Hello,
I have tried without success to fly my quadrotor in RC_CLIMB
mode. Firstly, I had some problems with the altitude measured from the
baro: i found out that the BMP085 mounted on the LISA M is sensitive to
sunlight! I protected the sensor with foam and the altitude measurement
is now OK.
After this, i still didn't manage to use correctly RC_CLIMB
mode: I found that the vertical speed (ins_zd) and accelerations
(ins_zdd) are wrong.Vertical speed "zd" is varying from 3 to 5 m/s with
the quad hovering.... And the acceleration "zdd" oscillates from -10 to
30 m/s² !!
Now, i guess that these vertical speed and acceleration errors
comes from the accelerometers used in the ins kalman filter. In that
case, i have several questions:
- What level of vibration/ noise of accelerometers is acceptable
on a quad hovering? I have around +/- 7m/s²
- I have mounted the Lisa M on gel pads (
http://www.rtom.com/moongel.htm),
balanced the propellers, what else can be done to minimize vibrations
on a quadrotor? Are there other things to pay attention?
- In case the vibrations can't be reduced, is it possible to
tune the Kalman filter in the INS ? I have seen that the filtering is
done in vff_float.c. In this file, the parameter ACCEL_NOISE is equal
to 0.5: are there any chance to improve the filtering results by tuning
this parameter?
- The vertical filter VFF can be turned off by removing the flag
USE_VFF? What would be the consequences of such tuning? in other words,
how ins_zd and ins_zdd are then calculated?
- I also updated aspirin.c file, in order to use +/-16 G range
on the accelerometers.
- Any suggestions to fix this problem?
Thanks for helping. I have started to describe the block
diagrams of the rotorcraft vertical control loops, in order to add them
to the paparazzi wiki. So I hope to contribute myself to the paparazzi
project soon!
Regards
Loïc
_______________________________________________
Paparazzi-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/paparazzi-devel