Eric D. Joseph

Linearization techniques for capacitive sensors

Capacitive sensors are essential elements in many instrumentation circuits. The widespread use of these instruments and the need for integrating sensors on the same chip as the signal processing circuitry has motivated the developement of linearization techniques for the sensor output. Presented here are four methods which can be used to linearize the essentially nonlinear nature of capacitive transducers. Described first is the polynomial linearization scheme, in which the Taylor series expansion of the sensor output is linearized by Khachabs method. An improvement in accuracy is obtained by the direct linearization scheme which scales the sensor output by the nonlinear denominator. Though this scheme is useful in cases where the signal processing circuitry must be separated from the interface circuitry, the hardware requirements are pretty intense. An alternative approach is the Summer-Divide scheme, wherein the difference and the sum of the top and bottom capacitances are ratioed to yield the linearized output. The last method is the constant charge injection scheme. Here a constant charge is put on top of the top of the plates and the subsequent voltages are differenced to yield a linear output. Nonlinearity associated with parasitic capacitances can be easily removed by feedback schemes.

Design of a Σ-Δ converter-based automotive sensor

The design of a second order (Sigma) - (Delta) converter based microaccelerometer is presented in this paper. This converter has a large dynamic range, i.e., can measure acceleration from 0.5 g to 500 g, and provides directly digital output. The force-rebalance circuitry constraints the sensor mass to move within 10 percent of the distance between the polysilicon layers. The system is inherently linear and has a simple electronic interface.