Modeling of Molded Discrete Power Packages Geometric Detail Based on 3D-Mechanical Sectioning Image Reconstruction Method

E-mobility and automotive applications drive electronic circuits towards higher switching frequencies, thus higher power densities, nowadays. Whether traction inverters, onboard chargers, battery management systems, or even power electronics in cars' entertainment systems show increasing demands for high power densities shrinking and miniaturizing electronic systems. When it comes to high power densities in electronic systems, thermal cross-talk between single circuit devices (molded discrete power packages) should be considered. Passive devices, such as a capacitor placed in close proximity next to a molded discrete power package (active device), can get hot even by parasitic heating only. It is well known that electrical circuits, passive and active semiconductor components, can drastically change their electrical parameters due to increased temperatures. When designing such power dense systems, the most crucial question is how hot do components get? Indeed, simulations should accurately capture temperature distribution and thermal cross-talk in electronic assemblies to prevent system failures and detect possible design issues. For that reason, the internal structures of electronic devices can critically influence the circuit behavior. Typically, in power electronic simulations, the used mechanical 3D (CAD) models rely on pure datasheet information. Especially molded discrete power packages need more accurate modeling of internal structures. There are indeed hidden details in an epoxy mold compound, such as bond wires, die, heat slug, which is not described within the datasheet. This thesis proposes a straightforward and economically feasible technique to obtain high-resolution geometric information of the details hidden in the enclosed discrete power packages. High fidelity detailed internal structures are retrieved to create a 3D volume from the stack of 2D sequential images obtained using the 3D-Mechanical Sectioning Image Reconstruction method, which is a combination of two well-known conventional methods; parallel grinding and image reconstruction.