Abstract :
Test cost means different things to different teams at different times, in terms of investment, spending and affordability. Long design and production life-time cycles continue to give rise to scenarios where, while the test cost estimated at design time was considered expendable, the actual test cost incurred during production was not, resulting in debates, analysis and optimisations, (often implemented grossly and sub-optimally, due to lack of planning and lack of support in design as well as test infrastructure). Besides, as the life-cycle expands, i.e. devices are manufactured, sold and supported over a longer time period, the impact of test cost (as a percentage of manufacturing cost) on profitability can be even more profound. The adoption of the right techniques and infrastructure to guarantee functionality upon manufacturing, and during the operation life-cycle (as applicable) is crucial for profitability. Design support is often the most important enabler. Correspondingly, this drives the extent and scope of the functionality in the manufactured chip to be verified, the selection of tests for this verification, and the total number of chips being subjected to such tests. Test cost is perceptibly incurred “after manufacturing” and, in different contexts, considered “dispensable”. As a result, it is a big variable and often the target of ongoing optimisations. The lesser the affordability, (dependent on what is considered essential versus what is considered as an overhead), the higher is the focus on such optimisations. Good test cost models and estimates can, therefore, help to set the right bounds on affordability and optimisations.
