A Multi-Phase Diagnostic Flow Combining Scan-Based Refinement and Targeted Functional Tests for Automotive SoCs

Modern automotive systems increasingly rely on complex digital System-on-Chip (SoC) devices, where logic faults can severely compromise both performance and safety. Accurate post-silicon diagnosis is challenging because scan-based evidence may lead to large and ambiguous candidate sets, while purely functional validation often lacks the structural observability required to pinpoint the root cause. This study explores a multi-strategy diagnostic framework that leverages a hybrid structural–functional diagnostic flow based on cross-domain correlation. Scan-based diagnosis and iterative diagnostic patterns are used to identify and refine structural candidates; in turn, the resulting ranked candidates guide the development of targeted functional scenarios, whose outcomes are exploited to filter the initial candidate set. The framework is validated on a batch of real-world failed Automotive SoCs produced by STMicroelectronics, consistently reducing the candidate set across all testable devices in the batch and achieving accurate fault localization, as it is physically confirmed for an illustrative case study, by Physical Failure Analysis.