Harish K. Krishnamurthy
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Harish K. Krishnamurthy has a total of 94 co-authors. Here are the top co-authors by number of shared publications:
- James W. Tschanz
Intel Labs, Intel Corporation, Hillsboro, OR, USA
16 shared publications - Krishnan Ravichandran
Intel Labs, Intel Corporation, Hillsboro, OR, USA
16 shared publications - Vivek De
Intel Labs, Intel Corporation, Hillsboro, OR, USA
16 shared publications Sheldon Weng
Intel Labs, Intel Corporation, Hillsboro, OR, USA
12 shared publications- Xiaosen Liu
Tsinghua University
10 shared publications - Khondker Zakir Ahmed
Intel Corporation, Hillsboro, OR, USA
8 shared publications
Latest Publications
- 5 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (4) , 1265-1276
A Monolithic 5.7 A/mm2 91% Peak Efficiency Scalable Multistage Modular Switched Capacitor Voltage Regulator for Base Die Vertical Power Delivery in 3D-ICs
The rapid increase of silicon die area and power requirements of modern processors, requires a minimal power delivery footprint. Unlike conventional lateral power delivery (LPD), vertical power delivery (VPD) with high input voltage offers a higher density solution, freeing up the top-side package area for more compute and memory dice, and significantly improving overall energy efficiency. However, the development of high-voltage and high-efficiency voltage regulators (VRs) with low-profile passives has lagged behind these advancements. This article presents a monolithic multistage modular switched capacitor VR (MMSCVR) in 16 nm CMOS for VPD. It features: 1) scalable design in both current and voltage domains; 2) optimal deadtime control with self-timed deadtime generator (STDG); and 3) energy efficient, electrical overstress (EOS) free start-up sequence with multilevel mid-rail voltage generator (MMVG). The proposed solution demonstrates 90.6% peak efficiency and 5.7 A/mm2 peak current density with a 3-to-1 V conversion— 4.9× higher than state-of-the-art. The 4-to-1 V variant is the first high-density fully monolithic 4-to-1 V SCVR, achieving 7.8× higher current density than any prior art using off-chip capacitors.
- 5 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (4) , 1265-1276
A Monolithic 5.7 A/mm2 91% Peak Efficiency Scalable Multistage Modular Switched Capacitor Voltage Regulator for Base Die Vertical Power Delivery in 3D-ICs
The rapid increase of silicon die area and power requirements of modern processors, requires a minimal power delivery footprint. Unlike conventional lateral power delivery (LPD), vertical power delivery (VPD) with high input voltage offers a higher density solution, freeing up the top-side package area for more compute and memory dice, and significantly improving overall energy efficiency. However, the development of high-voltage and high-efficiency voltage regulators (VRs) with low-profile passives has lagged behind these advancements. This article presents a monolithic multistage modular switched capacitor VR (MMSCVR) in 16 nm CMOS for VPD. It features: 1) scalable design in both current and voltage domains; 2) optimal deadtime control with self-timed deadtime generator (STDG); and 3) energy efficient, electrical overstress (EOS) free start-up sequence with multilevel mid-rail voltage generator (MMVG). The proposed solution demonstrates 90.6% peak efficiency and 5.7 A/mm2 peak current density with a 3-to-1 V conversion— 4.9× higher than state-of-the-art. The 4-to-1 V variant is the first high-density fully monolithic 4-to-1 V SCVR, achieving 7.8× higher current density than any prior art using off-chip capacitors.
- 7 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 59  (12) , 4114-4123
A Monolithic 12.7 W/mm2, 92% Peak-Efficiency Switched-Capacitor DC-DC Converter Using CSCR-First Topology
This article introduces the continuously scalable conversion-ratio (CSCR)-first topology for monolithic switched-capacitor voltage regulators (SCVRs), substantially improving the performance of CSCR SCVRs for higher input voltages. The topology combines fixed-ratio 2:1 stages together with a CSCR stage to limit the voltage across the CSCR stage while boosting the allowable input voltage higher without relying on stacking capacitors or transistors. Furthermore, by running the fixed-ratio and CSCR stage out of phase, the effectiveness of the CSCR’s intermediate rails and switches is doubled, thus reducing the required transistor footprint. This article discusses a scalable implementation of this technology, leveraging an array-able voltage regulator (VR) core that encompasses the required powertransistors, capacitors, as well as local signal generation circuits, that is then combined with a single centralized controller to meet the current demands of small and large power domains. Measurement results of the VR, realized on Intel 16 technology using high-density metal-insulator–metal (MIM) capacitors, demonstrate the transient performance and reliability of the proposed approach, while showcasing a record 12.7 W/mm 2 peak power density for monolithic voltage conversion, and 92% peak efficiency.
- 7 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 59  (12) , 4114-4123
A Monolithic 12.7 W/mm2, 92% Peak-Efficiency Switched-Capacitor DC-DC Converter Using CSCR-First Topology
This article introduces the continuously scalable conversion-ratio (CSCR)-first topology for monolithic switched-capacitor voltage regulators (SCVRs), substantially improving the performance of CSCR SCVRs for higher input voltages. The topology combines fixed-ratio 2:1 stages together with a CSCR stage to limit the voltage across the CSCR stage while boosting the allowable input voltage higher without relying on stacking capacitors or transistors. Furthermore, by running the fixed-ratio and CSCR stage out of phase, the effectiveness of the CSCR’s intermediate rails and switches is doubled, thus reducing the required transistor footprint. This article discusses a scalable implementation of this technology, leveraging an array-able voltage regulator (VR) core that encompasses the required powertransistors, capacitors, as well as local signal generation circuits, that is then combined with a single centralized controller to meet the current demands of small and large power domains. Measurement results of the VR, realized on Intel 16 technology using high-density metal-insulator–metal (MIM) capacitors, demonstrate the transient performance and reliability of the proposed approach, while showcasing a record 12.7 W/mm 2 peak power density for monolithic voltage conversion, and 92% peak efficiency.
- 3 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 75-84
A Monolithic, 10.5 W/mm2, 600 MHz Top-Metal and C4 Planar Spiral Inductor-Based Integrated Buck Voltage Regulator on 16 nm Class
A monolithic buck voltage regulator (VR) built with top-metal and industry-first C4 planar spiral inductors demonstrates 2.5 times higher power density than prior art, providing efficient alternative to low-dropout linear regulators (LDOs) powering disaggregated digital, IO and PHY IPs. The 1.1 V single-stack power stage, DPWM delay-line, self-triggered windowed-flash ADC, and fully digital and reconfigurable controller are designed for high frequency, delivering 10.5 W/mm $^{{2}}$ at 600 MHz, with 0.35–0.85 V operating range and 18%–25% improvement in efficiency over an ideal LDO.
- 3 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 75-84
A Monolithic, 10.5 W/mm2, 600 MHz Top-Metal and C4 Planar Spiral Inductor-Based Integrated Buck Voltage Regulator on 16 nm Class
A monolithic buck voltage regulator (VR) built with top-metal and industry-first C4 planar spiral inductors demonstrates 2.5 times higher power density than prior art, providing efficient alternative to low-dropout linear regulators (LDOs) powering disaggregated digital, IO and PHY IPs. The 1.1 V single-stack power stage, DPWM delay-line, self-triggered windowed-flash ADC, and fully digital and reconfigurable controller are designed for high frequency, delivering 10.5 W/mm $^{{2}}$ at 600 MHz, with 0.35–0.85 V operating range and 18%–25% improvement in efficiency over an ideal LDO.
- Top Cited PapersOpen Access
- 28 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 39  (3) , 3391-3410
Mini-LEGO CPU Voltage Regulator
This article presents the design and optimization of a miniaturized 48-V-to-1-V, 240-A linear-extendable group operated point-of-load (LEGO-PoL) CPU voltage regulator module (VRM)— Mini-LEGO —with a volume of 30 mm×11.2 mm × 8.4 mm. The Mini-LEGO converter provides vertical power delivery, and it achieves a peak efficiency of 87.1%, a full load efficiency of 84.1%, and a power density of 1390 W/in $^{3}$ when stepping down a 48 V input voltage to an output voltage of 1 V. The Mini-LEGO offers compact device implementation, simple gate drive circuitry, and a fast regulation-stage switching frequency of 1.5 MHz. The 9-mm-by-9-mm, 4-phase, 20-A-per-phase vertical coupled inductor has a per-phase transient inductance of 10.3 nH, with its coupling reducing the ripple by a factor of six, and a height of 2.5 mm. This vertical coupled inductor structure is designed following a systematic approach to minimizing its height while achieving significant current ripple reduction and maintaining sufficient saturation margin. Thermal simulations are performed to verify the cooling feasibility. A measurement characterization method for vertical coupled inductor structures is introduced and used to characterize the designed vertical coupled inductor.
- 28 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 39  (3) , 3391-3410
Top Cited PapersOpen AccessMini-LEGO CPU Voltage Regulator
This article presents the design and optimization of a miniaturized 48-V-to-1-V, 240-A linear-extendable group operated point-of-load (LEGO-PoL) CPU voltage regulator module (VRM)— Mini-LEGO —with a volume of 30 mm×11.2 mm × 8.4 mm. The Mini-LEGO converter provides vertical power delivery, and it achieves a peak efficiency of 87.1%, a full load efficiency of 84.1%, and a power density of 1390 W/in $^{3}$ when stepping down a 48 V input voltage to an output voltage of 1 V. The Mini-LEGO offers compact device implementation, simple gate drive circuitry, and a fast regulation-stage switching frequency of 1.5 MHz. The 9-mm-by-9-mm, 4-phase, 20-A-per-phase vertical coupled inductor has a per-phase transient inductance of 10.3 nH, with its coupling reducing the ripple by a factor of six, and a height of 2.5 mm. This vertical coupled inductor structure is designed following a systematic approach to minimizing its height while achieving significant current ripple reduction and maintaining sufficient saturation margin. Thermal simulations are performed to verify the cooling feasibility. A measurement characterization method for vertical coupled inductor structures is introduced and used to characterize the designed vertical coupled inductor.
- 19 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Letters
- Vol. 6, 273-276
A Monolithic 26 A/mm2 Continuously Scalable Conversion Ratio Switched-Capacitor Converter With Phase-Merging Turbo and Communication-Less Ganging
This paper introduces the Phase-Merging Turbo (PMT) technique, a method which significantly augments the output current capability of a Continuous Scalable Conversion-Ratio (CSCR) Switched-Capacitor Voltage Regulator (SCVR). The research also proposes a unique method for implementing communication-free ganging with these converters, enhancing their scalability across a wide range of power domain sizes. Fabricated using a 4nm class CMOS technology, this study achieves a current-density of 26A/mm2 for monolithic capacitive voltage regulators, and a peak efficiency of 88.5%.
- 19 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Letters
- Vol. 6, 273-276
A Monolithic 26 A/mm2 Continuously Scalable Conversion Ratio Switched-Capacitor Converter With Phase-Merging Turbo and Communication-Less Ganging
This paper introduces the Phase-Merging Turbo (PMT) technique, a method which significantly augments the output current capability of a Continuous Scalable Conversion-Ratio (CSCR) Switched-Capacitor Voltage Regulator (SCVR). The research also proposes a unique method for implementing communication-free ganging with these converters, enhancing their scalability across a wide range of power domain sizes. Fabricated using a 4nm class CMOS technology, this study achieves a current-density of 26A/mm2 for monolithic capacitive voltage regulators, and a peak efficiency of 88.5%.
- 10 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Letters
- Vol. 6, 101-104
A Digital LDO in 22-nm CMOS With a 4-b Self-Triggered Binary Search Windowed Flash ADC Featuring Analog Layout Generator Framework
This paper presents an analog layout generator based DLDO with a self-triggered binary search windowed flash ADC in 22nm CMOS. A self-triggered binary search mechanism with a delay-based architecture is proposed to reduce the exponentially growing kickback noise and energy consumption of a traditional flash ADC down to the level of a SAR ADC while maintaining its high-speed feature. To conquer the complexity bottleneck of SoC development in FinFET technology, a practical analog layout generation framework is proposed to maximize the productivity of implementing analog circuit blocks in scaled CMOS process. To meet the performance, area and reliability specifications across a variety of circuits, the methodology allows varying levels of constraints from designers, thus significantly improving the physical design time & effort up to 60× compared with conventional manual approach. The DLDO features 3.55ps FoM and fully automatic generation.
- 10 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Letters
- Vol. 6, 101-104
A Digital LDO in 22-nm CMOS With a 4-b Self-Triggered Binary Search Windowed Flash ADC Featuring Analog Layout Generator Framework
This paper presents an analog layout generator based DLDO with a self-triggered binary search windowed flash ADC in 22nm CMOS. A self-triggered binary search mechanism with a delay-based architecture is proposed to reduce the exponentially growing kickback noise and energy consumption of a traditional flash ADC down to the level of a SAR ADC while maintaining its high-speed feature. To conquer the complexity bottleneck of SoC development in FinFET technology, a practical analog layout generation framework is proposed to maximize the productivity of implementing analog circuit blocks in scaled CMOS process. To meet the performance, area and reliability specifications across a variety of circuits, the methodology allows varying levels of constraints from designers, thus significantly improving the physical design time & effort up to 60× compared with conventional manual approach. The DLDO features 3.55ps FoM and fully automatic generation.
- 6 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2-4
A Digital LDO in 22nm CMOS with a 4b Self-triggered Binary Search Windowed Flash ADC Featuring Automatic Analog Layout Generator Framework
Complex compute and communication SoCs in scaled CMOS process integrate synthesizable digital logic along with many analog/mixed-signal circuits such as PLLs and LDOs. While the digital logic blocks can be ported easily across process nodes, the analog/mixed-signal circuits demand custom manual re-design for physical layout, verification and optimization. Furthermore, fine-grain power management and noise isolation necessitate implementation of a large number (>50) of independent voltage domains on die. The manual layout effort and time for such a large variety of critical analog/mixed-signal circuit blocks becomes increasingly unaffordable in practice. Also, physical layout requirements like common-centroid and inter-digitization, and routing requirements such as matching differential nets, minimizing interconnect parasitics of critical leaf cells like differential pairs, current mirrors, resistors and capacitor arrays, necessitate multiple iterations with post-layout verifications, thus significantly impacting overall design time and time-to-market. Furthermore, availability of high-quality mask-design engineers can pose a major bottleneck to timely completion of the design, especially for the increasingly complex layout design rules in scaled process nodes. Multiple EDA frameworks, such as ALIGN [1], BAG [2] and MAGICAL [3] were reported to generate analog layouts for improving design productivity of analog circuits in scaled CMOS. However, there is still a need for a methodology that can accommodate the varying needs of analog circuits and generate high quality layout while leveraging automated placement and/or routing. In this paper, we present an analog layout generator (ALG) methodology based digital LDO (DLDO) with a self-triggered binary search windowed flash ADC in 22nm CMOS (Fig. 1) to maximize the productivity of implementing analog circuit blocks in scaled CMOS process, thus significantly improving the overall design time & effort, as well as time-to-market for complex SoCs. A self-triggered binary search mechanism with a delay-based architecture is proposed to reduce the exponentially growing kickback noise and energy consumption of a traditional flash ADC down to the level of a SAR ADC while maintaining its high speed feature.
- 6 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2-4
A Digital LDO in 22nm CMOS with a 4b Self-triggered Binary Search Windowed Flash ADC Featuring Automatic Analog Layout Generator Framework
Complex compute and communication SoCs in scaled CMOS process integrate synthesizable digital logic along with many analog/mixed-signal circuits such as PLLs and LDOs. While the digital logic blocks can be ported easily across process nodes, the analog/mixed-signal circuits demand custom manual re-design for physical layout, verification and optimization. Furthermore, fine-grain power management and noise isolation necessitate implementation of a large number (>50) of independent voltage domains on die. The manual layout effort and time for such a large variety of critical analog/mixed-signal circuit blocks becomes increasingly unaffordable in practice. Also, physical layout requirements like common-centroid and inter-digitization, and routing requirements such as matching differential nets, minimizing interconnect parasitics of critical leaf cells like differential pairs, current mirrors, resistors and capacitor arrays, necessitate multiple iterations with post-layout verifications, thus significantly impacting overall design time and time-to-market. Furthermore, availability of high-quality mask-design engineers can pose a major bottleneck to timely completion of the design, especially for the increasingly complex layout design rules in scaled process nodes. Multiple EDA frameworks, such as ALIGN [1], BAG [2] and MAGICAL [3] were reported to generate analog layouts for improving design productivity of analog circuits in scaled CMOS. However, there is still a need for a methodology that can accommodate the varying needs of analog circuits and generate high quality layout while leveraging automated placement and/or routing. In this paper, we present an analog layout generator (ALG) methodology based digital LDO (DLDO) with a self-triggered binary search windowed flash ADC in 22nm CMOS (Fig. 1) to maximize the productivity of implementing analog circuit blocks in scaled CMOS process, thus significantly improving the overall design time & effort, as well as time-to-market for complex SoCs. A self-triggered binary search mechanism with a delay-based architecture is proposed to reduce the exponentially growing kickback noise and energy consumption of a traditional flash ADC down to the level of a SAR ADC while maintaining its high speed feature.
- 28 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 57  (4) , 1090-1099
A 32-A, 5-V-Input, 94.2% Peak Efficiency High-Frequency Power Converter Module Featuring Package-Integrated Low-Voltage GaN nMOS Power Transistors
A buck converter using a low-voltage GaN nMOS power transistor with 5–$10\times $ superior switching figure-of-merit over Si laterally-diffused MOS (LDMOS) and stacked CMOS at 5 V has been employed to build a high-frequency (3–20 MHz), high-density (9 A/mm 2 ) buck converter for high-performance-compute applications. The GaN-based converter is co-packaged with a CMOS Companion Die on a 4 mm$\times $ 4 mm package and employs on-die gate clamps to minimize the impact of package parasitics for high efficiency. For 5-V input to 1-V output conversion, the converter achieves 94.2% peak efficiency at 3.1-MHz switching frequency with a 40-nH inductor, and >80% peak efficiency at 20 MHz with an air-core inductor.- 28 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 57  (4) , 1090-1099
A 32-A, 5-V-Input, 94.2% Peak Efficiency High-Frequency Power Converter Module Featuring Package-Integrated Low-Voltage GaN nMOS Power Transistors
A buck converter using a low-voltage GaN nMOS power transistor with 5–$10\times $ superior switching figure-of-merit over Si laterally-diffused MOS (LDMOS) and stacked CMOS at 5 V has been employed to build a high-frequency (3–20 MHz), high-density (9 A/mm 2 ) buck converter for high-performance-compute applications. The GaN-based converter is co-packaged with a CMOS Companion Die on a 4 mm$\times $ 4 mm package and employs on-die gate clamps to minimize the impact of package parasitics for high efficiency. For 5-V input to 1-V output conversion, the converter achieves 94.2% peak efficiency at 3.1-MHz switching frequency with a 40-nH inductor, and >80% peak efficiency at 20 MHz with an air-core inductor.- 18 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 41  (11) , 4927-4938
EM SCA White-Box Analysis-Based Reduced Leakage Cell Design and Presilicon Evaluation
This work presents a white-box modeling of the electromagnetic (EM) leakage from an integrated circuit (IC) to develop EM side-channel analysis (SCA)-aware design techniques. A new digital library cell layout design technique is proposed to minimize the EM leakage and is evaluated using a high-frequency structure simulator (HFSS)-based framework. Backed by our physics-based understanding of EM radiation, the proposed double-row power grid-based digital cell layout design shows $>5\times $ reduction in the EM SCA leakage compared to the traditional digital logic gate layout design. Furthermore, exploiting the magneto-quasistatic (MQS) regime of operation of the EM leakage from the CMOS circuits, the HFSS-based framework is utilized to develop a pre-silicon (Si) EM SCA evaluation technique to assess the vulnerability of cryptographic implementations against such attacks during the design phase itself.
- 18 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 41  (11) , 4927-4938
EM SCA White-Box Analysis-Based Reduced Leakage Cell Design and Presilicon Evaluation
This work presents a white-box modeling of the electromagnetic (EM) leakage from an integrated circuit (IC) to develop EM side-channel analysis (SCA)-aware design techniques. A new digital library cell layout design technique is proposed to minimize the EM leakage and is evaluated using a high-frequency structure simulator (HFSS)-based framework. Backed by our physics-based understanding of EM radiation, the proposed double-row power grid-based digital cell layout design shows $>5\times $ reduction in the EM SCA leakage compared to the traditional digital logic gate layout design. Furthermore, exploiting the magneto-quasistatic (MQS) regime of operation of the EM leakage from the CMOS circuits, the HFSS-based framework is utilized to develop a pre-silicon (Si) EM SCA evaluation technique to assess the vulnerability of cryptographic implementations against such attacks during the design phase itself.
- Top Cited PapersOpen Access
- 14 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 37  (6) , 6305-6322
Vertical Stacked LEGO-PoL CPU Voltage Regulator
This article presents a 48–1 V merged-two-stage hybrid-switched-capacitor converter with a linear extendable group operated point-of-load (LEGO-PoL) architecture for ultrahigh-current microprocessors, featuring 3-D stacked packaging and coupled inductors for miniaturized size, fast speed, and vertical power delivery. The architecture is highly modular and scalable. The switched-capacitor circuits are connected in series on the input side to split the high input voltage into multiple stacked voltage domains. The multiphase buck circuits are connected in parallel to distribute the high output current into multiple parallel current paths. It leverages the advantages of switched-capacitor circuits and multiphase buck circuits to achieve soft charging, current sharing, and voltage balancing. The inductors of the multiphase buck converters are used as current sources to soft-charge and soft-switch the switched-capacitor circuits, and the switched-capacitor circuits are utilized to ensure current sharing among the multiphase buck circuits. A 780-A vertical stacked CPU voltage regulator with a peak efficiency of 91.1% and a full load efficiency of 79.2% at an output voltage of 1 V with liquid cooling is built and tested. The switched capacitor circuits operate at 286 kHz and the buck circuits operate at 1 MHz. It regulates output voltage between 0.8 and 1.5 V through the entire 780-A current range. This is the first demonstration of a 48–1 V CPU voltage regulator to achieve over 1-A/mm $^\text{2}$ current density and the first to achieve 1000-W/in $^3$ power density.
- 14 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 37  (6) , 6305-6322
Top Cited PapersOpen AccessVertical Stacked LEGO-PoL CPU Voltage Regulator
This article presents a 48–1 V merged-two-stage hybrid-switched-capacitor converter with a linear extendable group operated point-of-load (LEGO-PoL) architecture for ultrahigh-current microprocessors, featuring 3-D stacked packaging and coupled inductors for miniaturized size, fast speed, and vertical power delivery. The architecture is highly modular and scalable. The switched-capacitor circuits are connected in series on the input side to split the high input voltage into multiple stacked voltage domains. The multiphase buck circuits are connected in parallel to distribute the high output current into multiple parallel current paths. It leverages the advantages of switched-capacitor circuits and multiphase buck circuits to achieve soft charging, current sharing, and voltage balancing. The inductors of the multiphase buck converters are used as current sources to soft-charge and soft-switch the switched-capacitor circuits, and the switched-capacitor circuits are utilized to ensure current sharing among the multiphase buck circuits. A 780-A vertical stacked CPU voltage regulator with a peak efficiency of 91.1% and a full load efficiency of 79.2% at an output voltage of 1 V with liquid cooling is built and tested. The switched capacitor circuits operate at 286 kHz and the buck circuits operate at 1 MHz. It regulates output voltage between 0.8 and 1.5 V through the entire 780-A current range. This is the first demonstration of a 48–1 V CPU voltage regulator to achieve over 1-A/mm $^\text{2}$ current density and the first to achieve 1000-W/in $^3$ power density.