An Automated Layout Generation Framework for Power Electronics PCBs

Abstract

The layout design for power electronics printed circuit boards (PCB) remains a critical step in the design automation toolchain for modern power electronics. At present, almost all power electronics PCBs are still manually laid. It is time and material-consuming that experienced engineers take iterations of development and testing stages before the PCB designs go to mass production. This thesis develops an automated layout generation framework named "PEPCB" (the abbreviation for Power Electronics PCB) that allows power PCB design methods using sophisticated component footprints and copper shapes, and has been validated that can generate usable PCB designs within minutes using limited computing resources (e.g. laptops). In contrast, manual design can take tens of minutes to hours. Together with other design automation tools, they will enable the industry to efficiently tailor-make power stages for every individual application. Meanwhile, even though the overall goal of PEPCB is to achieve a fully automated layout design, the framework still allows user interventions to better accommodate valuable engineer experience. The thesis first presents a comprehensive and transparent iterative automated layout design workflow, considering both component real footprint shapes and copper envelopes. On this basis, a hierarchical component grouping framework, with elaborated mathematical derivations for more stable and faster automated layout design is constructed. A SPICE-output-based grouping method is tailored for power electronics PCBs, emphasizing circuit parasitic parameters. With a detailed discussion of user interventions, the grouping framework ensures users to define hard rules by modifying specific design parameters. For the integrity and usability of PEPCB, the thesis proposes at least two design algorithms for each layout design procedure. The aim is to efficiently generate usable PCBs with limited computing resources. A thorough electromagnetic interference (EMI) analysis on grid-based copper representations is finally conducted in the thesis, covering signal integrity (SI), radiation, and crosstalk, while optimizing the copper design algorithms. Additionally, PCB thermal optimization methods and layout structural adjustment strategies for complex circuits are provided. With all these achievements, the thesis guarantees high-quality power PCB designs for complicated power circuits using PEPCB, meeting high frequency, high efficiency, low EMI, and high power density requirements.

Date of Award	29 Aug 2024
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Gus Cheng Zhang (Main Supervisor) & Andrew Forsyth (Co Supervisor)