Stochastic counting MEMS sensor using white noise oscillation for high temperature environment

This paper describes a stochastic counting MEMS sensor, which will be used in low S/N environments like in high temperature plants. A mass which vibrates between two counter electrodes by white voltage noise is "pulled-in" to either of the electrodes by the application of pulse voltage to the mass. The direction of the pull-in is determined stochastically, and the probability that the mass is pulled-in to a particular side depends on mechanical strain applied to the sensor structure. This sensing principle was confirmed by simulation, and then the sensor was prototyped using an SOI wafer. The probability of pull-in to a particular side was tuned by bias voltage applied to the counter electrodes, as predicted by the simulation. When the frequency of the pulse voltage for pull-in increases, the behavior of the sensor becomes unintended, because the mass is pulled-in during dumping vibration after releasing from the previous pull-in. This limits the sensing speed, but strain sensing is possible just counting the number of pull-in to a particular side, which is easy even in low S/N environments.