Glossary term
Negative Feedback
Feedback that subtracts a portion of output from input to reduce error and stabilize behaviour.
Definition
conceptFeedback that subtracts a portion of output from input to reduce error and stabilize behaviour.
Negative feedback measures part of a system output and feeds it back with opposite sign so that the controller acts on the error between reference and output. It is the core mechanism behind regulation, disturbance rejection, accuracy improvement, and many stability trade-offs in control systems and electronics.
Negative feedback compares a desired reference with a measured output and drives the actuator according to the error. If the output rises above the reference, the feedback action tends to reduce the command; if the output falls below the reference, it tends to increase the command. This sign convention is what distinguishes regulation from runaway amplification.
For a linear loop with plant G(s) and feedback path H(s), the closed-loop transfer function is commonly written as:
The denominator shows the basic trade-off. Increasing loop gain can improve tracking and disturbance rejection, but it can also reduce stability margin when delay, phase lag, actuator limits, sampling, or unmodeled dynamics become important.
Engineering use
Negative feedback is used in thermostats, motor drives, aircraft control, process control, voltage regulators, operational amplifiers, robotics, hydraulic actuators, and speed governors. It can reduce sensitivity to component variation, reject disturbances, linearize nonlinear elements over a working range, and improve accuracy without requiring perfect open-loop models.
The feedback signal must be measured with enough bandwidth, resolution, and reliability for the control objective. Sensor noise, quantization, filtering, delay, dead time, and mechanical backlash can all be fed back into the control action. In practice, controller design balances tracking performance, noise amplification, actuator effort, robustness, and safety behavior.
Stability perspective
A negative-feedback loop can become unstable if the phase shift around the loop approaches 180 degrees while loop gain remains high. At that point, the feedback effectively becomes positive at some frequency. Bode plots, Nyquist plots, root locus, gain margin, phase margin, and sensitivity functions are common tools for checking this risk before hardware is exposed to aggressive gains.
Common mistakes
A common mistake is a sign error in wiring, coordinate convention, encoder direction, or software scaling. The loop then reinforces error instead of correcting it. Another mistake is assuming negative feedback always improves behaviour: with saturation, rate limits, integral windup, excessive delay, or noisy sensors, the same loop can oscillate or damage equipment. A careful review checks loop sign, units, sensor polarity, actuator direction, margins, saturation handling, and fail-safe behavior.