PID Controller Interview Questions

Q1. What are the three terms of a PID controller and what does each do?

The proportional term produces an output proportional to the current error, the integral term accumulates past error to eliminate steady-state offset, and the derivative term reacts to the rate of change of error to damp overshoot and improve transient response. In a furnace temperature controller, the P term provides immediate corrective heating, the I term eliminates the offset between setpoint and actual temperature, and the D term prevents overshoot when the setpoint changes. Removing any one term degrades a specific aspect of performance — without I the temperature stays a few degrees below setpoint, without D the response overshoots and oscillates.

Follow-up: What happens if you set the derivative gain too high?

Q2. What is integral windup and how do you prevent it?

Integral windup occurs when the actuator saturates — for example a valve fully open — while the integrator continues accumulating error, building up a large integral term that causes a prolonged overshoot once the setpoint is reached. In a pneumatic pressure controller driving a control valve, if the setpoint is changed by a large step, the valve saturates at 100% open while the integrator winds up to a very large value, causing the pressure to overshoot significantly when the setpoint is finally reached. Anti-windup circuits prevent this by freezing or back-calculating the integral whenever the actuator output is saturated, which is implemented in virtually every industrial DCS and PLC.

Follow-up: Describe two specific anti-windup implementation methods.

Q3. How do you tune a PID controller using the Ziegler-Nichols method?

In the closed-loop Ziegler-Nichols method, you set I and D to zero, increase proportional gain until the system oscillates with constant amplitude, record the ultimate gain Ku and ultimate period Tu, then use the Z-N table formulas: Kp = 0.6Ku, Ti = 0.5Tu, Td = 0.125Tu. On a temperature control loop for an injection molding machine, this procedure might give Ku = 8 and Tu = 40 seconds, yielding Kp = 4.8, Ti = 20 s, Td = 5 s as starting tuning values. Ziegler-Nichols tends to give aggressive tuning with 25% overshoot, so most engineers then detune the result by reducing Kp by 20–30% for industrial applications.

Follow-up: What are the main disadvantages of the Ziegler-Nichols tuning method?

Q4. What is the effect of derivative action on noisy sensor signals?

The derivative term amplifies high-frequency noise because differentiation in the s-domain multiplies by s, boosting noise that is already present in the sensor signal in proportion to its frequency. A pressure transmitter with ±0.1% measurement noise feeding a PID controller with high derivative gain produces large, rapid actuator movements that wear out the control valve and cause process instability. The standard solution is to apply a first-order low-pass filter to the derivative term, implemented as Kd·s/(τd·s + 1) rather than pure Kd·s, with the filter time constant τd typically set to 1/10 of the derivative time.

Follow-up: How do you choose the cutoff frequency of the derivative filter?

Q5. What is the difference between position form and velocity form PID?

Position form PID directly calculates the absolute output signal u(t) = Kp·e + Ki·∫e·dt + Kd·de/dt, while velocity form (incremental form) calculates the change in output Δu at each sample, which is then added to the previous output. A PLC controlling a motorized valve uses velocity form PID because it naturally handles actuator limits and bumpless transfer — if the controller is switched from manual to automatic mode, the valve stays at its current position without a sudden jump. Velocity form also avoids integrator windup by construction, since each step only adds a finite increment.

Follow-up: What is bumpless transfer and why is it important in industrial controllers?

Q6. Why is derivative action often omitted in controllers for process industry applications?

Derivative action is omitted in many process industry applications because process variables like temperature, pH, and flow are inherently noisy and slow, and the derivative term amplifies the noise more than it improves the response speed. A pH control loop in a water treatment plant typically uses only PI control because the measurement noise from the pH electrode is significant, and the slow process dynamics mean that the speed improvement from derivative action is not worth the noise amplification and valve wear. For fast, clean processes like servo position control, derivative action is valuable and routinely used.

Follow-up: Under what process conditions is derivative action actually beneficial?

Q7. What is the role of the integral gain in a PID controller?

The integral gain eliminates steady-state error by continuously accumulating the error signal and adding it to the output until the error reaches zero, effectively providing infinite DC gain in the controller. A proportional-only speed controller for a conveyor belt will always run slightly below setpoint because proportional action alone needs a non-zero error to produce output, but adding integral action drives the steady-state error to exactly zero. The tradeoff is that integral action slows the response and requires careful anti-windup management when the actuator saturates.

Follow-up: What type of disturbance can a PI controller reject completely in steady state?

Q8. How does a PID controller behave when proportional gain is set too high?

Excessively high proportional gain causes the closed-loop system to oscillate, or in extreme cases go unstable, because the large corrective action overshoots the setpoint and generates a new large error in the opposite direction. A motor position controller with Kp = 200 Nm/rad when the stable range is below Kp = 50 Nm/rad will exhibit continuous oscillation around the setpoint as the stiff proportional response repeatedly overshoots. The gain at which oscillation begins is the ultimate gain Ku used in Ziegler-Nichols tuning.

Follow-up: How would you diagnose excessive proportional gain from a closed-loop step response test?

Q9. What is set-point weighting in a PID controller?

Set-point weighting applies a coefficient less than 1 to the proportional and derivative terms acting on the setpoint, allowing the feedback loop to remain tightly tuned for disturbance rejection while reducing overshoot on setpoint changes. An industrial boiler drum level controller with set-point weighting factor b = 0.5 applies only half the proportional gain to a setpoint step, cutting the initial overshoot while retaining full disturbance rejection capability. Without set-point weighting, tuning for disturbance rejection and tuning for setpoint tracking are conflicting requirements.

Follow-up: How does set-point weighting affect the disturbance rejection performance of the controller?

Q10. What is a cascade PID control structure and when is it used?

Cascade control uses two PID loops — an outer (primary) loop controlling the main process variable and an inner (secondary) loop controlling a faster intermediate variable — where the outer loop's output is the setpoint for the inner loop. In a steam temperature control system, the outer loop controls steam temperature while the inner loop controls fuel flow rate, which responds ten times faster than the temperature, allowing disturbances in the fuel supply to be corrected before they affect the temperature. The inner loop must be tuned first and must be at least 3–5 times faster than the outer loop for cascade control to provide a significant benefit.

Follow-up: What is the main requirement for the inner loop speed relative to the outer loop in cascade control?

Q11. What is the transfer function of a PID controller in the s-domain?

The ideal PID controller transfer function is C(s) = Kp(1 + 1/(Ti·s) + Td·s), which can be rewritten as Kp·(Ti·Td·s² + Ti·s + 1)/(Ti·s), showing a double zero in the numerator and a pure integrator in the denominator. For Ti = 2 s and Td = 0.5 s, the two zeros are placed at s = -0.29 and s = -3.4, and the controller designer positions these zeros to cancel or bracket the dominant plant poles. The pure integrator in the transfer function is what guarantees zero steady-state error to step disturbances.

Follow-up: Where do the zeros of the PID transfer function need to be placed relative to the dominant plant poles?

Q12. How do you select initial PID gains for a new process without using Ziegler-Nichols?

The IMC (Internal Model Control) tuning method uses the plant step response to identify the gain K, time constant τ, and dead time θ, then sets Kp = τ/(K(θ + λ)), Ti = τ, and Td = θ/2, where λ is the desired closed-loop time constant chosen by the engineer. For a first-order-plus-dead-time process with K = 2, τ = 30 s, and θ = 5 s, choosing λ = 5 s gives Kp = 1.5, Ti = 30 s, Td = 2.5 s — a smooth, disturbance-rejecting response. IMC tuning is preferred over Ziegler-Nichols in process industries because it explicitly includes a robustness tuning parameter λ that the engineer can adjust.

Follow-up: How does increasing the IMC filter time constant λ affect the response speed and robustness?

Q13. What is derivative kick and how is it avoided?

Derivative kick is the large, instantaneous spike in controller output that occurs when a step change in setpoint hits the derivative term, because the derivative of a step is theoretically infinite. In a motor speed controller, changing the setpoint suddenly from 1000 to 1500 RPM causes the derivative term to produce a huge voltage spike that can damage the drive electronics or cause mechanical shock. The standard solution is to apply derivative action only to the measurement (process variable), not to the error, so a step in setpoint does not affect the derivative term at all.

Follow-up: What is the name for applying the derivative term only to the process variable rather than the error?

Q14. What is a parallel PID versus a series PID structure?

In parallel PID, the three terms are computed independently and summed: u = Kp·e + Ki·∫e·dt + Kd·de/dt, with each gain independent of the others. In series (interacting) PID, the proportional and derivative modes interact: C(s) = Kp·(1 + 1/(Ti·s))·(1 + Td·s), which is the structure implemented by many older pneumatic and electronic analog controllers. When converting tuning parameters between the two forms, the effective proportional gain changes, which is why PID parameters from an older analog controller cannot be directly entered into a modern digital DCS without conversion.

Follow-up: How do you convert series PID parameters to parallel PID parameters?

Q15. Why is pure derivative control never used alone?

Pure derivative control produces zero output for constant errors and infinite output for instantaneous changes, making it physically unrealizable and useless for eliminating steady-state error. A robotic joint controlled only by derivative action would ignore a constant position error entirely, since the derivative of a constant is zero, resulting in the joint never reaching its target. Derivative action is always combined with proportional or integral action to provide the steady-state correction that derivative alone cannot supply.

Follow-up: What is the minimum combination of PID terms needed to achieve zero steady-state error to a step input?