Improving Stress Quality For Soc Using Faster-Than-At-Speed Execution Of Functional Programs

At the end of the manufacturing cycle of digital circuits, a stress phase is mandatory in order to remove from the final device population the weak devices that may result in early life failures. Devices used in safety critical environments must undergo this phase that is usually accomplished by exploiting the Burn-In (BI) process. Unfortunately, BI has elevated costs for companies and current state of the art techniques are trying to reduce its cost.In recent days, Faster-than-at-Speed-Test (FAST) has become a useful technique to discover small delay defects. At the same time, overclocking methods to enhance system performances have been studied, which focus on temperature management to preserve system functionalities. In this contribution, a FAST technique is proposed with the aim of intentionally provoking a thermal over-heating in the microprocessor by mean of the execution of FAST functional test programs; in other words, functional procedures are executed at higher than nominal frequencies. The goal is to introduce an internal stress stronger than current procedures used during BI in order to speed up early detection of latent faults.Being the functional stress procedures executed at faster than nominal speed, the original behaviour may not be preserved and therefore non-functional states may be reached. In this contribution, it is illustrated how to avoid blocking configurations due to timing constraints violation and how to obtain a significant increase of the switching activity by carefully increasing the clock frequency. Furthermore, a novel strategy is proposed to generate a suitable set of Faster-than-At-Speed stress programs capable to thoroughly stress processor cores.Experimental results carried out on a MIPS-like architecture show major achievements of the methodology: the processor may work at frequencies up to about 20 times higher than the nominal one without falling into an unpredictable state and the switching activity is increasing up to 300% per ns.