Slide 1
Measure, compare and act
A feedback controller measures a system, compares the result with a target and changes an input. The loop repeats as the system responds.
More to Read
A disturbance can then be corrected without predicting its exact arrival. A room thermostat, for example, responds to a temperature change caused by an opened door.
Slide 2
Correction needs the right sign
Negative feedback acts against a measured difference from the target. If a heater increases output when a room becomes too hot, the loop instead makes the error worse.
More to Read
The correct sign is necessary but not sufficient for stability. Even a controller intended to oppose error can oscillate if its timing and strength are poorly matched to the system.
Slide 3
Strong correction can overshoot
Controller gain sets how strongly the controller reacts to error. Higher gain can reduce some differences from the target but make the response more aggressive.
More to Read
A system with stored energy may continue changing after the controller reduces its input. That inertia can carry the measured quantity beyond the target and provoke repeated corrections.
Slide 4
Delay makes old information dangerous
A controller acts on measurements that may lag behind the current state. Long delays can turn a useful correction into a badly timed one.
More to Read
Transport time, sensor filtering and computation all add delay. Tuning—adjusting how the loop responds—requires considering them together, not only the actuator’s speed.
Slide 5
The loop inherits its measurement limits
A controller cannot distinguish a real change from an error in its sensor unless the design provides other information. Better control begins with a trustworthy signal.
More to Read
Filtering can reduce noise—unwanted signal variation—but also delay the response. This creates a tradeoff between reacting quickly and following variations that are not useful changes.
Sources
- Designing Control Systems
MIT OpenCourseWare