SDG #212 Why you don't need 10000+ uF on your voltage regulator output

A common piece of forum advice holds that if a linear regulator’s output sags or overshoots under a sudden load change, the cure is ever larger electrolytics. This video dismantles that assumption, starting with what is actually happening inside a typical series pass regulator: an error amplifier, a voltage reference and a pass element that all have finite bandwidth and slew rate. Because the control loop needs time to react, the behaviour in the first few microseconds after a load step is determined almost entirely by the impedance of the output network, not by the regulator’s DC accuracy.

That distinction is where bulk capacitance stops helping. A 10,000uF electrolytic has significant ESR and ESL plus a self resonance well below the frequencies present in a fast current step, so it simply cannot supply charge quickly enough to fill the initial dip. Adding more of the same part mostly adds cost, board area and inrush current at power-up while doing little for the fast edge. Smaller ceramics placed close to the load, and attention to the loop area of the decoupling path, are what actually flatten the transient.

The measurements use a signal generator driving an electronic load style step, with the UNI-T UTG932 (also available as the UTG962) providing the switching waveform, captured on an oscilloscope so that the initial droop, ringing and recovery time can be compared across different output capacitor combinations. The result is a practical rule for anyone laying out a supply rail: choose capacitance for the frequency range that needs covering rather than the largest value available, and understand that regulator loop bandwidth sets a limit that no amount of bulk capacitance can overcome.