Hello guys,
I am trying to calibrate the accelerometer in the FlyMaple using an ArduPilot v3.3 modified, unfortunately, it seems that I cannot do it. I tried to get the samples from the board and this is the results:
Level 0.11 -0.19 -8.88
Left 0.12 9.84 1.03
Right 0.1 -10.46 1.02
Down -10 -0.33 0.88
Up 10.26 -0.17 0.39
Back 0.33 -0.35 10.68
With these values, the iteration cannot be converging. I checked it with a small Matlab program using the algorithm of ArduPilot. Can it be there are something wrong with the accel, have you guys ever met this situation ?
In the worst case, can I just ignore the accel calibration and test the auto mode in the real model ? How bad can it be if I ignore this offset ?
Replies
I am now having the same issue as you where the accelerometer calibration fails. HUD looks correct. Did you ever solve this problem?
When I attempt to calibrate I see:
offsets[0] = -1.08 0.34 -.03
Scaling[0] = 1.35 1.21 1.08
I see in the code that the offsets cannot be greater than 3.5 (ok) and the scaling cannot be outside the range 0.9 to 1.1 (which obviously mine are).
Do you see the same issue with the scaling?
Somehow my values are not even converging, as I mentioned before. However I decided to carry on, as long as the values when the system in level are good enough (nearly 0 for both pitch and roll)
So did you ever get the accel calibration routine to say "successful" or does it always say "failed"? I think I might have other issues too. Just sitting on my desk, the yaw reading spins around and around. They pitch and roll now appear wrong in the HUD (about 10 degrees both).
It might be the AHRS_TRIM_X and Y are large. Set them to zero and see if that fixes the lean.
These trim values are captured at the first stage of the accel calibration. I wonder if perhaps the AHRS_ORIENTATION is not correct, the IMU on the flymaple is not oriented correctly or if the code has the default orientation for the flymaple incorrect.
Interestingly, when my AHRS_ORIENTATION is 0, the horizon is perfectly level, but if I changeAHRS_ORIENTATION to 1 (yaw 45), my horizon goes quite tilted. If only yaw were changed I would expect the horizon to stay level. Do you know what could cause this?
I reset the trim values and now my HUD appears correct, but the calibration still fails. I doubt that the orientation is wrong. If I tilt the board downward/upward in pitch, or rightward/leftward in roll, I see the HUD respond correctly.
Similar to Quang, my gravity measurements are all over the place. The magnitude of his acceleration vectors in the 6 positions varied from 8.88m/s^2 in the level position to 10.69 in the back position. Mine varied from 9.0 in the level position to 10.8 in the left position. It seems like the accelerometer is highly direction dependent, which may be the reason the calibration fails the 0.9 < scale < 1.1 requirement.
'falied' every time. In your case, I have never met that situation. I think you should change the offset and the scale values back to their default to see if it can fix that problem.
Do you mind if I ask how you connected the board to Mission Planner?
I just purchased a flymaple board and followed the instructions here to upload the firmware v3.3 to the board. I have that completed, yet Mission Planner always times out on the connect.
Is there an additional step I need to take? Thanks.
What kind of connection are you using, TCP or USB port ? In fact, FlyMaple cannot connect with Mission Planner via SerialUSB, or at least I cannot find out a way to do that... The only way I know is via Wi-Fi (TCP). If you want to communicate with FlyMaple by USB port, you should use MapleIDE....
-
1
-
2
of 2 Next