Managing the sampling rate variability of digital MEMS accelerometers in dynamic calibration

The use of sensors with digital interface, within large or dense sensor networks, is nowadays widespread in many scientific and technological applications: from more traditional applications, such as infrastructure monitoring and predictive maintenance, up to the advanced ones, such as smart manufacturing, IoT, Machine Learning, and other emerging fields of fast-real-time interconnections, within the framework of digitalization. The technical and functional performance of these sensing infrastructures, if accurately identified, allows to enhance the trustworthiness, safety, and accuracy of the managed processes; based on metrological characterizations and calibration, it is possible to provide the actual sensitivity of digital sensors with respect to reference physical stimuli, within the proper uncertainty budgets and suitable covering factors. At present days, metrological characterization and proper calibration of digital sensors is still a technical and methodological challenge and several studies are oriented along with this perspective. In this paper, the sampling rate variability - depending on MEMS analog-to-digital converter, external microcontroller internal clock and their interaction - of digital MEMS accelerometers in dynamic calibration is investigated. The sampling rate variability is evaluated among 25 sensors of the same batch, and within every single sensor in time, and methods to manage the associated uncertainty and to avoid mismatches in calibration, are proposed and discussed.