Variation-Aware Design Verification For Automotive Semiconductor IP
TimeTuesday, July 12th1:45pm - 2:00pm PDT
Location2012, Level 2
DescriptionThe role of semiconductor content in automotive parts has increased significantly in recent years, demanding higher functional safety verification and increased simulation automation. Most automotive semiconductor designs today are mission-critical and require variation-aware simulation methods with a high level of coverage and accuracy to meet stringent regulations and ensure operational safety. This often renders brute-force Monte Carlo approaches impractical due to compute and runtime limitations.
This paper describes the design and verification of a voltage and temperature insensitive free-running RC oscillator used as a mission-critical boot clock and system monitor in a high-performance automotive application. A machine learning-enabled method for diagnosing a potential yield problem using variation-aware design verification methods enabled fundamental improvements in the design for high-sigma reliability and full PVT coverage, while reducing simulation runtime significantly.
This approach also provides design sensitivity analysis and re-verification with optimized parameters, to improve overall yield and design robustness.