Effects of response quantisation on the accuracy of transient response test results

Transient response testing has been shown to be an economical technique for testing linear macros in mixed-signal systems. Previously reported work on the analysis of transient responses has shown how the response from a linear network can be processed to generate a single figure, termed the index of functionality, which is an absolute measurement of device functionality and on which pass/fail decisions are subsequently made. However, if the circuit response is deeply buried in the mixed-signal system, it must first be extracted before it can be processed. The paper details the results of an investigation into the effects of sampling, quantising and extracting transient responses serially through a digital scan path before processing. By understanding the mechanisms by which errors introduced by the sampling and quantisation process manifest themselves in the index of functionality, it is possible to optimise the quantisation parameters for a particular test situation and test to any required standard.