Open Loop vs Closed Loop
Open-loop control commands blindly; closed-loop control measures the result and corrects, buying disturbance rejection and robustness.
Two Ways to Command a Plant
An open-loop controller computes its command purely from the reference and a model of the plant. A toaster on a timer is open loop: it heats for a fixed duration regardless of how brown the bread actually gets. The command never depends on the outcome.
A closed-loop (feedback) controller measures the actual output, subtracts it from the reference to form an error, and acts on that error. A thermostat is closed loop: it keeps heating only until the measured temperature reaches the setpoint.
Why feedback matters
- Disturbance rejection: feedback counters loads the model never anticipated.
- Robustness: it tolerates gain and parameter errors in the plant model.
- Stabilization: feedback can stabilize an inherently unstable plant.
- Accuracy: high loop gain drives steady-state error toward zero.
What feedback costs
Feedback is not free. It can turn a stable plant unstable if designed poorly, it injects sensor noise into the command path, and it demands a sensor at all. Open-loop control avoids these problems and remains the right choice when the plant is well known, stable, and disturbances are small.
A quick comparison
| Model error | Disturbance | Best choice |
|---|---|---|
| small | small | open loop |
| large | large | closed loop |
In practice the two combine. A well-tuned feedforward term (open loop, model-based) handles the predictable bulk of the command while feedback (closed loop) cleans up the residual error. Large fusion machines mix both: precomputed coil-current trajectories provide the feedforward while real-time feedback trims plasma position against unmodeled drifts.
The decision is ultimately about information: closed loop pays for a sensor to earn robustness, and the question is whether the disturbance and uncertainty justify that price.