Texas Devices (TI) debuted new design sources and power-management chips to assist corporations meet rising synthetic intelligence (AI) computing calls for and scale power-management architectures from 12V to 48V to 800 VDC. The brand new options will likely be on show at Open Compute Summit (OCP) Oct. 13-16 in San Jose, and embrace:
- “Energy supply trade-offs when getting ready for the following wave of AI computing development”: TI is collaborating with NVIDIA to develop power-management units to help 800 VDC energy structure, as IT rack energy is predicted to eclipse 1MW within the subsequent two to a few years. This white paper reexamines the ability supply structure inside the IT rack, and addresses design challenges and alternatives for top effectivity and excessive power-density power conversion at a system degree.
- Reference design: 30kW AI server power-supply unit: To help stringent AI workloads, TI’s dual-stage power-supply reference design encompasses a three-phase, three-level flying capacitor energy issue correction converter paired with twin delta-delta three-phase inductor-inductor-capacitor converters. The facility provide is configurable as a single 800V output or separate output provides.
- Twin-phase good energy stage:The very best peak energy density energy stage in the marketplace, TI’s CSD965203B provides 100A of peak present per part and combines two energy phases in a single 5mm-by-5mm quad flat no-lead bundle. The machine allows designers to extend part depend and energy supply throughout a small printed circuit board space, enhancing effectivity and efficiency.
- Twin-phase good energy module for lateral energy supply: The CSDM65295 module delivers as much as 180A of peak output present in a compact 9mm-by-10mm-by-5mm bundle, serving to engineers improve information heart energy density with out compromising thermal administration. The module integrates two energy phases and two inductors with trans-inductor voltage regulation (TLVR) choices, whereas sustaining excessive effectivity and dependable operation.
- Gallium-nitride intermediate bus converter: Able to delivering as much as 1.6kW of output energy in a quarter-brick (58.4-mm-by-36.8mm) kind issue, TI’s LMM104RM0 converter module provides over 97.5% input-to-output power-conversion effectivity and excessive light-load effectivity to allow energetic present sharing between a number of modules.
AI information facilities require architectures designed with a number of foundational semiconductors for environment friendly energy administration, sensing and information conversion. With new design sources and a broad power-management portfolio, TI is working alongside information heart designers to implement a complete method that drives environment friendly, secure energy administration – from energy technology on the grid to the elemental logic gates of graphics processing items.
“With the expansion of AI, information facilities are evolving from easy server rooms to extremely subtle energy infrastructure hubs,” stated Chris Suchoski, sector basic supervisor, Information Facilities at TI. “Scalable energy infrastructure and elevated energy effectivity are important to fulfill these calls for and drive future innovation. With units from TI, designers can construct modern, next-generation options which might be enabling the transition to 800 VDC.”
