High Current, High Power-Density Intermediate Bus Converters for Vertical Power Delivery to Next-Generation Processors

Advancements in artificial intelligence and machine learning are driving the development of a new generation of extremely powerful graphics processing units (GPUs) with exponentially increasing transistor counts, requiring larger die sizes and more power. Today, employing the conventional two-stage lateral power delivery with a 12 V intermediate bus voltage (IBV) requires most of the real estate in the GPU cards to be occupied by power delivery circuits. However, this is impractical for the next-generation GPUs with larger sizes and higher power consumptions, resulting in the interest to transition to vertical power delivery, where a lower IBV is used to reduce the size of the voltage regulators and move them vertically underneath the GPU. However, this increases the required step-down ratio for the first-stage intermediate bus converter (IBC), making it challenging to maintain high efficiency while constrained to a limited footprint. This paper proposes a high-density transformer unit-cell for the 1st-stage LLC-based IBC, one or more of which can be connected to achieve different IBVs. To demonstrate an implementation of the transformer, an 840 W 48 V/1.8 V LLC converter is designed with an array of 14 transformers built into seven unit cells, distributed along the edge of the GPU package to minimize the power distribution network (PDN) losses. The converter module can achieve up to 95.5% efficiency while showcasing a power density of 2200 W/in<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>.

Authors

Institutions

Publication Details

Journal
IEEE Transactions on Power Electronics
Published
2025-05-23
DOI
https://doi.org/10.1109/tpel.2025.3572841
Primary Topic
Advanced DC-DC Converters
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

High Current, High Power-Density Intermediate Bus Converters for Vertical Power Delivery to Next-Generation Processors

Yan Liang, Ahmed Nabih, Pranav Raj Prakash, Mostafa Mosa et al.
IEEE Transactions on Power Electronics
Advanced DC-DC Converters
article

High Current, High Power-Density Intermediate Bus Converters for Vertical Power Delivery to Next-Generation Processors

Yan Liang, Ahmed Nabih, Pranav Raj Prakash, Mostafa Mosa, C. Thomas Gray, Sudhir S. Kudva, Qiang Li
article en

Abstract

Advancements in artificial intelligence and machine learning are driving the development of a new generation of extremely powerful graphics processing units (GPUs) with exponentially increasing transistor counts, requiring larger die sizes and more power. Today, employing the conventional two-stage lateral power delivery with a 12 V intermediate bus voltage (IBV) requires most of the real estate in the GPU cards to be occupied by power delivery circuits. However, this is impractical for the next-generation GPUs with larger sizes and higher power consumptions, resulting in the interest to transition to vertical power delivery, where a lower IBV is used to reduce the size of the voltage regulators and move them vertically underneath the GPU. However, this increases the required step-down ratio for the first-stage intermediate bus converter (IBC), making it challenging to maintain high efficiency while constrained to a limited footprint. This paper proposes a high-density transformer unit-cell for the 1st-stage LLC-based IBC, one or more of which can be connected to achieve different IBVs. To demonstrate an implementation of the transformer, an 840 W 48 V/1.8 V LLC converter is designed with an array of 14 transformers built into seven unit cells, distributed along the edge of the GPU package to minimize the power distribution network (PDN) losses. The converter module can achieve up to 95.5% efficiency while showcasing a power density of 2200 W/in<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>.

IEEE Transactions on Power ElectronicsVol. 40(10)
Nvidia (United States) (US), Virginia Tech (US)
Affordable and clean energy
Openalex Percentile: Top 90%
Advanced DC-DC Converters
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.