High density power module for edge AI

Microchip Technology has introduced a cutting-edge point-of-load 12A power module designed specifically for edge AI applications. The MCPF1412 power module features a 16V VIN buck converter and supports I2C and PMBus interfaces, catering to the unique requirements of edge AI designs.

With a compact 5.8 mm × 4.9 mm × 1.6 mm Land Grid Array (LDA) package, the MCPF1412 significantly reduces board space by over 40% compared to traditional discrete designs. This space-saving design is achieved through an interleaved buck converter configuration that minimizes voltage stress on internal power components while maintaining switching losses similar to conventional designs of the same rating. Moreover, the power module incorporates an inherent current-sharing mechanism between its two phases.

One standout feature of the MCPF1412 is its versatility, offering standalone operation without the need for a digital interface. Designers can easily configure output voltages using simple resistor divider adjustments and monitor system performance through the Power Good output.

The power module also boasts a range of diagnostic functions, including over-temperature, over-current, and over-voltage protection, enhancing performance and reliability for edge AI servers, processing cards, and point of load (PoL) applications. Operating within a temperature range of TJ −40°C to +125°C, the MCPF1412 is equipped with an on-board embedded EEPROM for programming the default power-up configuration.

According to Rudy Jaramillo, vice president of Microchip’s analog power interface division, “The MCPF1412 seamlessly integrates with our FPGAs and PCIe solutions, offering Microchip customers a comprehensive solution. This innovative power module optimizes space utilization by reducing chip placements when combined with other Microchip devices.”

For developers looking to evaluate their designs, the MCPF1412 is supported by the EV37R94A Evaluation Board and GUI. Priced at $5.10 each in 10,000-unit quantities, this power module provides a cost-effective and efficient solution for edge AI applications.