@ Cédric
About the signs:
I've checked them, but maybe I've overseen something.
I'll check it again asap.
How to change the signs only for the magnetometer?
The only way to "correct" it is with the "BODY_TO_IMU" parameter,
right?
But this also mixes up the acceleration behaviour.
Those are my current settings: (see attached photo - the ground sign
is pointing to the front)
<define name="BODY_TO_IMU_PHI" value="0."
unit="deg"/>
<define name="BODY_TO_IMU_THETA" value="0."
unit="deg"/>
<define name="BODY_TO_IMU_PSI" value="90."
unit="deg"/>
I've also tried those settings, but with no difference in the drift:
<define name="BODY_TO_IMU_PHI" value="180."
unit="deg"/>
<define name="BODY_TO_IMU_THETA" value="180."
unit="deg"/>
<define name="BODY_TO_IMU_PSI" value="270."
unit="deg"/>
@Heinrich
The propellers aren't balanced.
Because of the kind of my frame it is very hard to twist the main
wires for the motor power.
I will do the test with the influence of the current with different
throttle values.
If there is a big influence, I think I will do the calibration of
the current.
But I already have drift without some power on the motors. Also if
the autopilot is not mounted to the frame.
I've checked the rate values but I will recheck it.
The last time I've checked the accelerometer (rotation) the values
were stable.
Because of my frame, It is not possible to keep the 10cm distance
from the power wires, but I don't think I this is the problem,
because the drift is present, even the frame is not mounted on the
frame.
Thank you
Bruzzlee
On 04.04.14 08:34, Cédric Marzer (MRSA)
wrote:
Are you sure the signs are right on both the accelerometer and the magnetometer ?
On 03.04.14 13:17, Prof. Dr.-Ing.
Heinrich Warmers wrote:
Hi Bruzzle,
have you balanced all propellers?
Have you made the wiring of the DC and motor wires so, that
there is
not opening a space?
If you have a power switch you have to twist the plus and ground
wire
to get a small open space and wire both lines to to switch.
Please fix the multikopter to ground and look to the magnetometer
measurements, that there is no influence of the current with
different
throttle values.
Please look to the rate values.
Please look to the accelometer values.
Same other multikopter projects keep the magnetometer about 10
cm
away from every power wire.
good luck
Heinrich
Bruzzlee schrieb:
Hi Loic
Thx for the hint.
Today I did the calibration outside. (without mobile phone and
any
disturbing elements in the near)
It looks better now, but sometimes I still have a small drift.
Because of the drift, the Quadrotor wants to correct its
position and
spins up two rotors, even if I take the throttle back to zero
again.
Then I have to use the kill switch... (The first time it
"jumped" into
a wall become of it ;-) )
Maybe I have to tune the AHRS parameters... Any suggestions?
By the way, is there a simply way to deactivate the magnetometer
on the
Aspirin? (online)
Mag calibration should be perfomed outside, far away from
any
metallic material (cars, watch, belt, ...)
The gyro bias can be measured with gyro scaled mean value
when not
moving.
This value should be always equal to the bias estimated by
the AHRS
filter. Otherwise, it means that the gyro and mag
measurements don't
agree.
In most case this is due to a bad mag calibration.
Regards
Loic
Le 31 mars 2014 20:46, "Bruzzlee"
<address@hidden>
a
écrit :
Hi Heinrich and
Alonso
Today I did the calibration about 30 times. (on
different places)
Now I have calibration which I think I can work with.
Even the globe was not fully surrounded with measurement
dots.
(calibration graphic)
The phi and the bias is now almost stable.
Those values are all out of your range but I think it
will work anyway,
don't you?
Thank you for you hints
Bruzzlee
On 31.03.14 09:58, Prof. Dr.-Ing. Heinrich Warmers
wrote:
Hi,
can you calculate the bias in °/s?
I have found that a bias in the range of 0.5-2°/s is
normal.
If the bias brake this border change the sensor
element.
regards
Heinrich
Bruzzlee schrieb:
Hello
paparazzi community
After 8h of trying to get rid of my drift on phi you
are my last hope.
Every hint is useful!
I think it has something to do with the AHRS.
Compensation of psi drift with MAG
The plot shows a fast 90° turn clockwise and back
again.
In the lower plot you also can that the bias changes
a lot, is that
"normal"?
Lisa M 1.0 + Aspirin IMU 1.5
(I did the calibration of Accel and mag a few time
with similar
results)