• ATE
• calibration
• low-cost architectures
• MEMS
• sensor fusion
• testing
• trimming

During the last years smart sensors based on Micro-Electro-Mechanical systems (MEMS) are widely spreading over various fields as automotive, biomedical, optical and consumer, and nowadays they represent the outstanding state of the art.
The reasons of their diffusion is related to the capability to measure physical and chemical information using miniaturized components.
The developing of this kind of architectures, due to the heterogeneities of their components, requires a very complex design flow, due to the utilization of both mechanical parts typical of the MEMS sensor and electronic components for the interfacing and the conditioning.
In these kind of systems testing activities gain a considerable importance, and they concern various phases of the life-cycle of a MEMS based system. Indeed, since the design phase of the sensor, the validation of the design by the extraction of characteristic parameters is important, because they are necessary to design the sensor interface circuit. Moreover, this kind of architecture requires techniques for the calibration and the evaluation of the whole system in addition to the traditional methods for the testing of the control circuitry.
The first part of this research work addresses the testing optimization by the developing of different hardware/software architecture for the different testing stages of the developing flow of a MEMS based system. A flexible and low-cost platform for the characterization and the prototyping of MEMS sensors has been developed in order to provide an environment that allows also to support the design of the sensor interface. To reduce the reengineering time requested during the verification testing a universal client-server architecture has been designed to provide a unique framework to test different kind of devices, using different development environment and programming languages. Because the use of ATE during the engineering phase of the calibration algorithm is expensive in terms of ATE’s occupation time, since it requires the interruption of the production process, a flexible and easily adaptable low-cost hardware/software architecture for the calibration and the evaluation of the performance has been developed in order to allow the developing of the calibration algorithm in a user-friendly environment that permits also to realize a small and medium volume production.
The second part of the research work deals with a topic that is becoming ever more important in the field of applications for MEMS sensors, and concerns the capability to combine information extracted from different typologies of sensors (typically accelerometers, gyroscopes and magnetometers) to obtain more complex information. In this context two different algorithm for the sensor fusion has been analyzed and developed: the first one is a fully software algorithm that has been used as a means to estimate how much the errors in MEMS sensor data affect the estimation of the parameter computed using a sensor fusion algorithm; the second one, instead, is a sensor fusion algorithm based on a simplified Kalman filter. Starting from this algorithm, a bit-true model in Mathworks Simulink(TM) has been created as a system study for the implementation of the algorithm on chip.