Turning up feedback gain makes a system respond faster to a setpoint change — but push it too far and the same speed advantage turns into overshoot and oscillation.
A feedback control loop continuously compares a system's actual output against a desired setpoint, computes the error between them, and adjusts the system's input to drive that error toward zero. The interactive step-response diagram above shows the single most important design trade-off in feedback control: increasing controller gain speeds up the response, but past a certain point it also introduces overshoot and oscillation — you can't get arbitrarily fast response for free.
A standard feedback loop has four elements: the summing junction (computes error = setpoint − measured output), the controller (converts that error into a corrective action, scaled by gain), the plant/system (the actual physical process being controlled — a motor, a thermostat-controlled space, a chemical process), and the feedback path (measures the actual output and returns it to the summing junction). This closed loop — output continuously influencing input — is what distinguishes feedback control from simple open-loop control, where the input is set without regard to what the output actually does.
The controller gain scales how aggressively the system responds to a given error — a larger gain produces a larger corrective action for the same error, which drives the output toward the setpoint faster. This is genuinely useful up to a point: faster response means the system settles to a new setpoint more quickly after a disturbance or setpoint change.
Push gain too high, and the system's correction becomes so aggressive that it overshoots the setpoint before the feedback loop has time to register and correct for it — the system then overshoots in the other direction correcting for that, and can settle into oscillation before finally converging (or, in the worst case, diverge into instability if gain is pushed far enough). This trade-off — response speed vs. stability margin — is the central design tension in feedback control, and it's exactly what root-locus and Bode/Nyquist analysis (see the companion explainers) are tools for navigating systematically rather than by trial and error.
Open-loop control sets the input based purely on a predetermined plan, without measuring or reacting to the actual output — a simple timer-based sprinkler system is open-loop. Closed-loop (feedback) control continuously measures the actual output and adjusts the input based on the error between actual and desired output — a thermostat is closed-loop, since it measures actual temperature and adjusts heating/cooling accordingly.
Very low gain avoids oscillation but produces a sluggish system that responds slowly to setpoint changes or disturbances — in many applications (a fast-moving robotic arm, a critical process control loop), a slow response is itself a real problem. Control system design is about finding an appropriate gain (and often more sophisticated controller structures like PID) that balances speed against stability margin for the specific application's requirements, not simply minimizing gain.
A PID (Proportional-Integral-Derivative) controller is a more sophisticated controller than the simple proportional (gain-only) example shown here — it adds an integral term (eliminates steady-state error by accumulating past error over time) and a derivative term (anticipates future error based on its rate of change, improving stability). PID control is the most widely used control structure in industrial practice specifically because pure proportional control, as shown in this simplified demo, has real limitations that the integral and derivative terms address.
Try our STEM Learning Studio
More calculators, simulators, and guides for this discipline.