By Patrick Le Fevre
Marketing and Communication Director, Ericsson Power Modules
Reducing energy consumption in ICT applications using the dynamic bus voltage architecture
Today, the power architecture increasingly being used in the ICT industry is the Intermediate us Architecture (IBA) which was adopted as the standard in 2003. The I model differs from theIBA classic Distributed Power Architecture (DPA), which typically comprised a number of isolated DC/DC converters on each board that down-convert the -48 VDC line to values that suit the load circuitry, daisy-chaining additional regulators for expediency.
While the DPA model may still suit some small-scale applications, issues arise with efficiently down-converting -48 VDC to logic supply levels of 3.3 VDC or less in one step when traffic and loads varies from low to high. Dynamic bus voltage (DBV) technology is being foreseen as one of the most significant technological breakthroughs for the ICT industry.
System designers are seeking to reduce energy usage at the board level to make significant reduction in the environmental footprint of their systems at times of both high and low data-traffic demand. The introduction of low-power consuming silicon devices alone will not be enough to effectively limit the energy requirements of tomor- row’s network. However, the latest board- power consumption monitoring and control technologies can significantly aid this mission by enabling energy savings of between 3 and 10 percent at board level with the use of advanced DBV architectures. The benefits become increasingly obvious when saving 1W at the board level results in an average saving of 2 to 3W at the system level.
DBV is an evolution of IBA — it pro-
vides systems’ architects the ability to dynamically adjust the power envelope
to meet load conditions. It achieves this by adjusting the previously fixed 12 vDC intermediate bus voltage via the use of advanced digital power control and opti- mized hardware — the Ericsson Advanced bus Converters bMr456 and bMr457 — combined with an energy-optimizer series of algorithms. This can lead to a reduction in both energy consumption and power dissipation, which in turn contributes to a reduction in required cooling.