Front-end for capacitive sensors

This research focuses on a new powerful front-end able to reject environmental noise for capacitive sensors. Capacitive sensors can be used for indoor localization and so they can be adopted in many location-based services that improve life quality and safety. Room occupancy information can help reducing energy consumption by controlling the ambient temperature, lighting and water. Also, monitoring human activity for extended periods can detect behavioral changes, which can help recognizing the early onset of diseases like Parkinson disease. Moreover, presence monitoring systems can also be used to detect unauthorized intrusions. Capacitive sensors, operating in load mode, use just one-plate transducer and the human body as a constant-potential plate. However, load mode capacitive sensors are generally used to distances comparable to their plate dimensions, which are too short for indoor localization. Long-range capacitive sensors need to sense very small variations of the capacitance of the sensor plate. Their sensitivity and accuracy are often limited by the level of noise induced by environmental sources. To mitigate these effects, the sensor plate can be guarded by auxiliary fields to reduce the unwanted couplings of the sensor plate with the surrounding objects, and by post-processing sensor data to improve the reliability of long-range measurements. Here, I study a technique based on differential single plate measurements. This technique can reject low frequency noise which is a common mode signal for the differential acquisition and processing of sensor data. It is based on having a current, that charges discharges the plate, driven by a square wave. Thus the plate voltage is a triangular wave with slopes proportional to the capacitance of sensor. An Analog to Digital Converter, perfectly synchronized with the driving square wave, is used to sample plate voltage. To improve the accuracy of measurements and the rejection of drift noise I use oversampling and decimation technique that allows to increase the resolution (from 12-bit to 16-bit) of the analog to digital conversion and reduce the quantization error. Once the system for differential measurements was ready, I proceed in implementing an analog low-pass filter, with cut-off frequency between 15 kHz and 20 kHz, by using the operational amplifier available in the microcontroller and external capacitors and resistors. The purpose is reducing high-frequency noise. Due to the non-linearities of the active components, results after the filtering show some distortions, thus I evaluate the errors to minimize their influence on the measurements. Another element in the front end is a digital filter to reduce the noise outside the frequency bands of interest. A digital filter is a system that performs mathematical operations on a sampled, discrete-time signal to reduce or enhance certain aspects of that signal. They may be more expensive than an equivalent analog filter but they can reach very high orders. Many digital filters are based on the fast Fourier transform. I experiment different kind of digital filters : Finite Impulse Response, Chebyshev, and Elliptic. The mask used is the same for all of them and it consists in cut off frequency of 1 Hz or 3 Hz, -80 dB in stopband. They are designed with minumum order to match stopband characteristic.