PID controller tuning covers how proportional, integral, and derivative actions shape the response of a feedback control system. College engineering students use this reference to connect tuning parameters with rise time, overshoot, settling time, offset, and stability. The cheat sheet emphasizes practical formulas and rules that help students move from a plant model or test data to reasonable controller gains.
It is designed as a clean printable reference for controls labs, design projects, and exam review.
The core PID law combines present error, accumulated error, and predicted error trend to compute the control input. Classical tuning methods such as Ziegler-Nichols, Cohen-Coon, and relay autotuning provide starting gains, not final guaranteed designs. Robust tuning also checks gain margin, phase margin, actuator limits, noise sensitivity, and integral windup.
Good PID design balances performance with stability, disturbance rejection, and safe operation under uncertainty.
Key Facts
- The parallel PID control law is u(t) = Kp e(t) + Ki integral e(t) dt + Kd de(t)/dt.
- The ideal PID form is u(t) = Kc[e(t) + 1/Ti integral e(t) dt + Td de(t)/dt], where Ki = Kc/Ti and Kd = Kc Td.
- Increasing Kp usually decreases rise time and steady-state error, but too much Kp increases overshoot and can cause oscillation.
- Increasing Ki removes steady-state offset for constant disturbances, but too much Ki increases overshoot, settling time, and windup risk.
- Increasing Kd adds damping and can reduce overshoot, but derivative action amplifies measurement noise unless filtered.
- For Ziegler-Nichols closed-loop PID tuning, Kp = 0.6 Ku, Ti = 0.5 Pu, and Td = 0.125 Pu, where Ku is ultimate gain and Pu is ultimate period.
- A common filtered derivative term is Kd s/(1 + Td s/N), where N is often chosen between 5 and 20.
- A practical robustness target is phase margin about 45 degrees to 60 degrees and gain margin greater than about 6 dB for many industrial loops.
Vocabulary
- PID controller
- A feedback controller that combines proportional, integral, and derivative actions to reduce the error between a setpoint and a measured output.
- Proportional gain
- The gain Kp that multiplies the current error and gives an immediate control response.
- Integral action
- The control action that accumulates error over time to eliminate steady-state offset.
- Derivative action
- The control action that responds to the rate of change of error and can add damping to the closed-loop response.
- Ultimate gain
- The gain Ku at which a closed-loop system with proportional-only control produces sustained oscillations.
- Integral windup
- A condition where the integral term grows too large during actuator saturation and causes overshoot or slow recovery.
Common Mistakes to Avoid
- Using Ziegler-Nichols gains as final gains without testing robustness is wrong because the method often gives aggressive tuning with high overshoot.
- Increasing all three gains at once is wrong because it hides which term caused oscillation, noise amplification, or slow settling.
- Ignoring actuator saturation is wrong because the controller output may demand an impossible input, leading to integral windup and poor recovery.
- Applying derivative action directly to a noisy measurement is wrong because differentiation amplifies high-frequency noise and can make the actuator chatter.
- Confusing Ki with Ti is wrong because Ki = Kc/Ti, so increasing Ti actually weakens integral action in the ideal PID form.
Practice Questions
- 1 A Ziegler-Nichols closed-loop test gives Ku = 8 and Pu = 12 s. Find Kp, Ti, and Td for PID tuning.
- 2 A PID controller is written as u(t) = Kc[e(t) + 1/Ti integral e(t) dt + Td de(t)/dt] with Kc = 4, Ti = 10 s, and Td = 0.5 s. Find Ki and Kd in parallel form.
- 3 A temperature loop has a steady-state offset after a setpoint change using proportional-only control. Which PID term should be added or increased, and what new risk should be checked?
- 4 Explain why a controller with very fast rise time may still be a poor design if it has low phase margin, actuator saturation, or strong measurement noise.