A More Accurate Means of Comparing DC-DC Converter Performance

Apr 20, 2005 6:54 PM
By David Morrison, Editor, PETech

A new metric for comparing the performance of dc-dc converters provides a more accurate reflection of real world performance than converter efficiency alone. This measurement tool, called effective output loss resistance, was introduced by Fairchild Semiconductor’s Alan Elbanhawy in his plenary session presentation on April 18 at the Portable Power Developer's Conference in San Jose, Calif. Elbanhawy is the director of the computing and telecommunications segments, Advanced Power Systems Center at Fairchild.

In his talk, “Is Power Conversion Efficiency Running out of Steam as a Comparison Tool?” Elbanhawy described why efficiency is lacking as a means for comparing dc-dc converters from different sources. Elbanhawy notes that the typical efficiency-versus-load current graph is only a valid basis for comparison in an “ideal world” where a number of dc-dc converter parameters are fixed. In such a world, I/O parameters, such as input voltage, output voltage and load current range, would all be fixed. Moreover, switching frequency also would be fixed and only one heatsinking scheme would be used in all cases. Of course, for most dc-dc converters, all of these parameters vary from application to application.

As the speaker shows in presenting voltage regulator module (VRM) performance data, all of the above parameters affect converter efficiency. In particular, the output voltage and heatsinking scheme have a significant impact on efficiency. This effect is noticeable even when looking at the efficiency data for a given VRM model. Therefore, comparing the efficiency data of converters from different sources can be misleading when those converters are operating at different output voltages and with different heatsinking arrangements.

The new parameter, effective output loss resistance (Rol), can account for both output voltage and the effect of heatsinking. Rol is defined as the (power loss at a given load current)/(load current²). When Elbanhawy graphs a converter’s Rol versus load current at different output voltages, the resulting curves are tightly spaced. A similar result is obtained when a converter’s Rol is plotted versus load current at a fixed output voltage but under different heatsinking conditions. In both graphs, the tight spacing of curves makes it convenient to express Rol (at any given value of load current) as some value in milliohms plus or minus a tolerance that reflects the small variation in the curves.

Elbanhawy goes on to explain that curve fitting techniques may be used to derive an equation for Rol as a function of load current. This expression then can be used to analytically compare different dc-dc converter sources. This approach then overcomes the differences in test conditions that prevent direct comparisons of efficiency data published by the various dc-dc converter suppliers.

If users must generate the Rol data themselves, this represents extra work on their part. However, the dc-dc converter manufacturers could eliminate this burden by publishing the curves or the equation describing Rol as a function of load current. Doing so would permit easy comparisons of their products. Customers then would seek out the converter with the lowest value of Rol and smallest variation in Rol at the load current of interest.

For more information on this presentation, contact Louise Merriman, Integrated Marketing Manager for Fairchild Semiconductor at Louise.Merriman@fairchildsemi.com or visit www.fairchildsemi.com.