Power-supply designers routinely tackle new dc-dc converter design tasks with the goal of achieving the highest efficiency possible. Clearly, this task is a very complex one. Once the topology has been selected based on the specifics of the application, a designer must go through a very rigorous process of component selection. This process includes selection of the PWM controller, power MOSFETs and gate drivers as well as the input and output filter inductors and capacitors.

Usually, the one overruling goal of the designer is to achieve the highest power conversion efficiency at the maximum load current, while worrying much less about the converter’s efficiency at low current levels.

Depending on the application, a power designer may find this goal sufficient if the application requires the load current to be in the range of 70% to 100% of the full load. Such a design clearly will render some very satisfactory performance results since the efficiency curve of a well-designed dc-dc converter tends not to vary much in this range. So, the user will — more or less — get the same efficiency within approximately ±1.5%.

On the other hand, in a very demanding application such as the core converter in a notebook computer, the situation is very different. Here, the load histogram shows long periods of steady-state current in the lower percentiles of the load range interrupted by shorter periods of medium to high currents up to the full load.

Nonetheless, even in notebook applications it seems that dc-dc converter designers still place an overwhelming emphasis on the full-load-current efficiency. These designers focus to a much lesser degree on optimizing efficiencies at currents on the low end of the range.

If we examine this situation closely, one can see that lower efficiencies at low currents will directly negatively impact the battery life. That’s because, for a very large percentage of the time, the notebook computer operates at these low current levels.

Some very simple and effective techniques have been employed to partially remedy this situation by allowing the PWM controller to operate in pulse-skipping or constant-on time modes at low currents and move to full PWM control at higher currents. This results in improved efficiencies up to a certain limit, but much more can be done to improve overall efficiency.

Innovative ideas can be implemented to achieve a “flat” power conversion efficiency curve, where the dc-dc converter delivers high efficiency over the entire current range. Here, the enabling technology consists primarily of improved or “intelligent” gate-driver schemes. Several companies have already received patents for such schemes, which — when implemented — should produce robust designs that directly improve the notebook’s battery life without adding any extra burden on the cooling system.

The above-mentioned innovative ideas must balance the demands of the circuit over the entire current range. Such a feat is not easily achieved because the demand for high efficiency at full load requires smaller MOSFET which can only be achieved using larger die sizes with larger gate capacitances, C_{GS} and C_{GD}.

These capacitances play a major role in determining the dynamic losses; the larger the value of these capacitances, the larger the dynamic losses. The loss mechanisms at low currents then become dominated by the dynamic losses, which are directly dependent on the switching frequency. So, concerns over dynamic losses at light loads make the decision to move to higher switching frequencies a very difficult one.

The dominance of the dynamic losses is due to the selection of an small enough to generate low conduction losses at the maximum current. That minimization of makes the conduction losses at low currents almost negligible since conduction losses are approximately proportional to I^{2}.

In the face of these technical challenges, I believe that the move to higher efficiency over the entire load range in notebook computers has largely stalled. Why? Because designing for higher efficiency over such a wide load range has a negative impact on the cost of the overall bill of materials. Even as notebook computers continue to provide greater performance (including more efficient power conversion), notebook pricing continues to fall. Going forward, these are issues that component suppliers, computer manufacturers and even consumers will need to grapple with as energy efficiency continues to become a more pressing issue for everyone.