Power Supplies and Power Electronic Devices

Introduction

Power supplies convert electrical energy into the voltage, current, frequency, and isolation required by a load. Their design requires more than selecting a converter topology. A successful product must also manage semiconductor stress, magnetics, feedback stability, heat, electromagnetic interference (EMI), abnormal operating conditions, power quality, safety, and cost.

Linear supplies remain useful where low noise and simplicity dominate. Switched-mode power supplies (SMPSs) achieve much higher efficiency and power density by operating semiconductor devices primarily in their low-loss ON and OFF states. Modern systems increasingly employ digital control, bidirectional conversion, battery interfaces, and wide-bandgap devices such as silicon carbide (SiC) MOSFETs and gallium nitride (GaN) transistors.

Learning Objectives

After studying this chapter, students should be able to:

• design basic linear and switched-mode power supplies;
• analyze buck, boost, buck–boost, and flyback converters;
• estimate semiconductor, magnetic, and thermal stresses;
• distinguish MOSFETs, IGBTs, thyristors, SiC, and GaN devices;
• design transformer turns, inductance, air gap, and flux density;
• evaluate control-loop stability and compensation quantitatively;
• apply gate driving, isolation, protection, and EMI-control practices;
• assess efficiency, power factor, battery interfaces, and regulatory constraints.

1. Power-Supply Architecture and Specifications

A mains-powered electronic supply commonly includes an input fuse, surge suppressor, EMI filter, rectifier, bulk capacitor or power-factor-correction (PFC) stage, DC–DC converter, output filter, feedback system, and protection circuits. Isolation is required where safety standards demand separation between hazardous primary circuitry and accessible secondary circuitry.

Important specifications include:

Efficiency is

where η is dimensionless, and powers are in watts. At 90% efficiency, a 100 W supply dissipates 100/0.90-100=11.1 W.

2. Rectification and Linear Power Supplies

2.1 Rectifier and reservoir capacitor

A full-wave bridge reverses the negative half-cycle of an AC input. With a sufficiently large reservoir capacitor, the approximate peak-to-peak ripple is

where I_load is approximately constant, C is capacitance, and f_r is the ripple frequency. For full-wave rectification, f_r=2f_line. This approximation assumes small ripple and short capacitor-charging intervals.

For a 1 A load, 100 Hz ripple frequency, and 4700 μF capacitor,

The bridge must withstand repetitive current pulses much larger than the average load current. These narrow pulses also degrade power factor and increase harmonic distortion.

2.2 Linear regulators

A linear regulator continuously adjusts a pass transistor so that the output remains constant. For a regulator carrying output current I_o, its approximate dissipation is

where I_q is quiescent current. If I_q is negligible,

This is an upper-bound approximation and applies only when input and output currents are nearly equal.

Worked example—thermal design. A 12 V source supplies a 5 V, 0.8 A regulator with negligible quiescent current:

If ambient temperature is 40°C and maximum junction temperature is 125°C, the maximum total junction-to-ambient thermal resistance is

If junction-to-case and interface resistances total 3°C/W, the heat sink must provide no more than approximately 15.2-3=12.2 °C/W. Thermal design must also account for airflow, mounting orientation, and tolerance.

Low-dropout regulators (LDOs) require only a small input–output difference but may become unstable with unsuitable output-capacitor values or equivalent series resistance (ESR). Manufacturer stability requirements are therefore part of the design, not optional recommendations.

3. Switched-Mode Conversion

SMPS converters transfer energy using controlled switches, inductors, transformers, diodes, and capacitors. The following ideal relationships assume steady-state continuous-conduction mode (CCM), lossless components, and sufficiently small ripple.

3.1 Buck converter

During the switch ON interval, the inductor voltage is V_in-V_o. During the OFF interval it is -V_o. Inductor volt-second balance gives

therefore

The inductor ripple is

where D is duty ratio, L is inductance, and f_s is switching frequency.

For triangular capacitor current and negligible control interaction, buck output ripple can be estimated by

where r_C is capacitor ESR. The ESR term represents the approximate peak-to-peak step associated with triangular inductor ripple.

Example. For 24 V to 12 V, 5 A, 100 kHz, and a target ripple of 1 A peak-to-peak:

3.2 Boost converter

Ideal CCM volt-second balance gives

and

The output cannot normally be lower than the input through ordinary boost action. In CCM, the boost converter also has a right-half-plane zero: increasing duty initially deprives the output of energy before raising its long-term voltage. For an ideal boost converter with resistive load R and inductance L,

Feedback crossover is commonly kept well below this frequency, often below approximately one-fifth of it.

3.3 Inverting buck–boost converter

For the ideal inverting topology in CCM,

The negative sign indicates reversed output polarity. Non-inverting buck–boost converters use additional switches or coupled structures to provide either step-up or step-down operation without polarity reversal.

3.4 Flyback converter

A flyback converter stores energy in a gapped coupled inductor while the primary switch is ON and releases it to the secondary while the switch is OFF. Under ideal CCM conditions,

where N_p and N_s are primary and secondary turns. Diode drops, leakage inductance, winding resistance, dead time, and snubber losses alter the practical result.

Figure: Galvanically isolated flyback converter. Power crosses through the transformer and feedback crosses only through the optocoupler; no direct conductor crosses the isolation boundary.

Alternatives to an optocoupler include an isolated error amplifier, a certified digital isolator, or an auxiliary transformer winding with primary-side regulation. Safety depends on insulation rating, creepage, clearance, transformer construction, and certification—not merely on drawing a boundary in a schematic.

4. Semiconductor Power Devices

4.1 Loss estimation

MOSFET conduction loss is approximately

The resistance must be evaluated at operating junction temperature. A first-order hard-switching estimate is

where voltage and current are assumed to overlap linearly during transitions. This excludes output-capacitance loss, reverse recovery, gate-drive loss, parasitics, and ringing.

Gate-drive power is approximately

where Q_g is total gate charge at the relevant operating conditions and V_g is gate-drive swing.

For IGBTs, a useful conduction model is

for an approximately constant ON-state drop, where d is conduction fraction. Datasheet curves provide more accurate values.

4.2 Gate drivers

A gate driver must source and sink enough current to charge and discharge device capacitances rapidly. Approximate average gate current is Q_gf_s, whereas peak current is governed by desired transition time and total gate-path impedance. Excessively fast switching increases EMI and overshoot; excessively slow switching increases switching loss.

High-side N-channel devices may use bootstrap drivers, isolated drivers, or isolated bias supplies. Half-bridge drivers require dead time to avoid shoot-through. Miller-induced turn-on can be reduced using low-impedance turn-off paths, negative gate bias where permitted, Miller clamps, and careful layout.

5. Magnetics Design

Magnetics frequently determine the size, losses, isolation capability, and achievable switching frequency of a converter.

5.1 Flux and turns

Faraday's law gives the winding voltage

where N is turns, A_e is effective core area in m², and B is flux density in tesla. For an approximately constant primary voltage during an interval t_on,

Therefore, for a selected permissible flux excursion,

The maximum must be taken over the full input and control range. Flux excursion must include reset behavior, tolerances, transient duty limits, and possible flux walking.

5.2 Inductance and air gap

For a core whose reluctance is dominated by a uniform air gap,

where μ₀=4π×10^-7 H/m and l_g is gap length. The formula neglects core reluctance and fringing, so practical design uses manufacturer A_L data or finite-element correction.

Stored magnetic energy is

In a flyback converter, this energy is deliberately stored mainly in the gap. A conventional transformer ideally transfers energy directly and normally does not require a large energy-storage gap.

5.3 Copper and core losses

Copper loss is

At high frequency, R_ac exceeds DC resistance because of skin and proximity effects. Skin depth is

where ρ is conductor resistivity, ω=2πf, and μ is permeability. Foil, litz wire, and interleaved windings can reduce AC loss, although interleaving may increase primary-to-secondary capacitance and common-mode EMI.

Core loss is often estimated using an empirical Steinmetz relation:

where P_v is loss per unit core volume and k, α, and β are material- and waveform-dependent coefficients. Manufacturer data must be used with consistent units. Nonsinusoidal excitation may require an improved generalized Steinmetz method or measured loss curves.

5.4 Worked flyback magnetics design

Design an illustrative 24–36 V input, 12 V/2 A isolated flyback at 100 kHz. Assume 90% efficiency for current estimation, a 0.5 V secondary diode drop, A_e=80 mm², and choose N_p/N_s=3.

The secondary voltage reflected to the primary is

At 24 V input, ideal volt-second balance gives

Choose N_p=18 and N_s=6. The low-line flux excursion is

Input power is 24/0.90=26.67 W, so low-line average input current is 26.67/24=1.111 A. Since primary current flows for fraction D, its approximate mean during the ON interval is

Choose a peak-to-peak magnetizing ripple of 40% of this value: ΔI=0.728 A. Then

The approximate peak current is 1.82+0.728/2=2.18 A. For a gap-dominated core,

This is a preliminary value. Final design must verify saturation including DC bias, fringing loss, leakage inductance, winding RMS currents, insulation margins, core loss, temperature rise, and worst-case current limiting.

6. Feedback and Control-Loop Compensation

A regulated converter uses negative feedback to reject disturbances. The loop gain is

where G_c is compensator transfer function, G_m is PWM or modulator gain, G_p is power-stage transfer function, and H is feedback gain.

At the gain-crossover frequency f_c, |T|=1. Phase margin is 180°+∠T(j2πf_c). Practical designs commonly target about 45°–70°, but acceptable values depend on transient requirements, tolerances, sampling delay, and nonlinear operating modes.

For an ideal voltage-mode buck converter with output-capacitor ESR r_C, resistive load R, and CCM operation, the duty-to-output transfer function is

This model neglects parasitic inductor resistance, delays, and discontinuous conduction. The LC double pole can introduce nearly 180° phase lag, while capacitor ESR introduces a zero at

A type-III compensator can be written as

The integrator gives zero steady-state error to a constant disturbance, two zeros add phase near the LC resonance, and high-frequency poles reduce noise gain.

Quantitative compensation example

Consider a 24-to-12 V voltage-mode buck converter with L=47 μH, C=220 μF, R=6 Ω, r_C=0.05 Ω, f_s=100 kHz, and a 1 V PWM ramp. The LC resonant frequency is

and the ESR zero is

Select f_c=10 kHz, zeros at 1.25 and 1.96 kHz, and poles at 14.5 and 50 kHz. Evaluating the power stage at 10 kHz gives approximately

With unity PWM gain and feedback gain for the normalized example, the required compensator magnitude is 1/0.726=1.377. Evaluating the frequency-dependent factors in the selected compensator gives

Thus choose

The compensator phase at crossover is approximately +25.9°, producing an estimated loop phase of -117.8° and phase margin of 62.2°. Component tolerances, PWM delay, sampling, load range, current limiting, and measured frequency response must still be checked. Crossover should normally remain below about one-tenth of switching frequency and below restrictive right-half-plane zeros.

7. Protection, Isolation, and Thermal Reliability

Common protection functions include:

• cycle-by-cycle overcurrent limiting;
• short-circuit shutdown or hiccup restart;
• input undervoltage lockout;
• output overvoltage protection;
• overtemperature shutdown;
• reverse-polarity and reverse-current protection;
• surge suppression using MOVs, TVS diodes, or active clamps;
• snubbers and clamps for leakage-inductance energy.

A fuse protects wiring and limits fire risk; it does not necessarily protect a semiconductor quickly enough. Protection coordination must consider fault energy, interruption rating, component safe operating area, and restart behavior.

Isolation components require adequate working voltage, transient rating, creepage, clearance, insulation thickness, and common-mode transient immunity. Reinforced insulation may replace two independent basic-insulation barriers when certified for the application.

8. EMI and Layout

Differential-mode noise circulates between supply conductors. Common-mode noise flows through parasitic capacitance to chassis, protective earth, or surrounding structures. An EMI filter may use differential inductors, common-mode chokes, X capacitors across the line, and safety-rated Y capacitors across an isolation barrier.

The most effective EMI measure is usually controlling current-loop geometry. High-di/dt commutation loops and high-dv/dt switching nodes should be physically small. Gate-drive returns should not share inductive paths with power current. Kelvin source connections improve gate-control accuracy.

Input filters can interact with a converter's negative incremental input resistance. Damping may be required to prevent oscillation. Compliance testing uses specified equipment such as a line-impedance stabilization network (LISN); an oscilloscope trace alone does not establish regulatory compliance.

9. Power Quality and Battery Interfaces

For sinusoidal voltage and current with phase difference φ, displacement power factor is cosφ. With distorted current, true power factor is

Passive rectifier-capacitor inputs often have poor distortion power factor. Active PFC shapes line current to follow line voltage, commonly using a boost stage. Harmonic-current limits apply to many product categories.

Battery-connected converters must support chemistry-specific limits and battery-management-system commands. Lithium-ion charging usually applies constant current followed by constant voltage, with current tapering during the voltage-limited phase. A charger must not infer safety from terminal voltage alone; cell temperature, individual cell voltages, pack current, isolation, contactor state, and fault status may be required.

Bidirectional converters support regenerative braking, vehicle-to-grid operation, energy storage, and uninterruptible supplies. Direction changes require controlled current ramps and prevention of DC-bus overvoltage.

10. Applied Case: Isolated Battery-Supported Control Supply

Illustrative case. An industrial controller requires an isolated 12 V, 24 W rail from a nominal 24–36 V battery bus.

Problem. The supply must maintain isolation, exceed 88% efficiency, avoid transformer saturation, survive output shorts, and meet conducted-emissions requirements.

Analysis. A linear regulator would dissipate at least (24-12)2=24 W at minimum input and even more at high input, so it is unsuitable. An isolated flyback is appropriate at 24 W because it provides isolation with one magnetic component and modest circuit complexity. The earlier magnetics calculation gives an 18:6 turns ratio, approximately 201 μH magnetizing inductance, 0.102 T low-line flux excursion, and 2.18 A estimated peak primary current.

Approach. The design uses a 100 V MOSFET only if leakage overshoot is clamped sufficiently below its derated voltage limit; otherwise a higher-rated device is selected. Secondary output sensing drives an optocoupler, ensuring that the feedback path crosses the barrier only through a certified isolation component. Cycle-by-cycle current limiting, hiccup short-circuit protection, input undervoltage lockout, a primary RCD or active clamp, and thermal shutdown are included. A common-mode choke and differential input filter are added after measuring noise with a LISN. The filter is damped and checked for control interaction.

Outcome. Prototype acceptance requires efficiency and temperature measurements across input and load range, transformer hipot testing, loop-gain measurement, short-circuit testing, surge testing, and pre-compliance EMI scans. Passing nominal bench operation alone is insufficient.

Learning points. Topology, magnetics, semiconductor rating, feedback isolation, compensation, protection, and EMI form one coupled design problem. Optimizing any one in isolation can compromise another.

11. Contemporary Developments and Professional Practice

SiC and GaN devices enable higher switching frequencies and smaller passive components, but their fast edges magnify parasitic inductance, common-mode current, insulation stress, and measurement error. Suitable drivers, low-inductance packaging, controlled gate resistance, and high-bandwidth isolated probes are essential.

Digital power controllers can implement adaptive dead time, telemetry, fault logging, nonlinear control, and parameter estimation. Machine-learning methods may support predictive maintenance or anomaly detection, but safety-critical shutdown should remain deterministic, independently verifiable, and robust against corrupted data.

Designers must comply with electrical-safety, EMC, environmental, and energy-efficiency regulations applicable to the market. Ethical engineering includes honest efficiency reporting, adequate component derating, traceable safety analysis, repairability where feasible, and avoidance of protection shortcuts that transfer risk to users or service personnel.

Chapter Summary

Power-supply design integrates energy conversion, control, magnetics, semiconductor operation, thermal management, isolation, EMI, and protection. Linear regulators offer low noise and simplicity but become inefficient at large voltage drops. SMPS topologies use switched inductive energy transfer to achieve high efficiency. Volt-second balance determines ideal conversion ratios and magnetic flux excursion, while current ripple determines inductance and device stress.

MOSFETs, IGBTs, thyristors, SiC, and GaN devices serve different voltage, frequency, and power ranges. Their selection requires conduction, switching, gate-drive, thermal, and fault analysis. Magnetics design requires verified turns, air gap, saturation margin, winding loss, core loss, and insulation. Stable regulation depends on the complete loop gain; compensation must account for LC poles, ESR zeros, switching delay, and right-half-plane zeros. Safe professional designs also require explicit isolation in every barrier-crossing power, sensing, or control path, coordinated protection, compliant EMI filtering, and validated battery and power-quality behavior.

Key Terms

Buck converter: A topology that ideally reduces DC voltage.

Boost converter: A topology that ideally increases DC voltage.

Continuous-conduction mode: Operation in which inductor current never reaches zero during a cycle.

Duty ratio: Fraction of a switching period for which a controlled switch is ON.

Flyback converter: An isolated converter that stores energy in a gapped coupled inductor.

Galvanic isolation: Prevention of direct conductive connection between circuit domains.

Gate driver: Circuit that supplies controlled gate voltage and current to a power switch.

IGBT: Insulated-gate bipolar transistor used mainly at higher voltage and power.

Line regulation: Output sensitivity to input-voltage variation.

MOSFET: Voltage-controlled field-effect transistor widely used for power switching.

Optocoupler: Isolation device that transfers information using light.

Phase margin: Angular distance between loop phase and −180° at gain crossover.

Power factor: Ratio of real power to RMS apparent power.

Right-half-plane zero: A nonminimum-phase zero that initially drives output opposite to its final response.

Skin effect: Concentration of AC current near a conductor surface.

Snubber: Network that limits switching transients or damps ringing.

Thyristor: Latching semiconductor switch triggered through a gate.

Volt-second balance: Steady-state requirement that average inductor voltage over a period is zero.

Wide-bandgap device: Semiconductor device, notably SiC or GaN, with a larger bandgap than silicon.

Review Questions

Short Answer Questions

1. Why does a linear regulator become inefficient when its input–output voltage difference is large?
2. State the ideal CCM conversion ratio of a buck converter.
3. What is the purpose of dead time in a half-bridge?
4. Distinguish differential-mode EMI from common-mode EMI.
5. Why is an air gap commonly used in a flyback magnetic component?

Long Answer Questions

1. Derive the ideal CCM voltage ratio of a boost converter using volt-second balance.
2. Compare MOSFETs, IGBTs, SiC MOSFETs, and GaN devices in terms of loss, frequency, voltage, and application.
3. Explain a complete transformer or coupled-inductor design process, including turns, flux, gap, copper loss, core loss, and insulation.
4. Explain how LC poles, ESR zeros, and right-half-plane zeros influence converter compensation.
5. Discuss the protection, isolation, thermal, and regulatory requirements of a mains-powered SMPS.

Analytical / Application Questions

1. Design the inductor for a 36-to-12 V, 4 A buck converter at 150 kHz with 30% peak-to-peak current ripple. State all assumptions.
2. A MOSFET carries 8 A RMS with R_DS(on)=40 mΩ at operating temperature. It switches 200 V and 8 A with a combined linear transition time of 50 ns at 100 kHz. Estimate conduction and hard-switching losses.
3. Propose an isolated 48 V battery-to-15 V, 60 W supply, identifying topology, semiconductor ratings, feedback isolation, magnetics checks, protection, compensation, and EMI validation.

Multiple Choice Questions with Answers

Question 1

An ideal CCM buck converter operates from 30 V with duty ratio 0.4. What is its output voltage?

A. 7.5 V

B. 12 V

C. 30 V

D. 75 V

Correct Answer: B. 12 V

Question 2

A full-wave rectifier connected to a 50 Hz source produces a reservoir-capacitor ripple frequency of:

A. 25 Hz

B. 50 Hz

C. 100 Hz

D. 200 Hz

Correct Answer: C. 100 Hz

Question 3

A linear regulator converts 15 V to 5 V at 1 A with negligible quiescent current. Its approximate power dissipation is:

A. 3.33 W

B. 5 W

C. 10 W

D. 15 W

Correct Answer: C. 10 W

Question 4

Which component explicitly transfers feedback across the isolation boundary in the chapter's isolated flyback architecture?

A. Ordinary copper trace

B. Optocoupler

C. Output capacitor

D. Gate resistor

Correct Answer: B. Optocoupler

Question 5

What is the principal magnetic purpose of the air gap in a flyback coupled inductor?

A. To eliminate winding resistance

B. To store energy and control inductance

C. To create galvanic isolation

D. To remove all leakage inductance

Correct Answer: B. To store energy and control inductance

Question 6

A MOSFET carries 5 A RMS and has R_DS(on)=20 mΩ at operating temperature. Its conduction loss is:

A. 0.1 W

B. 0.5 W

C. 2.5 W

D. 5 W

Correct Answer: B. 0.5 W

Question 7

Which converter has a right-half-plane zero under CCM operation?

A. Ideal voltage-mode buck converter

B. Linear regulator

C. Boost converter

D. Resistive voltage divider

Correct Answer: C. Boost converter

Question 8

At loop gain crossover, a converter has a loop phase of −125°. What is its phase margin?

A. 35°

B. 55°

C. 125°

D. 305°

Correct Answer: B. 55°

Question 9

Which device normally remains ON after its gate trigger is removed until current falls below its holding current?

A. MOSFET

B. IGBT

C. SCR

D. Zener diode

Correct Answer: C. SCR

Question 10

A 100 W converter operates at 80% efficiency. What is its total power loss?

A. 20 W

B. 25 W

C. 80 W

D. 125 W

Correct Answer: B. 25 W