Litcius/Paper detail

A 9x Matrix Autotransformer Switched-Capacitor DC-DC Converter for Datacenter Application

Haoran Meng, Maohang Qiu, Zhongshu Sun, Xiaoyan Liu, Dong Cao

202310 citationsDOI

Abstract

Datacenter applications have grown rapidly in recent years. Power converters with a high-power density and high efficiency for datacenter applications are in high demand. This paper presents a nine times conversion ratio, i.e., 9x matrix autotransformer switched-capacitor DC-DC converter (MASC) for data center applications. MASC has both advantages of switched tank converter and LLC converter. The high voltage side of MASC takes advantage of switched tank converter. The low voltage rectifier side of the MASC leveraged a new magnetic integrated autotransformer design that is like a current doubler rectifier like an LLC converter. Compared with the switched tank converter, MASC needs less switches on the low voltage rectifier side to achieve the same conversion ratio, thus leading to higher efficiency. Through appropriate design of parasitic inductance, all switches can achieve zero current switching or zero voltage switching. Therefore, total power loss of MASC can be reduced for high current and low voltage data center applications compared with switch tank converter. Compared with traditional LLC converter, the use of matrix autotransformer instead of isolation transformer further reduces the power loss of MASC by removing primary side windings. A 48V-5.33V MASC hardware prototype with a peak efficiency of 98.54%, a full load efficiency of 95.6%, and 654W/in <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> power density is designed, built, and tested.

Topics & Concepts

AutotransformerElectrical engineeringConvertersRectifier (neural networks)Switched capacitorCapacitorComputer scienceBoost converterTransformerVoltageForward converterElectromagnetic coilEngineeringDistribution transformerRecurrent neural networkArtificial neural networkStochastic neural networkMachine learningAdvanced DC-DC ConvertersMultilevel Inverters and ConvertersMicrogrid Control and Optimization