Nonlinearity effects on MEMS on-chip pseudorandum testing
Résumé
Microsystems are usually affected by multiple failure sources. A faulty behavior caused by different types of defects and failure sources can exhibit small functional errors that are difficult to detect using structural testing. From here stems the necessity to apply specification-based functional testing on the basis of a method that carries enough information about the physical behavior of the Device Under Test (DUT). Such a method can be attained by the Impulse Response (IR) measurement of the DUT. In this paper we explain three existing techniques to measure the IR of Linear Time-Invariant (LTI) devices. Only simple techniques that do not require the presence of a Digital Signal Processor (DSP) on-chip are considered. A detailed comparison according to noise and distortion immunity is carried out to demonstrate our choice for a BIST (Built-In Self-Test) approach. Then we discuss the validity of the pseudorandom technique if used to test purely nonlinear and other nonlinear systems. Results are compared on the basis of the Volterra modeling method used to identify nonlinear systems.