Uncertainty reduction techniques in microcontroller-based time measurements

This paper proposes and analyses different methods for reducing the uncertainty in microcontroller-based time measurements, which are increasingly common in interface circuits for sensors. The main uncertainty sources in these measurements are quantisation and trigger noise. The former depends on the technique used by the CPU for detecting the external events that start or stop the timing process, whereas the latter is in part attributable to the CPU activity. The sleep mode, a feature designed to reduce power consumption in microcontrollers, offers a simple and efficient solution to improve the measurement uncertainty because it reduces both the interrupt response time and the noise resulting from the CPU activity. These improvements are experimentally validated by testing an 8-bit microcontroller (AVR AT90S2313).