Date of Graduation
Doctor of Philosophy in Engineering (PhD)
Computer Science & Computer Engineering
Second Committee Member
Third Committee Member
EDA Tool for Power Electronics, Power Module Layout Optimization, Power-CAD, PowerSynth
Moving towards an electrified world requires ultra high-density power converters. Electric vehicles, electrified aerospace, data centers, etc. are just a few fields among wide application areas of power electronic systems, where high-density power converters are essential. As a critical part of these power converters, power semiconductor modules and their layout optimization has been identified as a crucial step in achieving the maximum performance and density for wide bandgap technologies (i.e., GaN and SiC). New packaging technologies are also introduced to produce reliable and efficient multichip power module (MCPM) designs to push the current limits. The complexity of the emerging MCPM layouts is surpassing the capability of a manual, iterative design process to produce an optimum design with agile development requirements. An electronic design automation tool called PowerSynth has been introduced with ongoing research toward enhanced capabilities to speed up the optimized MCPM layout design process. This dissertation presents the PowerSynth progression timeline with the methodology updates and corresponding critical results compared to v1.1. The first released version (v1.1) of PowerSynth demonstrated the benefits of layout abstraction, and reduced-order modeling techniques to perform rapid optimization of the MCPM module compared to the traditional, manual, and iterative design approach. However, that version is limited by several key factors: layout representation technique, layout generation algorithms, iterative design-rule-checking (DRC), optimization algorithm candidates, etc. To address these limitations, and enhance PowerSynth’s capabilities, constraint-aware, scalable, and efficient algorithms have been developed and implemented. PowerSynth layout engine has evolved from v1.3 to v2.0 throughout the last five years to incorporate the algorithm updates and generate all 2D/2.5D/3D Manhattan layout solutions. These fundamental changes in the layout generation methodology have also called for updates in the performance modeling techniques and enabled exploring different optimization algorithms. The latest PowerSynth 2 architecture has been implemented to enable electro-thermo-mechanical and reliability optimization on 2D/2.5D/3D MCPM layouts, and set up a path toward cabinet-level optimization. PowerSynth v2.0 computer-aided design (CAD) flow has been hardware-validated through manufacturing and testing of an optimized novel 3D MCPM layout. The flow has shown significant speedup compared to the manual design flow with a comparable optimization result.
Al Razi, I. (2022). Constraint-Aware, Scalable, and Efficient Algorithms for Multi-Chip Power Module Layout Optimization. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/4602