An approach to linear model-based testing for nonlinear cascaded mixed-signal systems

Linear model-based test and diagnosis (MbT&D) has been successfully applied to single-block modules like digital-to-analog converters (DACs) with a static non-linear transfer characteristic. For multi-block modules, a diagnosis methodology is needed that can deal with cascades of several linear and nonlinear blocks. In contrast to non-linear methods, linear MbT&D methods only require matrix operations associated with relatively low computational effort. A modification of the linear MbT&D in combination with Volterra series is presented that can be applied to cascaded non-linear systems, for example, a DAC followed by a low-pass filter. A simultaneous identification of numerous frequency domain Volterra kernels is enabled, and thus, to test the compliance to data sheet specifications.