Delay sensing for long-term variations and defects monitoring in safety–critical applications

Abstract

The impact of parametric variations on digital circuit performance is increasing in nanometer Integrated Circuits (IC), namely of Process, power supply Voltage and Temperature (PVT) variations. Moreover, circuit aging also impacts circuit performance, especially due to Negative Bias Temperature Instability (NBTI) effect. A growing number of physical defects manifest themselves as delay faults (at production, or during product lifetime). On-chip, on-line delay monitoring, as a circuit failure prediction technique, can be an attractive solution to guarantee correct operation in safety---critical applications. Safe operation can be monitored, by predictive delay fault detection. A delay monitoring methodology and a novel delay sensor (to be selectively inserted in key locations in the design and to be activated according to user's requirements) is proposed, and a 65 nm design is presented. The proposed sensor is programmable, allowing delay monitoring for a wide range of delay values, and has been optimized to exhibit low sensitivity to PVT and aging-induced variations. Two MOSFET models--BPTM and ST--have been used. As abnormal delays can be monitored, regardless of their origin, both parametric variations and physical defects impact on circuit performance can be identified. Simulation results show that the sensor is effective in identifying such abnormal delays, due to NBTI-induced aging and to resistive open defects.