviation bwetween baro alt and factual alt.
i just did this test flight - i can see clearly how it ‘fixes’ difference between baro reported alt and the lidar reported alt when it moves vertically, but then as it stabilizes in the loiter - somehow it creates a delta, and it is not a constant delta - in the loiter period on the lef tit was less then after getting up and down.
in reality it did not affect model`s behavior- it was glued solid to the position in the air. still, it looks odd to me.
It’s probably clear but just in case, the barometer altitude is an altitude above the EKF origin which may not be at exactly the same altitude as the takeoff altitude (or home).
Today I did several flight tests with my Mini Quad 210mm (it’s not a racing drone).
1st Test.
With the parameter GND_EFFECT_COMP set to value 1, and in the flight mode “PosHold” I made a takeoff, configured for a height of 2 meters.
Results:
Altitude Hold, very bad.
2nd. Test.
With the parameter GND_EFFECT_COMP set to the value 0, and in the flight mode “PosHold” I made a takeoff, configured for a height of 2 meters.
Altitude Hold, very bad.
3rd Test.
With the parameter GND_EFFECT_COMP set to value 1, and in the flight mode “PosHold” I made a takeoff, configured for a height of 5 meters.
Altitude Hold, acceptable.
4th Test.
With the parameter GND_EFFECT_COMP set to the value 0, and in the flight mode “PosHold” I made a takeoff, configured for a height of 5 meters.
Altitude Hold, acceptable.
5th. Test.
With the parameter GND_EFFECT_COMP set to value 1, and in the flight mode “PosHold” I made a takeoff, configured for a height of 15 meters.
Altitude Hold, Excellent
6th Test.
With the parameter GND_EFFECT_COMP set to the value 0, and in the flight mode “PosHold” I made a takeoff, configured for a height above 15 meters.
Altitude Hold, Excellent
Conclusion: In my case I have come to believe that the parameter GND_EFFECT_COMP is not having any effect.
