Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Zhongshu Sun_Thesis final__final format approved LW 7-30-2024.pdf (2.93 MB)

Digital Accessibility Report

File List


Acrobat Accessibility Report.pdf (70.7 KB)

ETD Abstract Container

Abstract Header

Matrix Auto-Transformer Switched Capacitor Dc-Dc Converter For Ai Training Computing Application

Sun, Zhongshu
http://orcid.org/0009-0007-9772-4652
http://rave.ohiolink.edu/etdc/view?acc_num=dayton172250979914251

Abstract Details

2024, Master of Science in Electrical Engineering, University of Dayton, Electrical Engineering.
This paper presents two DC-DC converters optimized for data center applications, focusing on the 48V DC bus input to low voltage outputs. The first converter discussed is a 48V to 3.2V (15x) MASC DC-DC Converter, which incorporates a high voltage side structure similar to that of a Switched Tank Converter (STC) and a low voltage side utilizing a current doubler circuit comparable to the secondary side of an LLC converter. This innovative design is aimed at minimizing both transformer winding losses and conduction losses on the low voltage side. The reconfigured structure allows for a theoretical reduction of the transformer's winding loss by 44.9% compared to conventional LLC converters, assuming the same core size. Additionally, the conduction loss on the low voltage side devices is reduced by half when compared to traditional STC designs. The experimental results of prototype achieved maximum efficiency of 98.75%. The second converter introduced a 48V to 3.3V Matrix Autotransformer Switched Capacitor DC-DC Converter with Partial Power Processing Regulator. It combines the circuit called matrix autotransformer switched capacitor converter (MASC) with buck converter input in series and output in parallel to achieve regulation purpose. The MASC can work at DCX mode to realize higher efficiency, and only 17.9% of the total power are processed by buck converter for regulation purpose. In this way, the power loss is reduced and the overall performance of the prototype is improved. A hardware prototype is designed and built. Simulation and experimental results are provided to verify and demonstrate the performance of the MASC with buck regulator converter. The prototype maximum measured efficiency can reach 96.2%.
Dong Cao (Committee Chair)
Bradley Ratliff (Committee Member)
Jitendra Kumar (Committee Member)
45 p.
Electrical Engineering; Engineering
switched-capacitor; Matrix Autotransformer; DC-DC; datacenter; LLC; Switch tank converter;

Recommended Citations

Citations

  • Sun, Z. (2024). Matrix Auto-Transformer Switched Capacitor Dc-Dc Converter For Ai Training Computing Application [Master's thesis, University of Dayton]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=dayton172250979914251

    APA Style (7th edition)

  • Sun, Zhongshu. Matrix Auto-Transformer Switched Capacitor Dc-Dc Converter For Ai Training Computing Application. 2024. University of Dayton, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=dayton172250979914251.

    MLA Style (8th edition)

  • Sun, Zhongshu. "Matrix Auto-Transformer Switched Capacitor Dc-Dc Converter For Ai Training Computing Application." Master's thesis, University of Dayton, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=dayton172250979914251

    Chicago Manual of Style (17th edition)

dayton172250979914251
61
© 2024, all rights reserved.
This open access ETD is published by University of Dayton and OhioLINK.