PowerSynth 2: Physical Design Automation for High-Density 3-D Multichip Power Modules

From Graph Theory to Graph Neural Networks (GNNs): The Opportunities of GNNs in Power Electronics

“…(i) Component-level: In [20], a graph model is constructed to represent the complex, heterogeneous layouts of multichip silicon carbide power modules, capturing interconnectivity and design constraints, and using this model in conjunction with integer programming and genetic algorithms for systematic and efficient optimization of module layouts. In the latest framework of design automation technique PowerSynth 2 [59], constraint graphs are constructed to enable bottom-up constraint propagation for synthesizing layouts that respond to various design constraints, such as minimum width, enclosure, and spacing between components, leading to optimized and constraint-aware solutions. In addition, various graphtheory-induced algorithms are also implemented, such as longest path algorithm, path-finding algorithm, etc.…”

Factoring Interacting Stress Mechanisms in Design for Reliability of Extreme Environment Power Modules

“…Coupled EM and thermomechanical stressing experiments on aluminum wire bonds have been leveraged to integrate a multistress model into the PowerSynth 2 EDA workflow to emphasize the extended lifetimes that can be achieved through optimization for thermomechanical and electrical stress reduction. PowerSynth 2 is a 2D/2.5D/3-D MCPM layout optimization tool that can perform electro-thermal optimization to get a Paretooptimal solution set using fast, accurate, hardware-validated models [10,12]. The traditional 2D power module generally only uses a single substrate, such as direct bond copper (DBC) to support multiple devices connected horizontally using traces.…”

“…The traditional 2D power module generally only uses a single substrate, such as direct bond copper (DBC) to support multiple devices connected horizontally using traces. The latest breakthroughs in 3D modules are power modules that use multiple device or substrate layers that are stacked and connected vertically [12]. For these high-power density MCPM designs, EM-aware reliability optimization is essential for critical missions.…”

Influence of the rds,on temperature dependency of SiC MOSFETs on the optimal switching cell mechanical layout

“…Concerning multi-objective optimisation procedures for the mechanical design, the focus has mainly been on optimising the geometry of single components as magnetics [10], or capacitors [8], or of the cooling system for dissipating losses [12]. More recent studies propose also mechanical design optimisation of the switching module [13,14] and its control scheme [15], or the unified switching cell (SC) block (i.e. power semiconductor devices, gate drivers, decoupling capacitors, and cooling system) as an individual building block [16].…”

“…In [13,14] a multi-objective optimisation tool for the rapid design of power semiconductor modules is presented for optimising the module's layout in terms of performance and power density. This tool includes models of the electrical parasitics and the thermal coupling between the devices.…”