Embedded test and control of analogue/RF circuits using intelligent resources

In this study, we propose a new approach of alternate test and adaptive control of power supply of AMS blocks based on the identification of some parameters of a behavioural model. A complete behavioural model of AMS block is built, including a model structure and a set of parameters. Using a population of circuits obtained by Monte Carlo simulation, a regression relationship is then built to link the system performance to the parameters of its behavioural model. A virtual performance measurement can thus be obtained from the identified parameters of the behavioural model of the current system. The estimated performances are then used either to make the pass / fail decision for test purpose or as virtual measurements for performance control of in order to optimize energy consumption. Figure 1 gives an overview of our approach.