Structured Test, Then and Now

There are two broad approaches to device testing: Functional test – “Does the device do what it is supposed to do?” – and structural test – “Has the device been built correctly?” Historically, electronic devices were tested functionally. If a device met its specifications, then by definition it worked and test was complete. As the complexity of electronic systems increased, however, verifying that a device met its specifications became increasingly difficult. A syllogism was constructed to solve the testing problem: If the device’s design is correct, and if each part of the manufactured structure operates correctly, then the device works. With this, structured test was born. Staples of structured test have been in existence since the 1960s, including fault models (stuck-at faults date from the 1940s [1], although the first published reference is [2]), automated diagnostics [3], fault simulation [4], and automatic test pattern generation [5]. At present, some form of structured test is the norm for virtually all integrated circuits.