Control of Electrical Drives

Introduction

An electrical drive is an electromechanical energy-conversion system in which a power converter, motor, controller, sensors, and load operate together to produce controlled motion. Modern drives are required not merely to rotate a shaft, but to regulate torque, speed, position, energy consumption, vibration, and safety under changing operating conditions.

A drive may be classified as open loop or closed loop. In an open-loop drive, the control command is applied without measuring the controlled output. Such systems are simple and inexpensive but cannot automatically compensate for load disturbances or parameter variations. In a closed-loop drive, measured speed, position, current, or torque is compared with a reference, and the error is used to modify the actuator command.

This chapter develops the principal control structures used in industrial motor drives: cascaded current and speed loops, scalar control, field-oriented control, direct torque control, sensorless estimation, and servo control. It also addresses controller tuning, stability, implementation constraints, and the selection of an appropriate strategy.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Distinguish open-loop and closed-loop electrical-drive control.
2. Explain why current, speed, and position loops are commonly cascaded.
3. Apply scalar V/f control to induction motors and identify its limitations.
4. Describe the principles of field-oriented control and direct torque control.
5. Explain sensorless speed estimation and its low-speed difficulties.
6. Design basic proportional-integral controllers and assess stability.
7. Select a suitable drive architecture for an industrial application.
8. Recognize practical, safety, electromagnetic-compatibility, and implementation constraints.

1. Drive Control as a Feedback Problem

The mechanical dynamics of a motor-load system are commonly represented by

where J is the combined inertia in kg·m², ω_m is mechanical angular speed in rad/s, T_e is electromagnetic torque in N·m, T_L is load torque in N·m, and B is viscous-friction coefficient in N·m·s/rad. The equation assumes a lumped rigid inertia and viscous friction; Coulomb friction, elasticity, backlash, and disturbances may require additional terms.

The control objective is usually to make a measured variable y follow a reference r. The error is

A controller generates a command from this error. In a closed loop, the plant, sensor, controller, and actuator form a feedback system. Negative feedback generally improves disturbance rejection and reduces sensitivity to plant-gain variation, but it can introduce oscillation or instability if loop gain and delay are excessive.

Open-loop control

Examples include fixed-frequency operation of an induction motor, timed stepping-motor excitation, and a predetermined voltage-frequency ramp. Open-loop control is suitable when the load is predictable and exact speed regulation is unimportant. Its principal weaknesses are:

• speed changes with load torque and supply variation;
• acceleration and braking cannot be accurately controlled;
• disturbances are not automatically rejected;
• parameter variation can produce large errors.

Closed-loop control

A closed-loop drive measures one or more variables. Current feedback provides rapid torque limitation and protection; speed feedback improves speed regulation; position feedback enables servo operation. Feedback sensors may be encoders, resolvers, Hall sensors, tachogenerators, current shunts, or voltage sensors.

A useful design principle is to regulate the fastest physical variable in the innermost loop. Electrical current changes much faster than mechanical speed, so current control is normally placed inside the speed loop. Position is slower still and is placed outside the speed loop.

2. Cascaded Current and Speed Control

A typical cascade contains a current controller, a speed controller, and, when required, a position controller. Each outer controller supplies the reference for the inner controller. Limiting the output of an outer controller is essential: a speed controller should not request more current than the motor, inverter, or thermal system can safely provide.

Figure: A cascaded position–speed–current control structure. The current loop is fastest and the position loop is slowest.

Current control

For a separately excited DC motor, electromagnetic torque is approximately

where k_t is the torque constant in N·m/A and i_a is armature current in A, provided the field flux is constant. For an AC machine, the relationship is more complex, but current components can be selected to control torque-producing and flux-producing effects.

A PI controller is widely used:

where K_p is proportional gain, K_i is integral gain, e(t) is tracking error, and u(t) is controller output. In Laplace form, assuming zero initial conditions,

The proportional term gives immediate response; the integral term removes steady-state error for a constant reference or disturbance. Saturation must be accompanied by anti-windup, otherwise the integrator continues accumulating error while the actuator is already at its limit.

Speed control

The speed controller usually produces a torque or current reference. Its bandwidth is selected substantially below the current-loop bandwidth, commonly by a factor of five to ten in a well-separated design. Excessively fast outer loops interact with inner-loop dynamics and delay, producing overshoot or oscillation.

For a simplified mechanical plant with torque command as input,

The simplification neglects electrical dynamics and assumes torque command is accurately followed. It is useful for preliminary speed-loop design, not for final validation of a high-performance drive.

Position control

A position loop compares demanded and measured angular position. It may use a proportional controller or a more elaborate feedforward and feedback structure. The commanded speed must be limited, and acceleration and jerk limits are often imposed to reduce mechanical stress. Position control is especially important in robotics, machine tools, packaging, and semiconductor equipment.

3. Scalar Control of Induction Motors

Scalar control regulates magnitudes such as stator voltage and frequency rather than the instantaneous spatial vectors of flux and current. Its most common form is constant V/f control.

For an induction motor at frequencies above the low-speed compensation region, maintaining approximately constant air-gap flux requires

where V_s is stator-voltage magnitude and f_s is stator electrical frequency. This follows from the approximate transformer relation E_s ≈ 4.44 N_s f_s Φ, where E_s is induced rms voltage, N_s is effective turns, and Φ is air-gap flux. The approximation assumes sinusoidal fundamental quantities and neglects stator-resistance voltage drop.

Below rated frequency, a voltage boost is often added to compensate for stator resistance:

where V_boost compensates low-speed resistive drop and k_f is selected from rated operating data. Above base speed, voltage is limited by the DC bus and flux is weakened as frequency increases. This is the constant-power region, approximately, because torque capability decreases as speed rises.

Scalar control is inexpensive and robust, and is adequate for pumps, fans, compressors, and conveyors with moderate performance requirements. Its disadvantages include poor low-speed torque, slow disturbance response, dependence on motor parameters, and limited control during rapid acceleration or regenerative braking. Slip compensation can improve speed regulation by adding an estimated slip frequency to rotor speed or commanded frequency, but it does not provide the dynamic decoupling of vector control.

4. Field-Oriented Control

Field-oriented control (FOC), also called vector control, transforms three-phase variables into a rotating reference frame. In a suitable synchronous frame, the machine variables become approximately dc quantities in steady state. This permits independent regulation of flux-producing and torque-producing current components.

The Clarke transformation maps phase variables to stationary orthogonal axes. For balanced three-phase currents, one common amplitude-invariant form is

with i_a + i_b + i_c = 0. The Park transformation then rotates the stationary vector through electrical angle θ_e:

The d-axis is commonly aligned with rotor or stator flux. Under rotor-flux orientation in an induction motor, i_d primarily controls flux and i_q primarily controls torque. The separation is exact only under the selected machine model and correct angle estimation; cross-coupling and parameter errors remain in practical systems.

For a permanent-magnet synchronous motor, an often-used torque expression in rotor coordinates is

where p is the number of pole pairs, λ_m is permanent-magnet flux linkage in Wb, L_d and L_q are d- and q-axis inductances in H, and currents are in A. For a surface-mounted machine with L_d ≈ L_q, the reluctance term is negligible and torque is approximately proportional to i_q.

FOC requires rotor electrical angle, obtained from a position sensor or estimator. Current controllers regulate i_d and i_q, while decoupling terms may compensate back-emf and rotational cross-coupling. FOC offers rapid torque response, high efficiency, and accurate speed control, but requires more computation, machine parameters, coordinate transformations, and careful commissioning than scalar control.

5. Direct Torque Control

Direct torque control (DTC) estimates stator flux and electromagnetic torque directly and selects inverter voltage vectors using a switching table or predictive algorithm. The basic stator-flux model is

where λ_s is stator-flux vector in Wb, v_s is stator-voltage vector in V, R_s is stator resistance in Ω, and i_s is stator-current vector in A. The model assumes measured or reconstructed voltage and sufficiently accurate resistance.

Torque is related to the flux and current vectors by

for an amplitude-invariant stationary-axis convention. Alternative scaling conventions produce different numerical coefficients, so the transformation convention must be kept consistent.

DTC can produce very rapid torque response and does not require a modulator in its classical form. However, hysteresis control may cause variable switching frequency, torque ripple, acoustic noise, and sensitivity to stator-resistance error. Modern predictive DTC and constant-switching-frequency variants reduce these limitations.

6. Sensorless Estimation

A sensorless drive estimates rotor position or speed from measured currents and voltages rather than using a mechanical encoder. Benefits include lower cost, improved mechanical ruggedness, reduced wiring, and easier installation. Typical methods include:

• back-emf observers;
• model-reference adaptive systems;
• full-order and reduced-order observers;
• sliding-mode observers;
• extended Kalman filters;
• high-frequency signal injection.

At low speed, back-emf magnitude is small because back emf is approximately proportional to speed. Consequently, voltage-sensor offsets, inverter dead time, resistance uncertainty, and integration drift become relatively large. Permanent-magnet machines may use high-frequency saliency injection at zero or very low speed, but this requires saliency and produces additional acoustic and electrical effects.

An observer should be assessed for convergence, noise sensitivity, computational burden, parameter sensitivity, and behavior during regeneration and rapid transients. A practical product may combine sensorless estimation at medium and high speed with an initial-position method or low-speed sensor for reliable starting.

7. Servo Control and Motion Profiles

A servo drive is a closed-loop drive designed for precise position, speed, or torque control. Its performance is determined not only by bandwidth but also by following error, settling time, overshoot, repeatability, disturbance rejection, and mechanical resonance.

Motion commands should generally include acceleration and deceleration limits. A trapezoidal velocity profile has constant acceleration, constant velocity, and constant deceleration. A jerk-limited profile changes acceleration gradually, reducing excitation of flexible structures. Feedforward of desired velocity and acceleration can reduce tracking error, while feedback maintains robustness.

Mechanical resonance may be represented by a two-inertia model involving motor inertia, load inertia, shaft stiffness, and damping. A controller tuned using only a rigid-body model can excite this resonance. Notch filters, low-pass filters, friction compensation, and torque feedforward may be required, but every filter introduces phase lag and therefore reduces stability margin.

8. Tuning, Stability, and Digital Implementation

Controller tuning begins with an appropriate plant model and explicit requirements. The engineer should identify current, speed, and position-loop bandwidths; actuator limits; sampling frequency; sensor resolution; computation delay; and expected disturbances.

For a first-order plant

where K is plant gain and τ is time constant, a PI controller can be selected by pole-zero cancellation as an initial design: choose the controller zero near the plant pole, so that K_i/K_p ≈ 1/τ. This is only a starting approximation; saturation, delay, unmodeled poles, and noise must be checked by simulation and experiment.

Stability is not equivalent to good performance. Important measures include gain margin, phase margin, bandwidth, sensitivity, complementary sensitivity, overshoot, and settling time. A Bode-plot design should include sensor and computational delay. A nominally stable continuous controller can become poorly damped after discretization.

A digital PI controller may be implemented incrementally. With sampling interval T_s, one simple discretization is

where k indexes samples. The implementation must define output saturation and anti-windup. Common anti-windup methods include conditional integration and back-calculation. The sampling frequency should be sufficiently higher than the intended closed-loop bandwidth, and interrupt jitter should be minimized.

Current measurement requires attention to shunt placement, amplifier bandwidth, common-mode voltage, offset, and isolation. PWM dead time distorts the applied voltage, especially at low speed. DC-bus ripple, quantization, switching noise, thermal drift, and electromagnetic interference affect estimation and control. Protection functions should operate independently enough that a software fault cannot defeat overcurrent, overvoltage, overspeed, insulation, or emergency-stop protection.

9. Selection of a Control Strategy

The correct choice depends on the complete system rather than the motor alone. A high-performance algorithm cannot compensate for inadequate sensors, poor mechanics, insufficient cooling, or an undersized inverter.

10. Illustrative Case Study: Conveyor Servo Retrofit

Background

An illustrative manufacturing conveyor uses a 2.2 kW induction motor. The original system uses fixed-frequency operation and experiences speed reduction when boxes accumulate. The process requires speed regulation within ±1%, controlled acceleration, and safe stopping.

Problem analysis

The motor speed changes because the load torque increases while the supply frequency remains fixed. Open-loop operation cannot reject this disturbance. A scalar drive would improve starting and permit adjustable speed, but accurate regulation during rapid load changes would remain limited.

Approach

A voltage-source inverter with sensorless vector control is selected. The design uses:

1. inner d- and q-axis current loops;
2. an outer speed PI controller;
3. current and torque limits;
4. an acceleration-limited speed reference;
5. thermal protection and DC-bus overvoltage protection;
6. a low-speed fallback strategy for starting and reversal.

The motor parameters are identified from rated data and commissioning tests. Current-loop gains are tuned first with the rotor locked or under controlled conditions. The speed loop is then tuned at a lower bandwidth. Load-step tests are performed in both motoring and regenerative operation.

Outcome

The closed-loop drive maintains speed much more effectively during box accumulation and produces smoother acceleration. The design still requires validation of low-speed starting, estimator convergence, braking energy management, and electromagnetic compatibility. If the conveyor must hold position accurately, an encoder and outer position loop would be preferable.

Learning points

The case demonstrates that control architecture should be selected from performance requirements. It also shows why current limits, mechanical constraints, commissioning procedures, and safety functions are part of drive design rather than afterthoughts.

11. Safety, Governance, and Contemporary Developments

Electrical drives contain hazardous voltages, stored DC-link energy, rotating machinery, and potentially high fault currents. Design should follow applicable machinery, electrical-safety, functional-safety, and electromagnetic-compatibility requirements. Risk assessment should address unexpected restart, loss of feedback, overspeed, regenerative overvoltage, braking failure, and access to energized parts.

Cybersecurity is increasingly relevant because networked drives can be configured or commanded remotely. Authentication, secure firmware, access control, event logging, network segmentation, and controlled update procedures reduce risk. Artificial intelligence can assist with condition monitoring, parameter identification, and fault diagnosis, but safety-critical protection should not depend exclusively on an opaque statistical model. AI-based recommendations require validated data, defined failure responses, and human engineering oversight.

Current developments include wide-bandgap semiconductor inverters, model-predictive control, high-frequency sensorless techniques, integrated motor drives, digital twins, energy-aware motion planning, and cloud-connected condition monitoring. These technologies improve efficiency and observability but also increase switching-speed, thermal, EMC, software-validation, and cybersecurity challenges.

Best-Practice Design Sequence

1. Define torque, speed, position, duty-cycle, overload, and stopping requirements.
2. Model the motor, inverter, load, sensors, and mechanical transmission.
3. Choose open-loop, scalar, vector, DTC, or servo control from required performance.
4. Design the current loop before the speed loop, and the speed loop before the position loop.
5. Include saturation, anti-windup, feedforward, filtering, and fault handling.
6. Validate stability with realistic delay, sampling, parameter variation, and load models.
7. Test progressively: unloaded operation, low-speed operation, rated load, transients, regeneration, and fault cases.
8. Document assumptions, limits, maintenance procedures, and safe states.

Chapter Summary

Electrical-drive control is a feedback problem involving electrical, mechanical, computational, and safety constraints. Open-loop drives are simple but cannot reject disturbances. Closed-loop drives use feedback to regulate current, speed, torque, or position. Cascaded loops place fast current control inside slower speed and position control, enabling effective limiting and disturbance rejection.

Scalar V/f control is economical and suitable for moderate-performance induction-motor applications, whereas field-oriented control provides decoupled flux and torque regulation with superior dynamic behavior. Direct torque control acts directly on estimated flux and torque and can be very rapid, although classical forms may exhibit variable switching frequency and ripple. Sensorless methods reduce hardware but face severe challenges at low speed.

PI controllers, saturation management, anti-windup, sampling, delay, sensor quality, PWM nonlinearity, and mechanical resonance strongly influence practical performance. A successful drive is selected and validated as a complete system, with safety, EMC, cybersecurity, thermal limits, and fault behavior included from the beginning.

Key Terms

Open-loop drive: A drive whose control action does not use feedback of the controlled output.

Closed-loop drive: A drive that compares a measured output with a reference and acts on the error.

Cascade control: Nested control loops in which an outer loop supplies the reference for an inner loop.

Current loop: The fast loop regulating motor current or current components.

Speed loop: A loop regulating mechanical angular speed.

Position loop: A loop regulating shaft or machine position.

Scalar control: Control of magnitudes such as voltage and frequency without direct instantaneous vector regulation.

V/f control: Induction-motor control that approximately maintains the ratio of voltage magnitude to frequency.

Field-oriented control: Vector control using rotating coordinates to regulate flux- and torque-producing current components.

Direct torque control: Control that selects inverter voltage vectors from estimated torque and flux errors.

Sensorless control: Drive control that estimates speed or position without a mechanical motion sensor.

PI controller: A proportional-integral controller containing proportional and integral actions.

Anti-windup: A method preventing integrator accumulation during actuator saturation.

Flux weakening: Operation above base speed in which flux is reduced because voltage is limited.

Slip compensation: Frequency correction intended to offset induction-motor speed slip.

Servo drive: A closed-loop drive designed for accurate torque, speed, or position control.

Bandwidth: The frequency range over which a control system responds effectively.

Phase margin: A frequency-domain measure related to the additional phase lag required for instability.

Regeneration: Conversion of mechanical energy back into electrical energy by the drive.

Dead time: Deliberate switching delay in an inverter leg to prevent shoot-through.

Review Questions

Short Answer Questions

1. What is the principal difference between an open-loop and a closed-loop drive?
2. Why is the current loop normally placed inside the speed loop?
3. State the approximate condition used in constant-V/f induction-motor control.
4. What roles do the d-axis and q-axis currents commonly play in field-oriented control?
5. Why is sensorless control difficult at very low speed?

Long Answer Questions

1. Explain the structure and design rationale of cascaded current, speed, and position loops.
2. Compare scalar control, field-oriented control, and direct torque control in terms of performance, complexity, and limitations.
3. Derive and discuss the mechanical equation of a motor-load system, including the meaning and units of each term.
4. Explain PI-controller tuning, saturation, anti-windup, sampling, and delay in a digital drive.
5. Discuss the major safety, EMC, thermal, and cybersecurity issues in a modern industrial drive.

Analytical / Application Questions

1. A conveyor requires accurate speed regulation during sudden load increases but does not require precise stopping position. Select a control architecture and justify it against scalar control and servo control.
2. A PMSM drive shows oscillatory speed response after a controller upgrade. Identify at least five possible electrical, mechanical, sensing, or digital causes and propose a systematic diagnostic sequence.
3. Design a validation plan for a sensorless induction-motor drive that must operate from zero speed to above base speed, including tests for starting, load disturbance, regeneration, parameter variation, and fault response.

Multiple Choice Questions with Answers

Question 1

Which loop is normally the innermost loop in a cascaded electrical drive?

A. Position loop

B. Speed loop

C. Current loop

D. Energy-management loop

Correct Answer: C. Current loop

Question 2

In constant-V/f control of an induction motor, the main objective below base speed is to maintain approximately constant:

A. Rotor resistance

B. Air-gap flux

C. Mechanical friction

D. Slip power

Correct Answer: B. Air-gap flux

Question 3

In field-oriented control, the q-axis current is primarily associated with which function under appropriate flux orientation?

A. Torque production

B. Temperature measurement

C. DC-link balancing only

D. Mechanical position sensing

Correct Answer: A. Torque production

Question 4

What is the main purpose of the integral term in a PI controller?

A. To increase sensor resolution

B. To remove steady-state error under suitable conditions

C. To eliminate actuator saturation completely

D. To measure rotor position

Correct Answer: B. To remove steady-state error under suitable conditions

Question 5

Which condition creates a major difficulty for back-emf-based sensorless estimation?

A. Very high mechanical speed only

B. Zero or very low speed

C. Large encoder resolution

D. Constant DC-link voltage

Correct Answer: B. Zero or very low speed

Question 6

For the mechanical equation J dω_m/dt = T_e − T_L − Bω_m, the quantity J represents:

A. Motor inductance

B. Combined rotational inertia

C. Torque constant

D. Viscous friction coefficient

Correct Answer: B. Combined rotational inertia

Question 7

A major limitation of classical hysteresis-based direct torque control is:

A. It cannot control torque

B. It requires a mechanical gearbox

C. Variable switching frequency and torque ripple

D. It operates only with DC motors

Correct Answer: C. Variable switching frequency and torque ripple

Question 8

What is the purpose of anti-windup in a PI-controlled drive?

A. To prevent integrator accumulation during saturation

B. To increase motor inertia

C. To remove all measurement noise

D. To raise the DC-bus voltage

Correct Answer: A. To prevent integrator accumulation during saturation

Question 9

Which control strategy is generally most appropriate for high-precision shaft-position regulation?

A. Fixed-frequency open-loop control

B. Basic scalar control only

C. Closed-loop servo control with position feedback

D. Uncompensated slip control

Correct Answer: C. Closed-loop servo control with position feedback

Question 10

Why are outer speed-loop bandwidths normally lower than inner current-loop bandwidths?

A. Mechanical dynamics are generally slower than electrical current dynamics

B. Current sensors cannot measure fast signals

C. Speed has no effect on torque

D. The inverter operates only at low frequency

Correct Answer: A. Mechanical dynamics are generally slower than electrical current dynamics