Determining Accelerometer Zero Offset Errors during Bench Calibration
Static multi-position gravity inversion separates zero-g offset from scale factor while Allan variance bias instability defines maximum valid integration time.
Six Position Calibration
Spatial Orientation
Geometric registration defines the sensor frame by rotating an instrument through ninety degree intervals along orthogonal axes. This six position calibration accounts for mounting offsets and gravitational bias within an inertial measurement unit. Gravity acts as the absolute reference for the accelerometer set during the procedure.
Fixed mounting surfaces inside the test rig hold the device steady during data acquisition for each orientation. Residual misalignment causes errors in output vectors that remain static regardless of device motion.
Systematic Error
Sensor bias manifests as a constant shift from the expected gravitational constant during the stationary segments of the protocol. A six position calibration isolates these specific biases from the scale factor errors of each individual axis. Each measurement captures the response of the triad under local acceleration.
Subtracting the earth gravity value from the observed reading calculates the actual offset present in the hardware. Repeatability of these values defines the confidence in the sensor output for long term navigation tasks.
Mechanical Constraint
Thermal expansion of the mounting jig introduces variations that limit the precision of the derived correction coefficients. Every degree of temperature change alters the physical alignment of the six position calibration setup. Laboratories maintain temperature stability within one degree to prevent these fluctuations.
Rigid structural components minimize the vibration that might skew the readings during the sampling phase. Verification against a laser tracker ensures the truth of the orientation angles during the entire duration of the test.
Vector Correction
Implementation of the resulting matrix transforms raw input data into a corrected coordinate space. Algorithms apply the calculated biases to the live stream to ensure the sensor reports a value of zero acceleration at rest. Accurate nulling of these offsets allows the hardware to track motion without drift accumulating in the integration chain.
Internal processing speeds determine the latency of this correction logic. Stability of the bias over time represents the ultimate limit of the measurement quality.