control//feedback control//control loop//saturation

2026-10-01

Saturation is the point where an actuator cannot give more: a valve fully open, a heater at full power, a motor at its current limit. Every real actuator has one, and a loop tuned on paper as if the actuator were unlimited behaves differently, often badly, the moment a large error asks for more than the actuator can deliver.

Inside its limits a loop is linear and its response scales: a step twice as large gives a swing twice as large, with the same shape. At the limit the loop is broken open for a while. The controller keeps computing, the actuator stays pinned, the output moves at the fastest rate the plant allows, and nothing the controller computes during that time reaches the plant. That is why a loop can be fast and well damped for small steps and sluggish or overshooting for large ones.

Saturation sets the speed limit of the loop, not the tuning. If a heater at full power warms an oven by 5 degrees a minute, no controller brings it up faster, and a controller that tries only adds overshoot. Sizing the actuator is part of the control design and comes before the controller.

A controller with memory misbehaves at the limit: its integral keeps accumulating an error the actuator cannot act on, and the loop overshoots when the error finally reverses. That is integral windup, and every industrial PID has protection against it.

Rate limits are a second kind. A valve that takes ten seconds to travel end to end limits how fast the actuation can change, which behaves like an extra lag at large amplitudes; motors have slew limits in the same way.

The word has a second meaning in dynamics, a growth that levels off on its own; the two are told apart in feedback loop.