Efficient cross-sensitivity compensation in multisensor systems by half-blind calibration

Abstract

Sensor systems designed to determine external measurands often show parasitic sensitivities to other influences, so-called disturbances. Calibration of such a system aims for the extraction of disturbance-compensated measurand values from the system output signals. Often this goal is achieved by exposing the system to a sufficient number of load conditions involving well-controlled values not only of the measurands but also of the disturbances. In this work we show that the calibration effort focused exclusively on the extraction of disturbance-compensated measurand values can be considerably reduced in the case of systems with linear response. The conclusions therefore apply to Hall, piezoresistive stress, and temperature sensors, among others. During the calibration procedure, well-controlled measurand values need indeed to be applied to the system; however, while all disturbance parameters need to be varied during calibration, accurate knowledge of their values is not needed. By making use exclusively of the calibration measurand values and the concurrently extracted sensor signals, it is possible to determine a reduced calibration matrix which ensure the successful extraction of disturbance-compensated measurand values from sensor signals. The effectiveness of the method is demonstrated using a six-degree-of freedom linear force-moment microtransducer with redundant sensors. The method may save cost by simpler calibration setups and the time-saving procedure it proposes.