Glossary term
Closed-Loop Control
Control action in which measured output is fed back, compared with a reference, and used to adjust the input to reduce error or shape system behaviour.
Definition
conceptControl action in which measured output is fed back, compared with a reference, and used to adjust the input to reduce error or shape system behaviour.
Closed-loop control uses feedback to make a system respond despite disturbances, model uncertainty, load changes, and component variation. Its performance depends on sensing, actuation, controller design, loop dynamics, stability margin, delay, saturation, and noise.
Closed-loop control measures what a system is doing and uses that information to adjust the command. A reference r is compared with measured output y to form an error e. The controller computes an input u that drives the plant through an actuator. Sensors, signal conditioning, computation, and feedback path all become part of the control loop.
Engineering role
Closed-loop control is used when open-loop commands are not accurate enough because of disturbances, changing loads, uncertain models, wear, friction, temperature variation, or process drift. Examples include motor speed control, robot joints, aircraft autopilots, furnace temperature control, chemical processes, power converters, servo valves, and active suspension.
Performance goals
Design goals may include tracking accuracy, disturbance rejection, fast settling, limited overshoot, low steady-state error, robustness, low actuator effort, and noise rejection. These goals often conflict. Increasing loop gain can reduce error but worsen noise sensitivity, saturation, or stability margin. Adding integral action can remove steady-state error but introduce windup if actuator limits are ignored.
Stability and dynamics
A closed loop can oscillate or become unstable if delay, phase lag, high gain, resonance, or unmodelled dynamics dominate. Frequency-response tools such as Bode plots, phase margin, gain margin, and Nyquist plots help assess robustness. Time-domain tests such as step response and disturbance response reveal overshoot, settling time, rise time, and saturation behaviour.
Implementation
Real loops include sampling, quantization, sensor noise, actuator limits, computation delay, communication jitter, filtering, and fail-safe logic. The control design should specify sampling rate, measurement bandwidth, units, sign convention, scaling, anti-windup, limits, and safe behaviour on sensor or actuator fault.
Common mistakes
Common mistakes include closing a loop around the wrong variable, reversing feedback sign, ignoring actuator saturation, filtering so heavily that phase lag destabilizes the loop, and tuning only at one operating point. Another error is judging performance from simulation without testing disturbances, sensor noise, startup, shutdown, and fault cases.