Engineering · Learn · Drives
PID Control in VFDs
How does a drive hold a pressure, flow or speed steady while the load keeps changing?
Most modern VFDs have a PID controller built in. It compares a setpoint with a measured value from a sensor and keeps adjusting the output frequency until the difference, the error, is gone. This page shows what each of the P, I and D actions does, lets you tune a simulated loop and watch the result, and gives a practical tuning routine. Parameter names and ranges differ by brand, so the drive manual is the final word.
Two different things get called “PID in a VFD”. In process PID (fans, pumps, compressors, winders) the drive’s PID block takes a sensor signal such as pressure, flow, temperature or tension, and its output becomes the frequency command. In a vector or servo drive the speed (and position) loops inside the drive are PI or PID loops fed by an encoder; they are set up during auto-tune and commissioning. The tuning ideas below apply to both.
Output = Kp × error + Ki × (sum of error over time) − Kd × (rate of change of the measured value). Many drives show integral action as a time (Ti = Kp ÷ Ki, in seconds) and derivative action as a time (Td = Kd ÷ Kp). Charts here come from the simulator in the next section.
A simplified model for learning: a process with two lags and a short delay, such as a fan or pump with a duct or tank, at 40% of range. The setpoint steps to 50% at 0.5 s and a load disturbance of −5% hits at 6 s. The output (VFD frequency command) is limited to 0 to 100% and the integral term stops growing when the output is saturated (anti-windup). Derivative acts on the measured value so that a setpoint change gives no sudden kick. Real loops are slower, noisier and different, so tune the real machine from its manual.
| Method | How it works | Good for | Watch out for |
|---|
Ziegler–Nichols calculator
| Controller | Kp | Ti (s) | Td (s) | Ki = Kp ÷ Ti | Kd = Kp × Td |
|---|
Method: set I and D to zero, raise Kp until the loop oscillates steadily (no growth, no decay). That Kp is the ultimate gain Ku and the oscillation period is Tu. The classic settings are aggressive and give noticeable overshoot, so many people use the gentler “no overshoot” row as a start. On a real machine, do this only where sustained oscillation is safe; otherwise use the step-response method or manual tuning.
Fix a problem
| Parameter | What it does | Typical setting or range |
|---|
Names differ: gain may appear as P-gain or proportional band (gain = 100 ÷ band), integral as Ti or Ki, derivative as Td or Kd. Starting values for a slow pressure loop with a 4–20 mA sensor: proportional band 50 to 150% (gain about 0.7 to 2), integral time 10 to 30 s, derivative off or 0 to 5 s, input filter 0.5 to 2 s. Treat them as conservative starting points.
| Application | Controlled variable | Typical feedback | Notes |
|---|
Settings are typical starting points, not rules. Every drive brand and process behaves differently, so follow the drive manual and your site safety rules, and tune with the machine in a safe state.