Delay-Controlled Frequency Tuning of a MEMS Sensor for Neuromorphic Acoustic Sensing

Frequency tunability marks the ability to change the characteristic frequency of an oscillator. For MEMS sensors, this is usually achieved by exploiting a sophisticated geometry with a nonlinear stress-strain relationship. This is known as active frequency tuning. However, MEMS sensors with such geometries are difficult to manufacture and the characteristic frequency can have a lower limit, which is constrained by the material constants and geometry of the MEMS sensor. To address this issue, we propose a different approach to enable active frequency tunability, which is based on designing a feedback loop with a controllable time delay. We show by analyzing the necessary condition of the Hopf theorem that the characteristic frequency of this system can be increased indefinitely or decreased to 90% of its original value by appropriately adjusting the delay and the feedback strengths. These observations can be explained by combining with the undelayed and delayed feedback loop, which implies that the phase and the amplitude of the feedback signal can be controlled. In addition, the gain of the sensor becomes tunable, since the Andronov-Hopf bifurcation can be controlled with this feedback loop. These results are particularly interesting for mimicking the cochlea functionality, as the cochlear is assumed to exhibit an Andronov-Hopf bifurcation. Hence, this approach can be, e.g., used for neuromorphic acoustic sensing, while keeping the geometry of this MEMS sensor as simple as possible.