A reliable sampling method to reduce large sets of measurements: a case study on the calibration of digital 3-axis MEMS accelerometers

The large-scale production of digital sensors, based on MEMS/NEMS technology, and the increasingly widespread use in many surveys and innovative applications, such as smart manufacturing and Industry 4.0, poses new metrological challenges for the characterization of their technological performance. In fact, in specific applications, where measurement reliability is an essential requirement, it is necessary to have accurate and traceable sensors, therefore actually calibrated. In this work, a recently developed calibration system for digital tri-axial MEMS accelerometers is applied. The calibration procedure allows to simultaneously evaluate the main sensitivities and the transverse sensitivities, based on the single-axis excitation of an inclined plane, on which the sensor under calibration is fixed. In order to verify the effectiveness of the calibration system in dynamic conditions (from 5 Hz to 3000 Hz), 48 different configurations are explored, in order to exclude correlations between the independent variables; therefore, on the basis of a reliable sampling method, it is possible to reduce the calibration procedure to only 4 configurations, providing the values of main sensitivity and transverse sensitivity of the sensor, in compliance with the values obtained from the 48 configurations.