A review of recent accelerometry experiments points to the need for a careful consideration of the question of where, exactly, the accelerometer sensor itself is located within the device that hosts the services required for its operation. We propose a simple measure for characterising the regime of the experimental conditions under which these implied positional errors become significant. We then present a method for reliably determining the accelerometer position and apply it to a wide variety of host devices ranging from data loggers to smartphones and tablets. Although the obvious value of the method lies in the improved accuracy of the ensuing accelerometry, the preliminary process of discovering the accelerometer position becomes a valuable and complete pedagogical experience in itself, providing insight into both the principles of radial acceleration and the internal structure of current-generation digital devices.
References
Although some points may share the same magnitude or the same direction, no two points are identical in both characteristics.
In practise, gyroscopes are used to detect changes in direction, and accelerometers only for changes in linear acceleration, but the point still holds, given that accelerometer readings still need to be corrected for such changes in direction.
See Table III for a comparative list.
To be fair, the estimation of damping model parameters, which was the goal of the paper in question, is not affected by the value of the radial distance, as the latter remains a constant with no variation. Their uncertainty due to true sensor position would lie along the short side of the device, given the experimental conditions described. In the absence of details regarding exactly how radial position was measured, we estimate the uncertainty in position to be the half-width of the short side, which for the iPod Touch is roughly 30 mm.
To make this rule consistent with host devices running iOS requires a preprocessing step in which all acceleration values are multiplied by −1.