PID Control for Speed

PID Control for Speed: Output Frequency and Actual Shaft Speed Are Not the Same Thing
Speed control is fundamental in conveyors, mixers, extruders, winders, pumps, fans, machine tools, cranes and servo systems.
A simplified closed-loop structure is:
Speed setpoint → Actual speed → Speed error → PI/PID → Torque demand → Motor → New speed
But a modern variable-frequency drive - VFD usually does much more than increase or decrease output frequency according to one PID calculation.
In vector-control and servo architectures, the speed loop normally sits outside faster torque and current-control loops.
The motor speed regulator normally does not drive the motor phases directly. It converts speed error into a torque or current demand, while faster inner motor-control loops produce the required electromagnetic torque.
Where Does the Speed Reference Come From?
A drive can receive its speed command from:
an operator panel,
a potentiometer,
a PLC,
a fieldbus network,
preset speeds,
or another process controller.
For example, a PLC may command 1200 rpm.
The drive then compares that reference with actual or estimated motor speed and uses its internal speed regulator to develop the torque needed to follow the command.
Motor Frequency Is Not Mechanical Speed
For an induction motor, synchronous speed is approximately:
Ns = 120 × f / P
where:
Ns is synchronous speed,
f is electrical frequency,
P is the number of motor poles.
A four-pole motor therefore has a synchronous speed of 1500 rpm at 50 Hz.
But the rotor of a normal induction motor under motoring load runs slightly below synchronous speed.
The difference is called slip.
This is why controlling frequency alone does not guarantee identical shaft rpm at every load.
A 50 Hz command does not mean the shaft is automatically running at exactly 1500 rpm. Pole count, slip, load and motor-control method all affect actual mechanical speed.
What Is Induction-Motor Slip?
An induction motor needs a speed difference between the rotating stator field and the rotor in order to develop torque.
As load torque changes, this slip also changes.
In a simple open-loop V/f system, increased load can therefore cause some shaft-speed droop.
A drive may use slip compensation to estimate this behaviour and modify its internal command so that static speed regulation improves.
Slip compensation can be very useful, but it is not the same thing as directly measuring shaft speed with an encoder.
Scalar V/f Control Versus Closed-Loop Speed Control
In basic V/f or scalar control, the drive primarily controls motor voltage and electrical frequency according to a defined relationship.
This is sufficient for many:
fans,
pumps,
simple conveyors,
general machinery.
But where load changes must produce very little speed error, the application may require:
sensorless vector control,
closed-loop vector control,
encoder feedback.
Not every VFD application needs an encoder, but not every VFD control mode provides the same speed-regulation performance.
How Does Sensorless Vector Control Know the Speed?
Sensorless vector control estimates motor flux, torque and speed from electrical measurements and a motor model rather than using a physical shaft encoder.
Compared with basic V/f control, it can provide much stronger:
load-disturbance response,
low-speed torque,
speed regulation.
However, the motor model becomes increasingly important at very low speed.
Applications requiring very tight low-speed regulation, near-zero-speed operation or demanding torque performance may benefit from actual encoder feedback.
Encoder Feedback for Speed Control
An encoder can measure actual motor or machine speed directly.
A typical architecture is:
Speed SP → Speed regulator → Torque/current demand → Motor → Encoder → Actual speed
Encoder feedback can be especially valuable for:
very low speed,
tight speed regulation,
fast disturbance rejection,
high-performance torque control.
But encoder installation introduces its own engineering requirements:
correct direction,
correct PPR/resolution,
shielding,
grounding,
mechanical alignment.
Motor-Side Speed May Not Be the Machine Variable You Really Care About
Measuring 1500 rpm at the motor shaft does not prove that a conveyor belt, roller or production material is moving at the correct linear speed.
Between motor and product there may be:
a gearbox,
a belt drive,
a chain,
a roller,
mechanical slip.
If production quality depends on actual belt surface speed, a feedback device on the driven roller may be more meaningful than a motor-mounted encoder.
The best feedback signal is the one that represents the real controlled mechanical variable, not merely the easiest rotating shaft to measure.
How the Speed Loop Connects to Torque and Current
If motor speed falls under load, the controller cannot create rpm directly.
It must generate more torque.
A simplified control chain is therefore:
Speed error → Speed regulator → Torque reference → Current controller → Motor torque
This links speed control directly with torque control and current control.
Servo systems use the same layered principle, with position outside speed and speed outside current.
This is why position control and speed control are naturally connected.
The Speed Regulator Is Often PI Rather Than Full PID
Motor-drive speed loops do not always expose three classic Kp-Ki-Kd parameters.
Many servo and VFD systems use:
speed-loop proportional gain
and
speed-loop integral time
as the primary tuning parameters.
In other words, the practical speed regulator is frequently a PI controller.
Derivative is not mandatory simply because the control objective is called “PID speed control”.
The general P, I and D principles are discussed in P, I and D components.
Derivative Does Not Predict Future Speed
A common simplified explanation says that D predicts future motor speed.
That is not technically accurate.
Derivative responds to the rate of change of error or measurement.
Speed feedback can contain:
encoder quantization,
speed-estimation ripple,
mechanical vibration,
electrical noise.
D can amplify these high-frequency components.
Modern drives already obtain much of their damping from the internal torque/current and speed-control structure, so a well-tuned PI speed regulator is common.
What Happens When Load Torque Suddenly Increases?
Imagine material is suddenly loaded onto a conveyor.
The mechanical load torque rises.
Motor speed begins to fall.
The speed loop detects increased speed error and commands more torque/current.
If the motor and drive have sufficient reserve, actual speed returns toward the setpoint.
This ability is called load-disturbance rejection.
A useful speed-control test therefore does not only ask:
“Does the unloaded motor reach the setpoint?”
It also asks:
“How much speed droop occurs when the real load changes?”
PID Cannot Create Torque Beyond the Physical Limit
If the motor or drive has already reached maximum permitted current or torque, the speed regulator has no additional physical authority.
Speed can continue falling even though the controller demands maximum output.
Possible causes include:
undersized motor,
drive current limit,
excessive mechanical load,
machine binding,
excessive acceleration demand.
A larger speed error can produce a larger controller demand, but it cannot make the motor deliver unlimited torque. Current and torque limits are hard physical control boundaries.
This distinction is fundamental to PID troubleshooting.
Acceleration Ramp Is Not the Speed PID
A drive parameter such as:
0 to 50 Hz in 10 seconds
defines a reference ramp.
It is not a speed-loop gain.
The sequence is more like:
Raw speed reference → Acceleration/deceleration ramp → Speed regulator → Torque demand
The ramp determines how quickly the requested speed is allowed to change.
The speed regulator attempts to make actual motor speed follow that ramped reference.
What If Acceleration Is Too Fast?
A high-inertia load requires substantial torque to accelerate quickly.
If acceleration time is too short, the drive may reach:
current limit,
torque limit,
motor thermal limit.
Actual speed then falls behind the reference ramp.
Increasing speed-loop gain cannot eliminate the torque required by the physics of the load.
Load inertia, available peak torque and desired acceleration must be compatible.
What If Deceleration Is Too Fast?
A rotating load contains kinetic energy.
When the drive commands rapid deceleration, the motor can operate regeneratively and return energy to the drive's DC bus.
If that energy cannot be dissipated or returned to the supply, DC-bus voltage can rise enough to produce an overvoltage fault.
Depending on the system, solutions may include:
longer deceleration time,
a braking resistor,
a braking unit,
a regenerative drive.
A deceleration overvoltage fault is not primarily a speed-PID tuning problem.
Speed Feedback Filtering Has a Cost
Filtering can reduce:
encoder ripple,
estimated-speed noise,
small high-frequency disturbances.
But every filter introduces delay.
Too much filtering can make the speed loop respond late and reduce stability margin.
This is the motor-control version of the issues discussed in time delay and system dynamics.
Smoother feedback is not automatically better feedback if the filter makes it too late.
Motor Identification and Autotune
Sensorless vector and advanced motor-control algorithms depend on an accurate motor model.
Correct commissioning data includes:
rated voltage,
rated current,
rated frequency,
rated speed,
rated power.
Many drives can also perform static or rotating motor identification.
This allows the drive to estimate parameters needed by the internal flux, torque and speed model.
Incorrect motor nameplate data cannot be fixed by increasing speed-loop gain.
Base Speed and Field Weakening
Below base speed, a motor-drive system can often operate in a region where relatively high torque is available.
Above the voltage-limited base-speed region, the drive may enter field weakening.
Motor flux is reduced because additional voltage is no longer available in the same proportion.
Available torque therefore normally decreases as speed rises farther into this region, while operation approaches a constant-power characteristic.
The fact that a motor can be commanded above 50 or 60 Hz does not mean it can deliver full rated low-speed torque at every higher speed.
Mechanical motor limits must also be checked before overspeed operation.
Low-Speed Operation Can Become a Thermal Problem
A standard motor with a shaft-mounted cooling fan loses cooling airflow when motor speed falls.
If the motor produces high torque continuously at low rpm:
electrical loss can remain high
while
mechanical fan cooling becomes weak.
Possible solutions include:
forced ventilation,
motor thermal sensors,
drive thermal modelling,
a motor designed for inverter duty and low-speed torque.
Accurate speed regulation does not prove that the motor is thermally safe.
What Is Speed Droop?
In some mechanically coupled multi-motor systems, forcing every drive to regulate exactly the same speed can cause the motors to fight each other for load.
A controlled speed droop allows speed reference to shift slightly with torque so that mechanical load can be shared more naturally.
This is one example where:
zero speed error at every instant
is not necessarily the best system-level objective.
Drive Process PID and Motor Speed PI Are Different Controllers
This distinction is extremely important.
A modern VFD may contain an internal motor speed regulator.
The same drive may also offer a separate process PID.
For example, pressure control may use:
Pressure SP → Pressure transmitter → Process PID → Motor speed reference → Internal speed loop → Motor
The process PID controls pressure.
The internal motor regulator controls motor speed.
The VFD process PID and the motor's internal speed regulator are not the same PID. One controls a process variable such as pressure or flow; the other makes the motor follow its speed command.
For the process-control architecture, see PID control with a VFD.
Conveyor Speed Control
For a conveyor, production may depend on linear belt speed rather than motor rpm.
For a driven roller:
v = π × D × n / 60
where:
v is surface speed,
D is roller diameter,
n is roller rpm.
Gear ratio, actual roller diameter and belt slip can all introduce error.
If product velocity is critical, direct line-speed feedback may be more useful than motor-speed feedback.
Winders Are Not Constant-RPM Applications
In a winding system, roll diameter changes continuously.
If rpm remains constant while diameter grows, web linear speed increases.
Therefore a winder may need coordinated control of:
line speed,
roll diameter,
torque,
web tension.
This is where torque control and speed control become part of one larger motion system.
Fans and Pumps Often Use Speed as an Actuator, Not as the Final Controlled Variable
In many fan and pump systems, the real target is:
pressure,
flow,
level,
temperature.
The process controller changes motor speed to achieve that process target.
For example:
Pressure error → Process PID → VFD speed reference
while the drive's own motor-control structure handles the speed and torque internally.
This connects motor speed control with:
Mechanical Resonance Can Limit Speed-Loop Gain
A drive may have enough computational bandwidth to support high gain, while the mechanical machine cannot.
Flexible components such as:
couplings,
gearboxes,
belts,
long shafts
can introduce resonance.
Possible symptoms include:
audible motor noise,
speed hunting,
torque oscillation,
current oscillation,
mechanical vibration.
Solutions can involve:
lower loop bandwidth,
torque filtering,
notch filters,
mechanical improvement.
The related physical problem is covered in PID vibration control.
What Can a PLC Do?
A PLC can send motor speed references by:
analog output,
fieldbus,
Ethernet-based drive communication.
It can also manage:
start and stop,
speed limits,
machine sequencing,
process PID,
interlocks,
supervisory ramps.
But the fastest current, flux and torque-control functions normally remain inside the drive or servo amplifier.
This division of responsibility is important in PID control with PLCs.
Speed Control Is a Real-Time Function
High-performance speed, torque and current regulators depend on predictable execution timing.
Excessive feedback delay or timing jitter reduces the achievable control bandwidth.
This becomes especially important in:
servo systems,
synchronized axes,
fast machine motion.
The timing issues are discussed more generally in real-time PID control.
What Should Be Recorded During Commissioning?
A useful drive trace can include:
Speed reference
Actual or estimated speed
Speed error
Torque reference
Actual torque
Motor current
Output frequency
DC-bus voltage
Current / torque limit state
Encoder feedback
A normal SCADA trend is useful for slower operating behaviour, but fast speed-loop tuning is often better diagnosed with the drive's high-speed trace.
Looking only at rpm can hide the real cause of a speed-control problem.
Common Speed-Control Problems
Motor slows when load is applied: inspect control mode, slip compensation, motor capacity and torque/current limit.
Speed hunts around the reference: check speed-loop gain, integral time, mechanical resonance, feedback filtering and encoder signal quality.
Motor cannot follow the acceleration ramp: inspect load inertia, available torque and current limiting.
Low-speed operation is rough: verify motor data and consider vector control or encoder feedback instead of basic scalar control.
Motor rpm is correct but conveyor line speed is wrong: inspect gearbox ratio, roller diameter, belt slip and feedback location.
The drive trips on overvoltage during deceleration: investigate regenerative energy and braking capacity before changing speed-loop gains.
Torque falls above base speed: field-weakening operation and motor-drive limits should be reviewed.
The process PID looks unstable while motor speed tracks correctly: separate the process loop from the internal motor speed regulator during troubleshooting.
Actual speed signal is noisy: inspect encoder PPR, wiring, grounding, shielding, filtering and mechanical mounting.
The drive is continuously in current limit: check machine load and motor sizing before increasing controller gain.
Speed-Control Commissioning Check
1. Is the controlled variable motor rpm, load rpm or true line speed?
2. Are motor nameplate data entered correctly?
3. Is the correct V/f, sensorless-vector or encoder-vector mode selected?
4. If an encoder is used, are direction and PPR/resolution correct?
5. Does the feedback represent the real mechanical variable?
6. Are speed-loop PI parameters appropriate for motor and load inertia?
7. Are acceleration and deceleration ramps physically realistic?
8. Does the drive repeatedly hit torque or current limit?
9. Is braking or regenerative capacity adequate for deceleration?
10. Is motor cooling adequate at low speed?
11. Is sufficient torque available in the field-weakening region?
12. Is speed filtering adding excessive delay?
13. Is mechanical resonance limiting loop bandwidth?
14. Are process PID and internal speed regulator clearly separated?
15. Has tuning been verified at realistic minimum and maximum load?
16. Have speed, torque, current and DC-bus trends been reviewed together?
Related PID, VFD and Motor-Control Topics
PID and motor-control fundamentals: What Is PID? - P, I and D Components - Variable-Frequency Drives - Encoders
Inner motor-control layers: PID Torque Control - PID Current Control - PID Frequency Control - PID Position Control
VFD and process PID: PID with VFDs - PID Flow Control - PID Pressure Control - PID Level Control
Tuning and dynamics: PID Gain and Tuning - Time Delay and System Dynamics - PID Vibration Control - PID Troubleshooting
Control infrastructure: PID with PLCs - PID with SCADA - Real-Time PID - PID and Other Control Strategies
Reliable motor speed control depends on much more than one speed-loop gain. Motor identification, control mode, feedback location, load inertia, current and torque capacity, acceleration and deceleration profiles, braking energy, mechanical resonance, field weakening and motor thermal limits all determine the final machine performance.