IEEE.org | IEEE Xplore Digital Library | IEEE Standards | IEEE Spectrum | More Sites

The Switching Bus Converter: Toward 48-V-to-1-V Single-Stage Vertical Power Delivery for Data Center Applications

Date: 07/05/2024
Time: 11:00 am
Presenter: Yicheng Zhu
Abstract: High-performance processors (e.g., GPUs, CPUs, ASICs, etc.) serve as the engine of data center computing platforms and the foundation for technical progress in areas such as artificial intelligence and autonomous vehicles. Due to the fast-growing demand for greater computational power, the electric power consumption of processors has increased dramatically in recent years and is approaching 1000 W, with core logic voltages below 1 V and peak current demand beyond 1000 A. At such high current levels, the large power distribution network (PDN) of the existing two-stage lateral power delivery (LPD) architecture can lead to a dramatic voltage drop and unacceptable conduction losses, which significantly limits processor performance, reduces system energy efficiency, and hinders data center decarbonization and densification. In pursuit of a more efficient and compact alternative to the existing solution, this talk will introduce a family of high-performance 48-V-to-1-V hybrid switched-capacitor (SC) voltage regulators, named the switching bus converter (SBC). Based on a single-stage vertical power delivery (VPD) architecture, the SBC significantly reduces the PDN size and power conversion losses. Moreover, benefiting from the hybrid SC approach, the SBC effectively leverages the greatly superior energy density of capacitors compared to magnetic components, as well as the better figure-of-merit of low-voltage switching devices over high-voltage devices. Various hardware prototypes will be presented, including a 1500-A high-efficiency converter and a high-density converter with a practical form factor for VPD.
Yicheng Zhu
Yicheng Zhu is a Ph.D. candidate in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley, advised by Professor Robert Pilawa-Podgurski. He received his B.Eng. and M.S. degrees in electrical engineering from Tsinghua University in 2017 and 2020, respectively. His research interests include circuit topologies, control techniques, and analytical models of high-performance hybrid switched-capacitor converters with applications in data center power delivery. Yicheng was a recipient of the IEEE Power and Energy Society (PES) Outstanding Student Scholarship, the Outstanding Master’s Thesis Award of Tsinghua University, the Berkeley Fellowship, in 2020, and the 2023 NVIDIA Graduate Fellowship, given annually to five Ph.D. students worldwide involved in research that spans all areas of computing innovation. In 2023, he received the Best Paper Award at the IEEE 24th Workshop on Control and Modeling for Power Electronics (COMPEL) and the Best Paper Award at the 2023 Open Compute Project (OCP) Future Technologies Symposium.