RESUMEN
We show that the calibration of tri-axis accelerometers based on the device's intrinsic properties alleviates the uncertainty due to mounting misalignment in comparison to the use of the sensitivity matrix. The intrinsic properties of a tri-axis accelerometer are based on a (u, v, w) coordinate system that represent the direction of maximum sensitivities of each of the three accelerometers (U, V, W) and are assumed not to be perfectly orthogonal to each other. The calibration procedure requires rotation of the device in the gravitational field around each of the Cartesian coordinate (x, y, z) axes. One component in driving down the uncertainty of laboratory comparisons and calibration repeats relates to misalignment in mounting the device onto the calibration instrument. We show that the uncertainty of the cross-axis terms of the sensitivity matrix is a dominating factor affecting uncertainty down to a 0.01° misalignment at a 100 µV noise level. The misalignment component can be exacerbated when calibrating modern microelectromechanical systems (MEMS)-based accelerometers, which are typically a few millimeters in dimension.