Howard Herbert Nick

Diagnostic effectiveness in computer systems using deterministic random ESD

A new criterion for electrostatic discharge (ESD) testing was recently developed where qualification effectiveness for a given apparatus is shown to be a function of the coverage obtained through ESD event coincidence with the diagnostic software/hardware in use. The authors show the effectiveness of ESD testing on a large-scale computer system undergoing testing in a typical manufacturing environment using both the present, industry accepted, deterministic ESD unit and an ESD unit based upon the recently established criterion. In addition, they provide guidelines for maximizing error detection and fault isolation as a function of ESD events and diagnostic software coverage. They further provide results showing the fallacy of assuming that a deterministic ESD unit will eventually cover every segment of a similarly deterministic diagnostic routine given unconstrained runtime. They also show the attainment of 100% diagnostic coverage in all cases based upon a 0.9 confidence using a randomly exercised ESD unit.<<ETX>>