New DC-DC converters deliver up to 2.6kW with efficiency above 98%, helping reduce conversion losses and thermal stress in dense AI and high-performance computing systems.
OmniOn Power has introduced its second-generation Osprey QODN series of DC-DC bus converters, designed to address the increasing power demands of AI data centres, accelerator servers, high-density computing platforms, and 48V distributed power architectures. The modules provide regulated 48/54V-to-12V conversion while delivering up to 2600W of output power and peak efficiency above 98.5% per cent.
The family includes four models: QODN217 rated at 2600W, QODN167 at 2000W, QODN136 at 1600W, and QODN108 at 1300W. At 54V input and full load, the converters achieve efficiencies of 97.9%, 98%, 98.1%, and 98.1%, respectively. Their high and relatively flat efficiency across the operating range helps minimise power losses as artificial intelligence (AI) workloads vary, supporting better thermal management in continuously operating systems.
The key features are:
- Up to 2.6kW output power
- 40–60V DC input range
- Regulated 12V intermediate bus
- PMBus Rev. 1.2 digital interface
- –40°C to +85°C operating range
The converters operate from a 40V to 60V input range and provide regulated 12V output. They use digital control and synchronous rectification to improve conversion performance while maintaining the same modified DOSA digital quarter-brick footprint used by the previous-generation QODE modules. This allows designers to increase power capability without requiring additional board space.
For system monitoring and control, the QODN family integrates a PMBus interface supporting functions such as remote on/off, remote sensing, power-good signalling, firmware updates over I²C, and black-box fault logging. Protection features cover output overcurrent and overvoltage, overtemperature, and input undervoltage and overvoltage conditions.
The modules operate across –40°C to +85°C, making them suitable for demanding computing and networking environments. Their compact 58.4 × 36.8 × 14.5mm footprint also supports migration from earlier-generation modules. Applications include AI servers, hyperscale and enterprise data centres, networking equipment, supercomputers, storage systems, automatic test equipment, and other distributed power architectures.
By combining higher power density with high conversion efficiency, the QODN series can help reduce wasted energy and the associated cooling burden in next-generation computing infrastructure.
Click here for the original announcement.
