Hyper-Integrated CAD Database for One-Shot EOTPR Fault Localization in Heterogeneous SiP-on-Board Assemblies

Thursday, November 20, 2025: 8:20 AM
2 (Pasadena Convention Center)
Arpan Bhattacherjee , NVIDIA Corporation, Santa Clara, CA
Dr. Joy Liao , NVIDIA Corporation, Santa Clara, CA

Summary:

The increasing complexity of datacenter accelerators and artificial intelligence (AI) inference systems has driven rapid adoption of sophisticated heterogeneous system-in-package (SiP) architectures. These advanced packages integrate multilayer printed circuit boards (PCBs), organic substrates, silicon interposers, and high-density integrated circuits, delivering significant performance improvements. However, these densely integrated architectures introduce substantial challenges for accurate and timely failure analysis (FA). Electro-Optical Terahertz Pulse Reflectometry (EOTPR), a non-destructive diagnostic method, offers superior fault localization compared to traditional techniques. Nonetheless, its practical implementation in complex SiP systems has been limited by fragmented CAD data and iterative manual analyses, significantly extending diagnostic timelines. This paper presents a novel “hyper-integrated” CAD database methodology, where disparate design datasets are automatically consolidated into a unified navigation database. This approach creates a static digital twin of the failing board-level device, enabling precise, “one-shot” fault localization using EOTPR. Experimental validation on advanced datacenter accelerator hardware demonstrates substantial improvements in diagnostic speed, accuracy, and efficiency, emphasizing the practical benefits of the hyper-integrated approach for state-of-the-art semiconductor failure analysis.
See more of: Boards and System Level FA
See more of: Technical Program