Advanced Energy has introduced its ADH series of DC-DC converters aimed at 800 V DC power architectures for AI data centers, targeting higher rack-level power as operators push toward megawatt-class designs.
The ADH series takes an 800 V DC input and converts it to a 50 V DC output. Advanced Energy rates the converters for multi-MHz switching frequency and a peak power efficiency of 98.2%.
In a standard half-brick package, the ADH converters deliver up to 8 kW peak power and 6 kW full-load power. Advanced Energy also specifies peak power density of over 2,700 W/in3 and full-load power density of over 2,000 W/in3.
Thermally, the ADH series is designed to use liquid-cooled baseplates. The company also points to the standard form factor and the ability to parallel modules as a way to scale power. For high-voltage deployments, the ADH design includes insulation intended for systems operating at hazardous input voltage levels.
Moving bulk power at 800 V DC can reduce distribution current for a given kW target, but it also raises the bar for converter efficiency, heat removal, and safety design. Converters like these sit directly in the path between facility-level DC distribution and the low-voltage rails used by AI servers, so power density and thermal design aren’t marketing metrics—they drive enclosure layout, coolant routing, and serviceability.
“Working in close collaboration with several leading customers, we have developed and delivered samples of the ADH converters and plan to expand the portfolio with additional power levels and form factors,” said David Richmond, senior vice president of high-volume products at Advanced Energy.
Advanced Energy says the ADH series can be combined with its NDQ series of 50 V DC to 12 V DC non-isolated bus converter (NIBC) and a newly developed Hot Swap Control (HSC) module, which it says manages inrush current and provides additional protection functions, to form a complete 800 V DC power solution for higher-power, higher-density AI server racks.
Source: Advanced Energy












