Cube Blue and Black IMU pitch oscillation

The issue is the pitch of the aircraft oscillating by was much as +/-4 degrees when sitting still. This issue seems to be more of a issue on the newer Cubes leading to the idea that something has changed with the IMU sensors. Is this behavior of the IMUs expected?

This issue has been observed on one CubeBlue and 3 CubeBlack flight controllers. This issue was observed and measured after the cubes had a chance to warm up to reduce issues from the IMUs being cold. Unless there is a significant change in pitch or roll the cube was sitting perfectly still on a flat table without being disturbed. The table is made of 24 inch thick stone on a welded steel frame. It does not move. Telemetry radios were set to a lower power or unplugged during testing and showed no difference in measurements. The CubeBlue logs were taken before and after doing a re-calibration of the accelerometer. Firmware versions were 3.9.8 or 3.9.9 depending on where the aircraft was in our internal flight worthiness testing. Parameter values defined by SB_0000003 were used by all aircraft.

Any more suggestions for testing are welcome. I think I might be able to test at least one CubeBlack from each batch all the way back to 2017 if I dug around long enough.

CubeBlue bin file after initial accelerometer calibration: Here+ GPS without a lock 36 12-31-1969 6-00-00 PM.bin (3.0 MB)

CubeBlue bin file after second accelerometer calibration: Same GPS 2 12-31-1969 6-00-00 PM.bin (2.4 MB)

CubeBlack in a taildragger that is part of the SB_0000002 batch bin #1: 2019-09-06 12-01-57
CubeBlack in a taildragger that is part of the SB_0000002 batch bin #2: 2019-09-06 14-24-18.bin (1.6 MB)

CubeBlack on a desk purchased well before SB_0000002 batch bin: 2019-09-06 14-11-56.bin (2.3 MB)

CubeBlack on a desk purchased in circa October 2018 as a combo with a Here2: 2019-09-06 14-14-27.bin (2.6 MB)

Unless they are armed and powered off a battery, any test on the bench is not really valid.
Slight oscillation on the bench is a normal ardupilot thing.

There are no changes for cube Blue or black in the IMU.

New IMU is in the CubeYellow, CubeOrange, and CubeBlack+, not in the standard Black or blue

Get a gps lock, and arm the aircraft, then check.

The aircraft got a GPS lock. Kalman filter switched to using using GPS and everything snapped in to +/- 0.4 degrees. I hadn’t thought about using GPS for attitude determination. Any hints for finding the algorithm in the code. Now I’m more curious than concerned.

1 Like

It’s all in the EKF, code ain’t my area :slight_smile: I’m the hardware guy :slight_smile:

1 Like