Interfaces directas entre sensores capacitivos y microcontrolador

Resumo

The direct connection of sensors to a microcontroller (MCU) without using any intervening circuit is a simple, reliable and efficient solution to reduce footprint and cost. In this thesis, several direct interface circuits based on charge transfer have been analyzed and validated that accept simple, differential and, unlike most commercial interfaces, multi-terminal sensors, with one or none grounded electrode, and for any value range from 1 pF up. To this end, various circuit configurations and measurement techniques have been proposed, developed and evaluated. The theoretical and experimental results obtained have allowed us, on the one hand, to understand and improve the operation of these circuits and, on the other hand, to apply these interfaces also to other analog sensors (resistive, inductive, or self-generating), to which the charge-transfer method had never been applied before. The theoretical analysis has led to the identification of the parameters of the MCU and the circuit components that determine the transfer characteristic of the measurand-to-digital conversion, which for capacitive sensors is linear when the capacitance of the sensor is inversely proportional to the measurand. Further, some limitations and uncertainty sources have also been identified and design rules have been formulated to improve measurement accuracy. The uncertainty sources identified and analyzed include: 1) the parasitic capacitances associated with the MCU pins and the connection layout of the sensor and reference capacitor to the MCU, which have proven to be the main source of systematic zero deviations (offset), especially when measuring small capacitances; 2) the dielectric absorption in the reference capacitor, which introduces gain deviations and whose effect can be reduced, according to the experimental analysis, not only by using capacitors with low dielectric absorption and implementing long discharge times, but also by discarding the first readings of the measurement; 3) temperature changes and drifts in the supply voltage, which introduce gain deviations and do not depend on the measurement method; and 4) capacitive coupled interference, which introduces mainly non-linearity, and whose effect depends not only on the amplitude and frequency of the interfering voltage and the coupling capacitance between the interfering source and the measurement node but also on the measurement method and the topology of the interface circuit. With regard to accuracy, the main experimental results are as follows. 1) For simple capacitive sensors, a maximum deviation referred to the Full Scale Range (FSR) of 8 % for sensors between 2 pF and 10 pF and less than 0.8 % for sensors between 10 pF and 1 nF, which are common values in moisture, rain and liquid level sensors, among others. 2) For capacitive differential sensors with nominal capacitance of 100 pF and 1 nF, a maximum non-linearity deviation relative to the Full Scale Span (FSS, Full Scale Span) of less than 0.2 % and 1 %, respectively, when subjected to relative capacitive changes of up to 90 %. 3) For simple resistive sensors between 100 M¿ and 1 k¿, a relative deviation of less than 8 % for resistors close to 100 k¿ and less than 2 % for resistors between 220 k¿ and 1 M¿. Theoretical and experimental analysis further shows that these relative deviations are qualitatively the same for other decades of the measurement interval, e.g. the range of 100 k¿ and 10 M¿. These features match or overcome those of more complex interface circuits, and may be sufficient in environments where low cost, low power consumption, and compactness are essential, particularly in the Internet of Things (IoT) environment.