Background: My first ArduCopter (APM1/oilpan, mag, no sonar, ublox GPS, xbee 900, hoverthings X450 quad frame, 1000 Kv motors, 30A escs, 10*4.7 props, 4Ah 35C 1P4S LiPo, X cross configuration, code from trunk from about two weeks ago). Had a few flying sessions with some satisfactory episodes, but mostly dissatisfied with stability and control, even after carefully ballancing props, making sure everything is properly aligned, no slipping motor shafts etc.
The problem seems to be getting worse. After the last repair, I struggled to make simple hovers in the back yard, it would just flip when I throttle up, unless I gave it strong correction. I even resorted to making a "string cage", and it wouldn't pull up level without "aggressive climb" type throttle (it's amply overpowered). At hover/gentle climb throttle it would flop about like a rag doll.
Eventually, I sat it level for a few hours and logged the output with mavproxy. It might not have been perfectly level, but it was reasonably close and perfectly still. Plotting ATTITUDE.roll and ATTITUDE.pitch revealed something is definitely broken:
I'm new at ArduPilot (previously only used paparazzi), and don't really know where to go from here. I'd attach the full log but can't because it's >7MB zipped. I can start swapping out some components but don't have another IMU/Ardupilot handy. How to diagnose and fix this?
Replies
Chris and I looked at his logs in more detail at a MakeHackVoid meeting this evening, and found the problem was that he had not run the 'level' command to calibrate his accelerometers since changing oilpan board. The accelerometer calibration was quite a way off, which caused gyro drift due to accelerometer drift correction.
- bad imu
- bad mag
- heat problem
- miscondiguration
I have a new mag coming, and can experiment with better cooling, but need some suggestions wrt imu testing and configuration (or better theories).