Energy in, Energy out

I found the article "Kilowatts on order" [IEEE Spectrum, February 1993, p. 32-37] interesting and informative. However, it seemed to gloss over several crucial aspects of PWM [pulse-width modulation] amplifier operation.

For example, the sidebar on p. 34 states, "During the on period, the entire DC supply voltage, ... is applied to the load." Of course, this cannot be true, or else the output voltage would equal the supply voltage. In order for the output voltage to differ from the supply voltage, they must be separated by a component that generates a voltage drop. In fact, the supply voltage is applied to the output filter, and the inductor generates the necessary voltage drop.

On this analysis, we see that a PWM amplifier must hold off from the load whatever voltage it does not deliver, just as a linear amplifier must. The reason that a PWM amplifier is more efficient is that it does this with a component that stores energy (an inductor) rather than a component that dissipates energy (a transistor). The stored energy is later delivered to the load, or returned to the supply rails.

All of this is implicit in the statement, "Output filtering attenuates harmonics of the switching frequency...." However, those accustomed to thinking of filters as signal-conditioning devices may not fully appreciate the significance of the output filter in managing the flow of energy and current between the supply and the load.

IEEE Spectrum, 1993 March

Steven W. McDougall / resume / / 1993 May