High Step-Down Single-Stage DC-DC Converter with Improved Planar Matrix Transformer for High-Current Data Center Application

Chang Wang, Mingxiao Li, Ziwei Ouyang, Zhe Zhang, Gabriel Zsurzsan, Michael A.E. Andersen

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

The torrent of information technology assigns a significant role to the nowaday data center. Since substantial energy consumption is reported in data center operation, the point-of-load (POL) converter in data center desiderates high efficiency and high power density to output a large current for the specific server. A 4-phases LLC converter is designed by implementing an improved matrix magnetic core along with the interleaving and integration of windings. Balanced current sharing and less conduction loss are achieved by the interleaving and integration strategies. An 800 W LLC experimental prototype is built under 500 kHz switching frequency with 40V-48V/5V-6V voltage conversion. Peak efficiency achieves 98.1% along with power density of 240 W/in3 with the experimental result. Excellent thermal performance is achieved along with the potential for delivering more power.

Original languageEnglish
Title of host publicationProceedings of 2022 IEEE Applied Power Electronics Conference and Exposition
PublisherIEEE
Publication date2022
Pages709-715
ISBN (Print)978-1-6654-0689-5
DOIs
Publication statusPublished - 2022
Event2022 IEEE Applied Power Electronics Conference and Exposition - George R. Brown Convention Center, Houston, United States
Duration: 20 Mar 202224 Mar 2022

Conference

Conference2022 IEEE Applied Power Electronics Conference and Exposition
LocationGeorge R. Brown Convention Center
Country/TerritoryUnited States
CityHouston
Period20/03/202224/03/2022

Keywords

  • Data center
  • LLC
  • Magnetic integration
  • Matrix transformer
  • Point-of-load converter
  • POL converter

Fingerprint

Dive into the research topics of 'High Step-Down Single-Stage DC-DC Converter with Improved Planar Matrix Transformer for High-Current Data Center Application'. Together they form a unique fingerprint.

Cite this