Extracting the frequency response of switching DC-DC converters in CCM and DCM from time-domain simulations

Determining if a feedback loop is stable is crucial, and in the case of switching dc-dc converters, also considerably involved because some of the signals across the ac signal path are large signal and nonlinear. Today, designers rely on averaged and linearized small-signal models to approximate loop dynamics. This method is appropriate for well-tested dc-dc supplies like the buck and boost but less reliable and more difficult to apply to emerging non-conventional switchers used, for example, in energy harvesters. Overdesigning these circuits to overcome the uncertainty is often unacceptable because accuracy and efficiency, which are critical, normally suffer as a result. Extracting loop dynamics from actual time-domain waveforms is more accurate because, in averaging waveforms around a narrow bias point, conventional methods eliminate important loop information and constrain the model to small-signal variations - this paper proposes a way to extract small- and large-signal frequency response from time-domain simulations. The results from buck and boost converters under voltage- and current-mode control operating in CCM and DCM match theory well.