Electrical Drives & Control

Chapter 4
Power Electronic Converters and Modulation

Bachelors Program in Electrical Engineering
Electrical Drives & Control   •   Bachelor

Power Electronic Converters and Modulation

Introduction

Electrical drives convert electrical energy into controlled mechanical motion. The power electronic converter is the interface that determines how much voltage, current, frequency, and power reach the motor. Modern drives use semiconductor switches operated at high speed to synthesize waveforms from fixed or slowly varying electrical sources. Converter choice and modulation strategy strongly influence torque response, efficiency, electromagnetic interference, acoustic noise, harmonic distortion, and motor insulation stress.

This chapter examines semiconductor switching devices, rectifiers, dc–dc choppers, voltage-source and current-source inverters, matrix converters, multilevel converters, pulse-width modulation (PWM), space-vector modulation (SVM), filtering, protection, and harmonic performance.

Learning Objectives

After studying this chapter, a student should be able to:

  1. Explain the operating roles and limitations of power semiconductor switches.
  2. Analyze the principal switching states of rectifiers, choppers, and inverters.
  3. Calculate ideal converter output quantities under stated assumptions.
  4. Explain carrier-based PWM and space-vector modulation.
  5. Compare two-level, multilevel, matrix, voltage-source, and current-source converters.
  6. Select filters and protection measures for practical drives.
  7. Interpret harmonic distortion, switching loss, and electromagnetic compatibility requirements.

1. Power Semiconductor Switching Devices

A power converter is built from controllable or naturally commutated semiconductor devices. The ideal switch has zero voltage drop when on, zero current when off, and instantaneous transitions. Real devices have conduction loss, switching loss, finite voltage and current ratings, parasitic inductance and capacitance, and thermal limits.

1.1 Principal devices

DeviceControl characteristicTypical strengthCommon drive use
Power diodeUncontrolled; conducts when forward biasedHigh current, simple rectificationRectifiers and freewheeling paths
Thyristor (SCR)Turn-on controlled; turn-off normally by circuit commutationHigh voltage and currentLine-commutated converters
MOSFETVoltage-controlled, fast switchingLow-to-medium voltage, high frequencyLow-voltage choppers and inverters
IGBTVoltage-controlled gate; moderate switching speedMedium/high voltage drivesIndustrial motor inverters
IGCTGate-controlled turn-off thyristorVery high power and low conduction lossMedium-voltage converters
SiC MOSFET / GaN deviceWide-bandgap, fast switchingHigh frequency, high temperature, low loss potentialAdvanced compact drives and converters

An IGBT is commonly represented by a controllable switch with an antiparallel diode. The diode is essential in motor drives because inductive load current cannot change instantaneously. During a switching transition, current may transfer between the transistor and the diode.

1.2 Conduction and switching losses

For a device approximated by a constant voltage drop Von, its conduction loss is

Pcond ≈ Von Iavg

where Iavg is the average device current during conduction. If the device is modeled by v = V0 + r i, then

Pcond = V0 Iavg + r Irms2

Here V0 is the threshold-like voltage, r is the dynamic resistance in ohms, and Irms is the device rms current. These expressions are approximations and apply over the operating range for which the device model is valid.

A simplified switching-loss estimate is

Psw ≈ fs (Eon + Eoff)

where fs is switching frequency in hertz and Eon, Eoff are turn-on and turn-off energies in joules at specified voltage and current. In practice, these energies depend on temperature, gate resistance, load current, dc-link voltage, and commutation conditions.

The total semiconductor loss is approximately

Ploss = Pcond + Psw + Pother

where Pother includes gate-drive, reverse-recovery, leakage, and auxiliary losses. Thermal design must keep junction temperature below its rated limit:

Tj = Ta + Ploss Rθja

for a simplified junction-to-ambient model. Tj and Ta are junction and ambient temperatures in °C, and Rθja is thermal resistance in °C/W. A real design may use separate junction-to-case, case-to-heatsink, and heatsink-to-ambient resistances.

2. Rectifiers

A rectifier converts ac power to dc power. Rectifiers may be uncontrolled, semi-controlled, or fully controlled, and may be single-phase or three-phase.

2.1 Single-phase diode bridge

For an ideal single-phase full-wave bridge supplied by

vs = Vm sin(ωt)

the average output voltage is

Vdc = \frac{2Vm}{π} = \frac{2√2 Vs,rms}{π}

where Vm is the source peak voltage and Vs,rms is its rms value. This result assumes ideal diodes, a resistive or sufficiently smoothed load, and negligible source impedance.

With a large dc-side inductor, the current is nearly constant. The ac-side current then consists of rectangular pulses, producing low-order current harmonics and poor displacement or power factor unless power-factor correction is used.

2.2 Three-phase diode bridge

For a three-phase ideal six-pulse diode bridge,

Vdc = \frac{3√2}{π} VLL,rms ≈ 1.35VLL,rms

where VLL,rms is the line-to-line rms supply voltage. The dominant characteristic ripple frequency is six times the supply frequency. Source inductance causes commutation overlap, reducing the dc output and producing notches in the line voltage.

2.3 Controlled rectifiers

In a fully controlled three-phase bridge with continuous dc current and negligible overlap, the average output is

Vdc = 1.35 VLL,rms cos α

where α is the firing angle measured from the natural commutation point. For 0° < α < 90°, operation is normally rectifying. For larger firing angles, average power reversal may be possible if the dc source and current direction satisfy inverter-operation conditions. Safe operation requires appropriate commutation margin and control of circulating and fault currents.

A diode bridge is simple and robust but cannot normally return energy to the ac source. A controlled active front end uses self-commutated switches and PWM to regulate dc-link voltage, control input current, and enable regenerative braking.

3. DC–DC Choppers

A chopper applies high-frequency switching to convert one dc voltage level into another. The switching period is Ts, the on-time is Ton, and the duty ratio is

D = \frac{Ton}{Ts}

with 0 ≤ D ≤ 1.

3.1 Buck converter

For an ideal buck converter in continuous-conduction mode,

Vo = D Vin

The inductor volt-second balance gives this result. During the on interval, the inductor voltage is Vin - Vo; during the off interval it is -Vo. Equating average inductor voltage to zero yields D(Vin-Vo) + (1-D)(-Vo)=0.

The inductor current ripple is approximately

∆IL = \frac{(Vin-Vo)D}{L fs}

where L is inductance in henries and fs=1/Ts is switching frequency. The formula assumes ideal switching, constant input voltage, and continuous current.

3.2 Boost converter

For an ideal boost converter in continuous conduction,

Vo = \frac{Vin}{1-D}

The output voltage becomes large as D approaches unity; however, losses, switch stress, diode recovery, control sensitivity, and discontinuous operation limit practical duty ratio.

3.3 Four-quadrant choppers

A four-quadrant converter, commonly constructed from an H-bridge, can apply either voltage polarity and conduct current in either direction. It therefore supports forward motoring, forward braking, reverse motoring, and reverse braking. Antiparallel diodes or synchronous switches provide paths for inductive current during freewheeling and regeneration.

4. Voltage-Source Inverters

A voltage-source inverter (VSI) converts dc-link voltage into ac voltage. Its dc link is usually supported by a capacitor, so the dc-link voltage is relatively stiff. A three-phase, two-level VSI has six switches arranged in three legs. Each leg must never have both upper and lower switches conducting simultaneously, because this would short-circuit the dc link. A dead time is inserted between complementary gate signals.

Let the upper switch state of phases a, b, and c be Sa,Sb,Sc ∈ {0,1}, where 1 denotes the upper switch on. The eight switching states are:

StateSa Sb ScCondition
0000Zero vector
1100Active vector
2110Active vector
3010Active vector
4011Active vector
5001Active vector
6101Active vector
7111Zero vector

For a balanced three-wire load, the phase voltage relative to the load neutral can be written as

van = \frac{Vdc}{3}(2Sa-Sb-Sc)

with analogous expressions for phases b and c. This assumes an ideal split-free two-level bridge and a floating load neutral. The two zero states produce zero line-to-line voltage, while the six active states form six equal vectors spaced by 60 electrical degrees in the stationary reference plane.

The inverter output is not a pure sinusoid. The motor inductance filters much of the switching ripple, while the fundamental component produces useful torque and flux. Excessive common-mode voltage can cause bearing currents and insulation stress; careful switching sequences, shaft grounding, common-mode filters, and suitable cable practices may be required.

5. Pulse-Width Modulation

PWM controls the average value of a switched waveform by varying pulse widths. In sinusoidal PWM, a sinusoidal reference is compared with a triangular carrier. For a two-level inverter, the amplitude modulation index is commonly defined as

ma = \frac{Vref,peak}{Vtri,peak}

and the frequency ratio is

mf = \frac{fs}{f1}

where fs is carrier frequency and f1 is fundamental output frequency. In the linear modulation region of a sinusoidal three-phase VSI, the fundamental line-to-line rms voltage is approximately

VLL,1,rms = \frac{√3}{2√2} ma Vdc

for 0 ≤ ma ≤ 1, under the stated conventional definition of ma. Beyond this region, overmodulation increases the fundamental voltage but introduces additional low-order harmonics and ultimately approaches square-wave operation.

Third-harmonic injection adds a zero-sequence component to the three references. Because the three third-harmonic components cancel in line-to-line voltages, the dc bus is used more effectively while phase-reference peaks remain within the carrier range.

PWM design involves trade-offs. Increasing switching frequency generally reduces low-frequency current ripple and acoustic noise but increases switching loss, electromagnetic emissions, and gate-drive demands. Random PWM can spread spectral energy, whereas fixed-frequency PWM simplifies filtering and control synchronization.

6. Space-Vector Modulation

Space-vector modulation represents the three-phase reference as one rotating vector in a stationary two-axis plane. In each sampling interval, the desired reference is synthesized from adjacent active vectors and one or both zero vectors.

For a reference vector \vec V^* in a sector, let T1 and T2 be the dwell times of the adjacent active vectors, and let T0 be the total zero-vector time. The volt-second balance is

\vec V^* Ts = \vec V1 T1 + \vec V2 T2 + \vec V0 T0

with

T0 = Ts - T1 - T2

The dwell times must be nonnegative; otherwise the reference is outside the linear modulation region. A common implementation uses a symmetric sequence such as zero–active 1–active 2–zero–active 2–active 1–zero. This reduces certain low-frequency errors and limits abrupt changes in switching state.

Compared with sinusoidal PWM, SVM generally provides better dc-bus utilization and a systematic method for selecting switching states. Its implementation requires sector identification, coordinate transformations, timing calculation, and careful handling of dead time and minimum pulse widths.

7. Multilevel Converters

Multilevel converters synthesize several voltage levels rather than only +Vdc/2 and -Vdc/2. Important topologies include:

  • Neutral-point-clamped converter: clamping diodes connect switches to intermediate dc-link points.
  • Flying-capacitor converter: capacitors provide clamping voltages.
  • Cascaded H-bridge converter: isolated dc sources or converter cells are connected in series.
  • Modular multilevel converter (MMC): many half-bridge or full-bridge submodules are inserted or bypassed.

For a three-level neutral-point-clamped leg, the output can be positive, zero, or negative relative to the dc-link midpoint. Smaller voltage steps reduce dv/dt, motor insulation stress, and output filtering requirements. They also improve waveform quality at a given device switching frequency.

The disadvantages are increased component count, gate-drive complexity, capacitor-voltage balancing, unequal loss distribution, and more difficult fault management. MMCs are especially attractive in high-voltage dc transmission, medium-voltage drives, and grid interfaces because they provide scalable voltage ratings and high-quality waveforms.

8. Matrix Converters

A matrix converter directly converts fixed-frequency ac to variable-frequency ac without a bulky dc-link energy-storage capacitor. A three-phase-to-three-phase matrix converter contains nine bidirectional switches. It can provide adjustable output frequency, bidirectional power flow, and near-sinusoidal input current with suitable modulation.

Its switching constraints are strict:

  1. An output phase must not be disconnected from the load inductive current.
  2. Input phases must not be short-circuited.
  3. Commutation must account for current direction and switch delays.

Matrix converters offer compactness and long life because they avoid large electrolytic capacitors. Their limitations include a restricted maximum voltage transfer ratio, complicated commutation, sensitivity to input disturbances, and substantial control complexity. A related indirect matrix converter uses a virtual rectifier and inverter connected through a fictitious dc link, allowing established modulation concepts while retaining reduced energy storage.

9. Filters and Harmonics

A converter waveform contains the desired fundamental component and unwanted components at switching-frequency sidebands, low-order frequencies, common-mode frequencies, and resonance frequencies. Harmonic current can cause additional motor heating, torque pulsation, acoustic noise, transformer losses, and interference with communication or protection systems.

Total harmonic distortion of a voltage or current waveform is defined as

THD = \frac{\sqrt{Xrms2-X1,rms2}}{X1,rms} × 100\%

where Xrms is total rms quantity and X1,rms is the rms fundamental component. This definition assumes the waveform has a well-defined fundamental and that all remaining components are treated as distortion.

An L filter uses an inductor to oppose high-frequency current ripple. An LC filter adds a capacitor and provides stronger attenuation but introduces resonance. An LCL filter offers high attenuation with smaller inductance in grid converters, but resonance damping is essential. Filter design must consider rated current, reactive current, capacitor inrush, resonance, control-loop interaction, insulation, and standards.

For an ideal series inductor,

vL = L \frac{di}{dt}

Thus, rapid current changes require a voltage across the inductor. This relation explains why inductors smooth converter current and why stray inductance generates overvoltage during switching.

10. Protection, Gate Drive, and Practical Implementation

A robust converter includes electrical, thermal, mechanical, and software protection. Important measures include:

  • semiconductor fuses or fast circuit breakers;
  • desaturation or overcurrent detection;
  • dc-link precharge to limit capacitor inrush;
  • snubbers, clamps, and low-inductance busbars;
  • gate-driver isolation and undervoltage lockout;
  • controlled dead time and interlock logic;
  • overvoltage, overtemperature, and ground-fault detection;
  • safe shutdown and fault logging;
  • thermal sensors and correctly designed heat sinks.

Dead time prevents shoot-through but distorts the fundamental voltage, particularly at low speed where the commanded voltage is small. Compensation can improve accuracy, but it must account for current polarity, device delays, diode recovery, and minimum pulse width.

The dc-link capacitor is subject to ripple current. Its rms ripple-current rating, lifetime, temperature, and equivalent series resistance must be checked. Film capacitors provide high ripple capability and long life but may be larger or more expensive; electrolytic capacitors offer high energy density but are more temperature- and lifetime-sensitive.

Safety engineering requires isolation, protective earthing, accessible disconnects, discharge resistors, warning labels, and verification that the dc link has fallen to a safe voltage before servicing. Electromagnetic compatibility should be designed rather than treated as a final corrective measure.

Worked Example: Three-Phase VSI and DC-Link Selection

An ideal two-level inverter is supplied by a 600 V dc link. It uses sinusoidal PWM with ma=0.8. Estimate the fundamental line-to-line rms output voltage in the linear region.

Using

VLL,1,rms = \frac{√3}{2√2}maVdc

and √3/(2√2) ≈ 0.6124,

VLL,1,rms ≈ 0.6124 × 0.8 × 600 ≈ 294 V

Therefore, the estimated fundamental output is approximately 294 V rms line-to-line. The result assumes ideal switches, negligible dc-link ripple, balanced sinusoidal references, and operation within the linear modulation range. Device voltage drop, dead time, parasitic effects, and modulation nonlinearities would slightly alter the practical value.

Applied Case: Regenerative Medium-Voltage Drive

Background

Consider an illustrative 1 MW conveyor drive whose load frequently accelerates and decelerates. A conventional diode front end supplies a dc link for a three-level neutral-point-clamped inverter.

Problem

During braking, the motor returns energy to the dc link. A diode front end cannot normally send this energy back to the grid, so dc-link voltage rises and braking resistors dissipate energy. The plant also requires low input-current distortion.

Analysis and approach

The diode front end is replaced by an active front end using self-commutated switches. A current controller synchronizes the converter with the grid voltage. The controller commands near-sinusoidal grid current and controls active power. During motoring, power flows from grid to dc link; during braking, the current reverses and energy flows from the dc link to the grid. An LCL filter attenuates switching ripple, while an MMC or three-level topology limits device voltage stress and output dv/dt.

Protection includes precharge, semiconductor overcurrent detection, dc-link overvoltage trips, grid-loss detection, capacitor monitoring, and controlled shutdown. The modulation algorithm is constrained to preserve neutral-point balance and avoid excessive common-mode voltage.

Outcome

The active front end can reduce braking-resistor energy, improve power factor, and meet harmonic limits when correctly designed. It introduces additional switches, sensors, control software, and compliance requirements. The engineering decision therefore depends not only on efficiency but also on regenerative duty cycle, grid requirements, maintenance capability, and total lifecycle cost.

Learning points

  1. Converter topology must be selected from the energy-flow requirements, not only the motor rating.
  2. Regeneration requires a bidirectional power path and suitable control.
  3. Harmonic compliance is a system property involving topology, modulation, filters, and grid impedance.
  4. Protection and thermal design are as important as the ideal switching equations.

Challenges, Limitations, and Contemporary Developments

Ideal converter analysis does not predict all field behavior. Semiconductor parameter variation, magnetic saturation, cable reflections, common-mode currents, grid impedance, measurement delay, and digital sampling produce nonideal effects. A simulation that omits dead time, device recovery, thermal coupling, and parasitic inductance may give dangerously optimistic results.

Wide-bandgap SiC and GaN devices enable higher switching frequency, lower switching loss, and smaller filters. Their fast transitions can nevertheless increase electromagnetic emissions, bearing currents, and insulation stress. Advanced drives therefore combine fast devices with controlled gate slew rate, optimized layout, shielding, and common-mode mitigation.

Digital control and model-based algorithms permit predictive current control, adaptive dead-time compensation, sensorless operation, and condition monitoring. Artificial intelligence can assist fault classification and remaining-life estimation, but it does not remove the need for physical protection, validated datasets, explainable alarms, and fail-safe behavior. Safety-critical switching decisions should remain bounded by deterministic hardware and supervisory interlocks.

Best practice is to define operating scenarios, calculate worst-case voltage and current stresses, select switching frequency from loss and waveform requirements, verify thermal margins, test abnormal conditions, and validate conducted and radiated emissions. Standards and local electrical codes should be consulted for insulation coordination, electromagnetic compatibility, functional safety, grid connection, and machine protection.

Chapter Summary

Power electronic converters use semiconductor switches to control electrical energy in drives. Diodes and thyristors support simple or line-commutated rectification, while MOSFETs, IGBTs, IGCTs, and wide-bandgap devices enable controlled high-frequency switching. Rectifiers produce dc power, choppers regulate dc voltage and support multiple quadrants, and VSIs synthesize variable-frequency ac for motor control. Switching states determine instantaneous phase voltages, while PWM and SVM determine their average fundamental values and harmonic spectra.

Multilevel converters reduce voltage steps and improve waveform quality but require balancing and more complex control. Matrix converters provide direct ac–ac conversion and bidirectional operation without a large dc-link capacitor, at the cost of difficult commutation and control. Filters reduce ripple and harmonics, but resonance and control interaction must be considered. Practical designs must integrate thermal management, dead time, gate drive, precharge, overcurrent protection, dc-link safety, electromagnetic compatibility, and regulatory requirements.

Key Terms

Semiconductor switch: A device operated mainly in on or off states to control power flow.

Rectifier: A converter that changes ac power into dc power.

Chopper: A dc–dc converter using controlled switching.

Voltage-source inverter: An inverter supplied by a stiff dc-voltage source.

Current-source inverter: An inverter supplied by a stiff dc-current source.

Duty ratio: The fraction of a switching period for which a switch is on.

PWM: Pulse-width modulation used to control average converter voltage.

Modulation index: Ratio defining the reference magnitude relative to the carrier magnitude.

Space vector: A two-axis representation of a balanced three-phase quantity.

Dead time: Intentional delay preventing simultaneous conduction in one inverter leg.

Commutation: Transfer of current from one semiconductor path to another.

Freewheeling: Continued inductive-current circulation through a diode or switch path.

Multilevel converter: A converter that synthesizes more than two voltage levels.

Matrix converter: A direct ac–ac converter using bidirectional switches.

Harmonic: A sinusoidal component whose frequency is an integer multiple of the fundamental.

THD: Ratio of rms distortion content to rms fundamental content.

Common-mode voltage: Voltage shared by all phase conductors relative to a reference.

Active front end: A controlled rectifier capable of regulating current and often returning power to the grid.

Snubber: A circuit that limits switching overvoltage or rate of voltage change.

Review Questions

Short Answer Questions

  1. Why is dead time required in a two-level voltage-source inverter?
  2. State the ideal continuous-conduction voltage relation for a buck converter.
  3. What are the two zero states of a three-phase two-level inverter?
  4. Define total harmonic distortion.
  5. Give one advantage and one limitation of a multilevel converter.

Long Answer Questions

  1. Explain the construction, switching states, and output-voltage formation of a three-phase two-level VSI.
  2. Compare diode, controlled, and active rectifiers with respect to power flow and input-current quality.
  3. Explain sinusoidal PWM and space-vector modulation, including their principal differences.
  4. Describe the operation and practical constraints of a matrix converter.
  5. Discuss converter protection, thermal management, dead time, and electromagnetic compatibility.

Analytical / Application Questions

  1. A 400 V line-to-line rms three-phase supply feeds an ideal six-pulse diode bridge. Calculate the average dc output voltage using the continuous-current approximation.
  2. A buck converter has Vin=300 V, D=0.6, L=2 mH, and fs=10 kHz. Calculate the ideal output voltage and approximate inductor current ripple.
  3. A regenerative drive repeatedly brakes a large load. Recommend a converter arrangement and justify the choice in terms of energy flow, harmonics, protection, and control complexity.

Multiple Choice Questions with Answers

Question 1

For an ideal three-phase six-pulse diode bridge supplied by line-to-line rms voltage VLL,rms, the average dc output is approximately:

A. 0.45VLL,rms

B. 1.35VLL,rms

C. 2.00VLL,rms

D. πVLL,rms

Correct Answer: B. 1.35VLL,rms

Question 2

In an ideal buck converter operating in continuous conduction, the output voltage is:

A. Vin/(1-D)

B. (1-D)Vin

C. DVin

D. Vin/D

Correct Answer: C. DVin

Question 3

Which pair represents the zero switching states of a three-phase two-level VSI?

A. 100 and 011

B. 110 and 001

C. 000 and 111

D. 010 and 101

Correct Answer: C. 000 and 111

Question 4

The main purpose of dead time in an inverter leg is to:

A. Increase dc-link voltage

B. Prevent simultaneous conduction of upper and lower switches

C. Eliminate all output harmonics

D. Increase motor frequency

Correct Answer: B. Prevent simultaneous conduction of upper and lower switches

Question 5

For a 600 V dc link and sinusoidal PWM modulation index 0.8, the approximate fundamental line-to-line rms voltage is:

A. 184 V

B. 294 V

C. 480 V

D. 735 V

Correct Answer: B. 294 V

Question 6

Space-vector modulation synthesizes a reference voltage during each sampling interval by using:

A. Only the two zero vectors

B. Adjacent active vectors and zero-vector time

C. A resistor and capacitor network only

D. A line-frequency transformer

Correct Answer: B. Adjacent active vectors and zero-vector time

Question 7

Which converter topology directly converts ac to ac without a bulky dc-link energy-storage capacitor?

A. Buck converter

B. Diode bridge

C. Matrix converter

D. Single-phase half-wave rectifier

Correct Answer: C. Matrix converter

Question 8

A principal benefit of a multilevel converter is:

A. Higher voltage steps at the motor terminals

B. Reduced number of semiconductor devices in all cases

C. Improved waveform quality with smaller voltage steps

D. Elimination of capacitor-voltage balancing

Correct Answer: C. Improved waveform quality with smaller voltage steps

Question 9

If total rms current is 10 A and fundamental rms current is 8 A, the THD is approximately:

A. 20%

B. 60%

C. 75%

D. 125%

Correct Answer: B. 60%

Question 10

An active front end is especially useful in a drive that:

A. Never changes speed or torque

B. Requires regenerative power flow to the grid

C. Uses only a mechanical brake

D. Must prohibit bidirectional power flow

Correct Answer: B. Requires regenerative power flow to the grid

Electrical Drives & Control
Chapter 4 of 14