Active System Identification of a DC-DC Converter Using Digital Control

A technique has been developed to actively identify the open-loop characteristics of a DC-DC converter while the unit operates. A digital controller is used to perform both the closed-loop control of the converter and to periodically identify the “plant” using a pseudo-random digital signal. Experimental results are provided for a digitally controlled 90W forward converter to demonstrate accurate and effective identification of the converter control-to-output dynamic response. Hardware requirements and potential fast algorithms are discussed for implementing the real-time converter regulation and identification operations. Nomenclature h(m) = converter impulse response samples (duty cycle control to output voltage) H(jw) = converter frequency response H = converter impulse response vector PRBS = pseudo random binary sequence Ruu = auto-correlation function of input u Ruy = cross-correlation function between u and y u(m) = samples of the PRBS input perturbations U = PRBS input matrix v(m) = system noise samples W = Hadamard matrix WHT = Walsh-Hadamard transform WHT32 = 32-point WHT processing engine y(m) = converter output voltage samples