The demand for power handling delivered by board-mounted power supply "bricks" is going through the roof. As we all know, the amount of data transferred through networks is increasing in almost exponential leaps and bounds, in part due to the huge growth in mobile traffic.
Correspondingly, the power demand of computing and networking boards used in the Information and Communications Technology sector is increasing from 300 W up to 1.0 and 1.2 kW and is expected to go even higher; it's expected to reach 3 kW per board by the middle of the decade, with some studies forecasting levels of up to 5 kW by 2020.
When the high-power brick industry emerged a couple of decades ago, the practical limit was around 100 W to 150 W. But now there are suppliers offering quarter-brick intermediate bus converters (IBCs) with several hundreds of watts (in fact Ericsson's latest PKM-NH series now delivers up to 864 W) with power density figures that could not even be imagined a few years ago.
Ever increasing efficiencies, advanced components, and packaging are making power ratings ever higher. Power density is now an order of magnitude higher than that of technology used 15 years ago -- the latest power bricks in the market offer an impressive 37 W/cm3 (600 W/in3).
Some may question why we are reaching these levels, and why not split it into three smaller boards making it easier to both power and cool? While many have considered this approach, the solution is not so easy -- available rack space is an issue for example.
However, today, paralleling quarter-bricks is probably the smart choice for 1.2-kW boards. But even if a 1-kW quarter-brick is conceivable in the near future to power 3-kW or even 5-kW boards, systems and board-power architects are questioning the electrical and thermal benefits to parallel this type of brick and are increasingly inviting power supply manufacturers to rethink their board-power modules, which in terms of packaging actually have not evolved a great deal over the past couple of decades.
Board power architects will also have to consider how to connect the output of these power bricks to additional rails that will be needed to distribute 250 A or more to end loads. Another issue is that of thermal management, in the power brick as well as in end-user systems. Systems architects are having to consider new mechanical platforms based on bigger boards and a chassis designed to accommodate new cooling methods such as refrigerated cold-walls, making it possible to cool systems with more efficiency than that offered by conventional air ventilation.
Considering these electrical and thermal constraints, multi-kilowatt boards could mean a totally new type of board-power source that takes up more space to facilitate thermal exchange, and with interconnection that is better designed for very high currents, in addition to brand new assembly techniques. Some interesting challenges are ahead.