Side-by-side comparison
| Parameter | P | I |
|---|---|---|
| Action on error | Proportional to current error | |
| Steady-state error | Reduces but does not eliminate | |
| Speed of response | Moderate improvement | |
| Stability impact | Reduces stability as Kp rises | |
| Noise sensitivity | Low | |
| Typical Kp range (DC motor) | 1–10 | |
| Transfer function | Kp | |
| Bode plot effect | Shifts magnitude up uniformly | |
| Real IC example | Op-amp inverting amplifier |
Key differences
P control gives output = Kp × e(t); it always leaves a steady-state error unless the plant already contains an integrator. I control integrates error over time, which eliminates offset but adds a 90° phase lag, reducing phase margin. D control reacts to how fast the error is changing — it adds phase lead (up to 90°) and damps oscillations, but it should never be used alone and must always be filtered (typically a first-order filter with τ = Kd/N, N = 5–20) to limit noise amplification.
When to use P
Use P-only control when a small steady-state offset is acceptable and simplicity matters — a basic motor speed regulator where ±5% error is tolerable needs only a proportional gain stage.
When to use I
Use I action (PI or PID) when zero steady-state error is mandatory — a process temperature controller (e.g., Eurotherm 3216) in a furnace must eliminate offset entirely to meet product specifications.
Recommendation
For most student lab experiments and placement questions, choose PID — it covers all three deficiencies. Start with Ziegler-Nichols tuning: find Ku and Tu at the stability boundary, then set Kp = 0.6Ku, Ti = 0.5Tu, Td = 0.125Tu.
Exam tip: GATE frequently asks which control action eliminates steady-state error for a step input — the answer is integral (I) action, and you should also state that it reduces phase margin.
Interview tip: Interviewers expect you to explain what happens to system stability as Ki increases and to describe one practical method of filtering the derivative term to reduce noise amplification.