Optimization Tool for Layout Stack to ensure Even Current Sharing among Paralleled Switching Devices: Layout impedance balance for paralleled power devices
Translated title
Optimization Tool for Layout Stack to ensure Even Current Sharing among Paralleled Switching Devices
Author
Gupta, Shubham
Term
4. term
Education
Publication year
2025
Submitted on
2025-05-28
Pages
60
Abstract
Når flere effektsmoduler arbejder parallelt, kan selv små forskelle i impedans i det omkringliggende layout give ujævn fordeling af strømmen. Dette projekt præsenterer en praktisk metode til at opnå mere balanceret strømdeling ved at optimere det eksterne layout med en genetisk algoritme (en søgemetode inspireret af naturlig selektion). Målet er at opdage og visualisere impedansskævheder i komplekse systemlayouts tidligt i udviklingsforløbet, så designere lettere kan afhjælpe dem. En nøglekomponent er FastHenry, et open source-værktøj, hvis åbenhed giver brugerne kontrol over design og drift. I samspil med MATLAB bliver arbejdsgangen mere effektiv med automatiserede simuleringer, ligningsløsning og interaktive visualiseringer. Som fuldt open source-alternativ kan Python også bruges til en tilsvarende integration. Til validering genererede algoritmen flere layoutvarianter med forskellig kompleksitet, som blev afprøvet i et Double Pulse Test (DPT)-setup hos KK Wind Solutions. Fremtidigt arbejde omfatter at forbedre FastHenry-modellens nøjagtighed, øge optimeringsværktøjets konvergenshastighed og gøre det mere brugerinteraktivt.
When several power modules operate in parallel, even small differences in impedance in the surrounding layout can make currents split unevenly. This project proposes a pragmatic way to improve current sharing by optimizing the external layout with a Genetic Algorithm (a search method inspired by natural selection). The aim is to detect and visualize impedance imbalances in complex system layouts early in development, so designers can address them sooner. A key enabler is FastHenry, an open-source software tool whose openness gives users control over design and operation. Combined with MATLAB, it streamlines the workflow through automated simulations, equation solving, and interactive visualizations. For a fully open-source alternative, similar integration can be achieved with Python. To validate the results, the algorithm generated several layout variants of varying complexity, which were tested using a Double Pulse Test (DPT) setup at KK Wind Solutions’ facility. Future work includes improving the accuracy of the FastHenry model, speeding up the optimization tool’s convergence, and making it more interactive for users.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Documents