Test methodology for prediction of analog performance parameters

Abstract

Analog testing, the name itself signifies the detection of faults in analog circuits. The aim of this thesis is to increase the test effectiveness and work in the performance parameter space. There are many test methodologies which can detect the faults in the circuit under test (CUT), out of which the test methodologies which can determine CUT performance parameters resulting in enhanced test effectiveness are, predictive oscillation based test methodologies. To detect the catastrophic and parametric faults these methodologies are used. These test methodologies are preferred over other methodologies because the input test stimulus generation is not needed, which reduces the complexity if multiple inputs are applied to the circuit. These test techniques are implemented with prediction process using neural networks which will in turn increases the performance of the circuit under test. The thesis follows with the implementation of the techniques and understanding the methods to increase the test effectiveness. The design process is performed in CADENCE simulation tool with 180nm technology.