I have a BMI 270 chip that appears to have a different sensitivity scalar in one direction compared to the other. I tested for this with the following procedure: Compute bias manually when chip is in no-motion Apply bias manually in software Rotate chip 10 rotations clockwise, integrating gyro during motion Rotate chip 10 rotations counterclockwise, integrating gyro during motion Report end integrated gyro. Doing this test results in a consistent error of about 2 degrees, across the 5 tests I performed. If instead I were to vibrate the BMI on the same fixture for the same duration as the test, the error results in about 0.2 degrees, which appears to indicate that this error is a result of the turning in particular. The other BMI's I have tested with this exact procedure usually end up with 0.2 to 0.4 degrees error. So this leads to my questions: Is this expected behavior? Does the BMI 270's gyroscope sensitivity scalar change with respect to the direction? Is this accounted for with respect to the nonlinearity specification? Could this be due to mishandling of the chip, such as a drop or fall?
... View more
