Unless the current limit of the regulator clips the resulting spikes, the spikes are equal to the regulator’s output-voltage rate of rise multiplied by the sum of the parallel output caPACitances: IMAX=" dV/dt "×COUT, where IMAX is the maximum current, dV/dt is a differential in voltage with respect to a differential in time, and COUT is the output capacitance. The math suggests that the best strategy for limiting the regulator’s turn-on maximum current is to limit dV/dt. The circuit in Figure 1 relies on this trICk and works with industry-standard adjustable linear regulators, such as the popular low-dropout LM2941.

　　The basis of the dV/dt-limiting technique comprises the six added components: R3, R4, CT, D1, D2, and Q1. On power-up, the control current through R3, CT, and D2 delays the rise of the output voltage and thus prevents excessive maximum-current transients.

　　Here’s how it works. When VIN is on and Q1 is off, current through R3, CT, and D2 pulls the adjust PIN of the regulator to the reference. This action limits VOUT’s dV/dt to the rate of CT charging through the series resistance, (R3+R1R2/(R1+R2)), and thereby limits IMAX to any desired value using the design equations R3=(VIN–VREF–1)/VOUT, R4=<20 R3, and CT="COUTVOUT/"(IMAX(R3+R1R2/(R1+R2))). For example, given the circuit constants in the figure and assuming COUT="100" µF, dV/dt=2500V/s, and IMAX="0".25A. At the end of the modified power-up sequence, D1 and D2 decouple the dV/dt circuit from the regulator’s feedback network, preventing the coupling of ripple voltages from the input voltage into the output voltage.

　　英文原文地址： http://www.edn.com/article/CA6486029.html

上一页  [1] [2]