Distribution vs. Efficiency: Power Architect's Dilemma

Oct 1, 2003 12:00 PM
By Arnold Alderman, President, Anagenesis Inc., Los Angeles

The age-old saying “stepping stones can become stumbling blocks” certainly applies to large systems power distribution. Implemented as distributed power architecture (DPA), intermediate bus architecture (IBA), and now even factorized power architecture (FPA), power architects continue to struggle with conveying power from line to load. They try to balance cost, efficiency and performance. In all the discussion, little consideration is paid to the overall line-to-load efficiency (LTLE). In fact, if we look closely, the line-to-load efficiency hasn't really improved much over the past 20 years.

Each power-supply element is appropriately touted as having respectable efficiencies above 90%. But is that so wonderful? Not when we look at the overall system efficiency. Power is delivered to do a job. Several series power supplies or converters are in the system's DPA. When all the system elements are taken into account, the LTLE is 80%, 81% or a whopping 82%. That is still disappointing, as nearly 20% of power is wasted in heat. Thus, we have distributed power and distributed heat.

Many arguments are presented regarding the optimization of the bus voltages and elements to minimize cost. However, many are concerned about minimizing losses. Major ongoing regional, national and global efforts are trying to reduce power consumption. Agencies, such as the EPA, EPRI, E2I and CEC, are passionately trying to reduce consumption by championing improved internal system power-conveyance efficiency. An 8% LTLE improvement would have a significant impact on power consumption. Imagine an LTLE of 87%, 88% or even 90%. That would mean a savings of 120W on a 1.5kW system. Is this kind of improvement feasible? Definitely.

How? First, the efficiency of each element in series string might be increased — the front-end rectifier, the IBA converter, the brick or the POL. This is an obvious answer, but costly. Squeezing out that extra 3% might price you out of the market. Second, you can jump to a new architecture with fewer elements — a probable strategy, especially if intermediate bus voltages are essential for merging power sources such as standby battery power. Last, but least considered, reduce the number of series converter elements to one — direct conversion from ac to low-voltage dc.

Yes, direct ac-dc conversion is going right back to where we started, but doesn't it make sense? If we want to have the most efficient system, we need to accomplish the task most directly with the fewest elements. Now a 50W or a 150W, 115V to 1.5V ac-dc power supply can be readily designed to achieve higher than 90% efficiency. Even a low-cost version could achieve 89%

So, where are the stumbling blocks? First and foremost, it isn't acceptable safety-wise to have high voltage on the system board. That was solved long ago with low-voltage ac distribution (48Vac or 24Vac). The hot swap inrush transients will have to be minimized at the input to the system board. Multiple secondary voltages will be needed. What about battery backup?

This idea isn't so absurd. It's being seriously considered by leading power supply design groups for some applications. Recently introduced power-supply components and advance component-eliminating designs are enabling this approach to be feasible.