12.6 Capacitor-current-sensor calibration technique and application in a 4-phase buck converter with load-transient optimization

Abstract

For switching DC-DC converters, a large and rapid load-current transient ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">load</sub> causes a large output voltage undershoot ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">US</sub> and long settling time ts if the transient responses are slow [1]. Since the output capacitor current I <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Co</sub> instantly reflects ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">load</sub> , transient response optimization for minimizing ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">US</sub> and ts with given charging and discharging slopes of inductor current ILo can be achieved by obtaining an accurate I <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Co</sub> and well-controlling the I <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Lo</sub> charging time tch and discharging time tdch [1]. However, ICo sensing accuracy is degraded if the output capacitor's impedance Z <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Co</sub> , comprising the capacitance CO, equivalent series resistance RESR and inductance LESL, varies with different output voltage V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</sub> , process variation, and/or PCB parasitics, leading to non-minimized ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">US</sub> and ts. An off-chip capacitor-current-sensor (CCS) calibration technique is reported in [2]; however, LESL is not considered, so the high-frequency information of ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">load</sub> is inaccurately sensed when large and rapid ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">load</sub> occurs. On the other hand, to well-control the tch and tdch of I <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Lo</sub> , the transient-optimized feedback circuit reported in [1] minimizes ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">US</sub> and t <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">s</sub> . However, optimization can be achieved only under fixed C <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</sub> , output inductor L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</sub> , input voltage V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IN</sub> , and V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</sub> .