The Repair LibraryRead · Learn · Master

Switching Power Supply Theory

The earlier volumes taught the switching regulator as a circuit — the buck that steps down, the boost that steps up, the topologies and their waveforms; this chapter opens by taking that foundation up to the level a professional repair of a modern device demands, treating the switching supply not as a diagram but as the power heart of the device and, as the last chapter kept insisting, its loudest noise source. The theory a technician needs here is the theory of a system. A switching supply is an energy-transfer engine: a switch chops the input into pulses, an inductor stores energy in its magnetic field during each on-time and releases it during each off-time, and an output capacitor smooths the result into steady direct current — and the whole thing is efficient for one reason worth understanding deeply, that the switch is either fully on, dropping almost nothing, or fully off, passing nothing, and never lingers in the dissipative middle where a linear regulator burns the difference as heat. On top of that engine sits the control. Pulse-width modulation is the lever: a controller varies the fraction of each cycle the switch spends on, and that duty fraction sets the output, turned continuously by a feedback control loop that senses the output and corrects it thousands of times faster than any load can change. Modern supplies add one more refinement the section treats carefully, synchronous rectification, which replaces the old catch diode with a second switch to recover the energy that diode wasted — a real efficiency gain that also introduces a real new set of failure modes. The section's purpose is not to re-derive the topologies but to give a repair technician the working model that makes a switching supply diagnosable: the cycle that moves the energy, the modulation that controls it, the synchronous switch that sharpens it, and the loop that holds it steady — because the health of every one of those shows in signals a technician can measure, and the chapters that follow, on the power-management ICs and the sequencing and failure analysis of real devices, all rest on this model of how the power is actually made.

AdvancedLow Risk23 min read

What You Will Learn

  • You will learn the switching supply as an energy-transfer engine — switch, inductor, and capacitor moving energy in cycles.
  • You will learn why a switching supply is efficient — the switch is fully on or fully off, never dissipating in the middle.
  • You will learn pulse-width modulation and how the duty fraction of each cycle sets the regulated output.
  • You will learn synchronous rectification — the second switch that replaces the catch diode — and the failure modes it adds.
  • You will learn the control loop that senses the output and corrects it, and how its health shows in the output a technician measures.

What You Will Be Able To Do

  • You will be able to describe the switching cycle as energy stored in the inductor and released to the output.
  • You will be able to explain why a switching supply runs cool where a linear regulator runs hot.
  • You will be able to read pulse-width modulation and relate the duty fraction to the regulated output voltage.
  • You will be able to recognize a synchronous rectifier on a board and name the failure modes it introduces.
  • You will be able to reason about a control loop's health from the output's ripple and transient response.

Required Tools

  • A donor board with a visible switching regulator — an inductor, a controller, and the switching node to trace as a system
  • An oscilloscope — the instrument that makes the switching cycle, the ripple, and the control loop's response visible
  • A donor device with a synchronous buck converter — a high-side and low-side FET pair to recognize as synchronous rectification
  • The notebook — this section records the switch, inductor, capacitor, modulation, and loop of each supply studied

When NOT to Attempt This

Do not attempt this section if any of the following apply to you:

  • You are not comfortable working with small surface-mount components.
  • You have not completed the prerequisite sections for this skill.
  • You do not have the required tools in working condition.

Section Overview

The earlier volumes taught the switching regulator as a circuit; this chapter opens on it as the power heart of the device — and its loudest noise source (switching-regulators-buck-boost-buck-boost). A switching supply is an energy-transfer engine. A switch chops the input, an inductor stores energy in its field each on-time and releases it each off-time, and a capacitor smooths the result to steady DC (capacitor-and-inductor-failure-modes). Its efficiency has one root. The switch is fully on, dropping almost nothing, or fully off, passing nothing — never in the dissipative middle where a linear regulator burns the difference as heat (voltage-regulators-linear-and-switching). The control is pulse-width modulation. The fraction of each cycle the switch spends on sets the output, turned continuously by a feedback loop. Modern supplies add the synchronous rectifier — a second switch replacing the catch diode, recovering its wasted energy and adding new failure modes. And the control loop holds it all steady. It senses the output and corrects it thousands of times faster than any load can change, its health visible in the output a technician measures (electromagnetic-interference-basics). The cycle, the modulation, the synchronous switch, and the loop — the model that makes a switching supply diagnosable.

Why This Matters

Every chapter that follows rests on this model of how a device's power is actually made (electromagnetic-interference-basics). This matters because the switching supply is the power heart of the device: nearly every rail in a modern device comes from a switching converter, so a technician who understands the cycle, the modulation, and the loop can reason about the power that everything else depends on, while one who treats it as a black box is blind to the most failure-prone system on the board (voltage-regulators-linear-and-switching). This matters because the model is what makes the supply diagnosable: the switch, the inductor, the rectifier, and the loop each leave a signature in signals a technician can measure — the switching node's waveform, the output's ripple, the response to a load step — so understanding the theory turns an oscilloscope trace into a diagnosis (switching-regulators-buck-boost-buck-boost). It matters because synchronous rectification changed the failure map: the second switch that made modern supplies efficient also added gate-drive faults, shoot-through, and a second power device to fail, so a technician who still pictures a simple diode misreads the modern converter (capacitor-and-inductor-failure-modes). And it matters because this is the foundation for the whole chapter: the power-management ICs, the sequencing, and the failure analysis that follow all assume a working picture of how a single switching rail is made and regulated, so the theory here is the ground the rest of the chapter stands on. Understand the energy cycle, the modulation, the synchronous switch, and the loop — and the power system stops being a black box and becomes a set of signals that tell a technician what is wrong.

Required Prerequisites

Before starting this section, you should have completed:

  • Note cards and a marker — the switch, inductor, capacitor, modulation, and loop of each supply are written out, because the working model is held as those five parts, not as a memorized schematic.
  • Printed reference waveforms — the switching-node square wave, the inductor's triangular current, and the output ripple are kept at hand, so a real trace is read against what healthy looks like.
  • Low-tack labels — the controller, the inductor, the switching node, and the feedback path are tagged on a donor board, so the theory maps onto a physical converter.
  • A donor board with a visible switching regulator — a controller, an inductor, and a switching node, so the energy-transfer engine is traced on real hardware.
  • A device with a synchronous buck converter — a high-side and low-side MOSFET pair, so synchronous rectification is recognized rather than pictured as a diode.
  • An oscilloscope and a bench supply — to see the switching node's square wave, the output ripple, and the loop's response to a changing load, turning the theory into observed signals.

Real-World Applications

This model is what a technician uses to reason about the power system before ever touching it. A repairer facing a dead rail on a laptop board pictures the energy cycle and knows to look for the switch, the inductor, the rectifier, and the controller as a system, not to guess at a single part (switching-regulators-buck-boost-buck-boost). A technician probing a switching node with a scope reads the square wave against the model — is it switching, at what duty, with what ripple — and turns the trace into a statement about which stage is working (voltage-regulators-linear-and-switching). A bench diagnosing a modern converter that runs hot recognizes a synchronous rectifier and reasons about gate drive and shoot-through, rather than searching for a diode that is not there (capacitor-and-inductor-failure-modes). And a tech chasing an output that sags whenever the load steps reasons about the control loop — too slow, or poorly compensated — instead of condemning a healthy power stage (electromagnetic-interference-basics). The confusions this prevents: a power system treated as a black box, a switching-node trace read without a model, a synchronous converter mistaken for a diode circuit, and a loop fault blamed on the power stage.

Common Challenges

  • The supply is a system, not a part. A dead rail invites a hunt for the one bad component — the switch, inductor, rectifier, and loop work together, and the fault is found by reasoning about the system, not by swapping parts (switching-regulators-buck-boost-buck-boost).
  • The waveforms need a model to read. A switching node is a fast square wave and the inductor current a triangle — without the energy-cycle model a scope trace is just a shape, and with it the shape is a diagnosis (voltage-regulators-linear-and-switching).
  • Synchronous rectification hides in plain sight. A modern converter looks like it should have a diode and does not — the second MOSFET is the rectifier, and missing it means misreading the whole stage (capacitor-and-inductor-failure-modes).
  • The loop is invisible until it misbehaves. A control loop leaves no obvious part to point at — its health shows only in the output's ripple and transient response, which must be read as the loop's signature (electromagnetic-interference-basics).

Safety Notes

Risk Level: Low. This section builds a model and observes waveforms — it reworks nothing — but power circuits carry real energy, so the standing bench law and live-power caution apply.

  • Real energy on low-voltage rails — a switching rail can source large currents, so a careless probe slip across the switching node can do damage even at low voltage.
  • Bulk capacitors hold charge — treat every bulk capacitor as charged until proven otherwise, and keep any high-voltage input section out of scope for this study.
  • Proper scope grounding — the switching node is a fast, hard-driven signal; probe it with correct grounding, never by bridging it to a nearby node.

Professional Tips Before Starting

  • Hold the five-part model. Switch, inductor, capacitor, modulation, loop — carry those five, and any switching supply resolves into a system you can reason about (switching-regulators-buck-boost-buck-boost).
  • Read waveforms against the model. The square wave, the triangle, the ripple — know what healthy looks like, and a real trace becomes a diagnosis (voltage-regulators-linear-and-switching).
  • Look for the second switch. A modern converter's rectifier is a MOSFET, not a diode — find it, and the synchronous stage reads correctly (capacitor-and-inductor-failure-modes).
  • Judge the loop by the output. Ripple and transient response are the loop's signature — read them, and the invisible loop becomes observable (electromagnetic-interference-basics).
  • Respect the energy. Even a low-voltage rail sources real current — probe carefully and treat bulk capacitors as charged.

The Working Model of a Switching Supply

The Energy Cycle — Switch, Inductor, Capacitor

The foundation the earlier volumes built gave the topologies; the model a professional needs treats them as one energy-transfer engine (switching-regulators-buck-boost-buck-boost). A switching supply does not divide voltage — it pumps energy in cycles. The switch, a MOSFET, connects the input for a fraction of each cycle. During that on-time, current builds through the inductor, and the inductor stores that energy in its magnetic field — not passing it straight through, but banking it. When the switch opens, the inductor cannot stop its current instantly, so it drives that stored energy onward to the output through the rectifier during the off-time, its field collapsing as it delivers. The output capacitor is the smoothing reservoir. It charges on the pulses of delivered energy and discharges into the load between them, ironing the pulsed delivery into the steady direct current the load actually sees, with only a small residual ripple at the switching frequency. One refinement keeps the picture honest across topologies. The bank-then-release image is exact for the boost and flyback types, whose rectifier feeds the output only during the off-time; in a buck — the topology this chapter leans on — the inductor sits in series with the load and feeds it throughout, merely charging harder on-time and discharging off-time, so the load is never starved between pulses, while the inductor remains the energy reservoir the switch fills and the output draws from. This is the whole engine. Input chopped into pulses, energy banked in the inductor and released to the output, and the capacitor smoothing the result — a pump that moves energy from input to output in packets, cycle after cycle, hundreds of thousands to millions of times a second. Switch, inductor, capacitor, cycle — the energy-transfer engine that every rail in the device is built on, and the first half of the model a technician carries to the bench.

Why It Runs Cool — and Why It Modulates

Two truths complete the engine: why a switching supply is efficient, and how it is controlled (voltage-regulators-linear-and-switching). The efficiency has one clean explanation. A switch dissipates power only as the product of the voltage across it and the current through it, and a switching supply keeps that product near zero: fully on, the voltage across the switch is almost nothing, so little is lost though the current is high; fully off, the current is nothing, so nothing is lost though the voltage is high; and the switch crosses the dissipative middle — where both are appreciable — only briefly at each transition. A linear regulator lives in exactly that middle. Its pass element holds a steady voltage across itself while the full load current flows, burning the input-to-output difference as heat — which is why a linear regulator dropping a large difference runs hot and a switching supply of the same power runs cool. The control is pulse-width modulation. The output depends on how much of each cycle the switch spends on — its duty fraction — because a longer on-time banks and delivers more energy per cycle, so varying that fraction varies the output. A controller does exactly that, continuously. It modulates the width of the on-pulse cycle by cycle, widening it to raise the output and narrowing it to lower it, which is the lever every switching supply is steered by. Two states for efficiency, a modulated pulse for control — the engine runs cool because the switch never lingers in the middle, and it regulates because the width of its on-pulse is the throttle.

The Synchronous Rectifier and the Control Loop

Two refinements turn the basic engine into the modern converter a technician actually meets: synchronous rectification and the feedback loop that steers the whole thing (capacitor-and-inductor-failure-modes). The first is the synchronous rectifier. In the basic converter, the off-time current returns through a catch diode, which wastes energy as its forward voltage drop times the current — a real loss at the high currents modern rails carry. The modern supply replaces that diode with a second MOSFET, switched on in complement to the main switch so it conducts the off-time current through its low on-resistance instead of a diode drop, recovering most of that wasted energy and raising efficiency. But the second switch adds a second set of failure modes. The two MOSFETs must never be on at once — that would short the input to ground, a destructive shoot-through — so a small dead-time separates them, and a gate-drive fault, a failed synchronous MOSFET, or lost dead-time is a new way for the modern converter to fail that the old diode circuit could not (voltage-regulators-linear-and-switching). The second refinement is the control loop. Everything so far runs open — but a supply must hold its output steady against changing input and load, so a feedback loop closes around it: it senses the output through a divider, compares it to a reference, and adjusts the pulse-width modulation to correct any error, thousands of times faster than any load can change. The loop's design is a balance. Too slow or poorly compensated, and the output sags on a load step or oscillates; well designed, it holds the output flat through everything — and because the loop leaves no single part to point at, its health is read in the output's ripple and its response to a load transient (electromagnetic-interference-basics). Synchronous switch for efficiency, feedback loop for regulation — the two refinements that complete the model of the converter a technician diagnoses.

Common Mistakes

  • Treating the supply as a black box. The power system is skipped as too complex to reason about — it is the most failure-prone system on the board, and the model makes it diagnosable (switching-regulators-buck-boost-buck-boost).
  • Reading a switching node without the model. The square wave is noted but not interpreted — duty, amplitude, and ripple each mean something against the energy-cycle model, and mean nothing without it (voltage-regulators-linear-and-switching).
  • Picturing a diode where a MOSFET rectifies. A modern converter is read as if it had a catch diode — the synchronous rectifier is a second MOSFET, and missing it misreads the stage and its failure modes (capacitor-and-inductor-failure-modes).
  • Blaming the power stage for a loop fault. An output that sags or oscillates is treated as a bad power component — the control loop, too slow or unstable, is often the cause, and its signature is in the transient response (electromagnetic-interference-basics).
  • Forgetting the energy in a low-voltage rail. A low voltage is treated as harmless — a switching rail sources real current, and a probe slip or a charged bulk capacitor can still do damage.

Troubleshooting Guidance

  • A rail is dead and the cause is unclear — reason about the system: picture the switch, inductor, rectifier, and controller as an energy-transfer engine and ask which stage of the cycle has stopped, rather than swapping parts, because the fault lives in a system the model makes legible (switching-regulators-buck-boost-buck-boost).
  • A switching-node trace is ambiguous — read it against the model: check whether it is switching at all, at what duty fraction, and with what ripple, because each reads as a statement about the modulation and the stage when set against the energy cycle (voltage-regulators-linear-and-switching).
  • A modern converter runs hot or fails repeatedly — suspect the synchronous stage: recognize the second MOSFET as the rectifier and reason about gate drive, dead-time, and shoot-through, the failure modes synchronous rectification added over the simple diode (capacitor-and-inductor-failure-modes).
  • An output sags or oscillates under changing load — look to the control loop: a loop too slow or poorly compensated sags on transients or oscillates while the power stage is healthy, so read the ripple and transient response as the loop's signature before condemning a component (electromagnetic-interference-basics).

Verification & Testing Methods

Confirm your working model before moving on:

  • [ ] I can describe the switching cycle as energy stored in the inductor and released to the output, smoothed by the capacitor.
  • [ ] I can explain why a switching supply runs cool where a linear regulator runs hot.
  • [ ] I can read pulse-width modulation and relate the duty fraction of each cycle to the regulated output.
  • [ ] I can recognize a synchronous rectifier on a board and name the failure modes — shoot-through, gate drive, a second power device — it introduces.
  • [ ] I can reason about a control loop's health from the output's ripple and its response to a load transient.

Then try the practice exercises below — model-building and waveform reading only; scenarios differ from the quiz.

Practice Exercises

  1. Trace the energy engine (5 minutes, donor board). On a donor switching regulator, identify and label the switch, the inductor, the rectifier, and the output capacitor, and describe in one line how energy moves through them each cycle, so the engine is concrete on real hardware (switching-regulators-buck-boost-buck-boost).
  2. Reason the efficiency (5 minutes, desk work). Write why the same rail made by a linear regulator would run hot while the switching version runs cool, in terms of the switch's two states and the pass element's dissipative middle, so the efficiency is understood, not memorized (voltage-regulators-linear-and-switching).
  3. Find the second switch (5 minutes, donor board). On a synchronous buck converter, locate the high-side and low-side MOSFETs, identify which is the rectifier, and note the failure modes the second switch adds over a diode, so synchronous rectification is recognized on sight (capacitor-and-inductor-failure-modes).
  4. Read the loop from the output (5 minutes, oscilloscope on a low-voltage rail). Observe a rail's output ripple and its response to a changing load, and describe what each says about the control loop's speed and stability, so the invisible loop is read from its signature (electromagnetic-interference-basics).

These core steps — the traced engine, the reasoned efficiency, the found synchronous rectifier, and the loop read from the output — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • A switching supply is an energy-transfer engine — a switch banks energy in the inductor each on-time and releases it to the output each off-time, and the capacitor smooths the pulses to steady DC — not a resistive divider (switching-regulators-buck-boost-buck-boost).
  • It runs cool because the switch is fully on or fully off, dissipating almost nothing, while a linear regulator holds its pass element in the dissipative middle and burns the input-to-output difference as heat (voltage-regulators-linear-and-switching).
  • The throttle is pulse-width modulation — the fraction of each cycle the switch spends on sets the output — and a feedback loop turns that fraction continuously to regulate against input and load (electromagnetic-interference-basics).
  • The synchronous rectifier replaces the catch diode with a second MOSFET to recover its wasted energy, raising efficiency but adding failure modes — shoot-through, dead-time, gate drive, and a second power device (capacitor-and-inductor-failure-modes).
  • The control loop senses the output and corrects the modulation faster than any load can change, and because it leaves no single part to point at, its health is read in the output's ripple and transient response.

Skills Learned

After completing this section, you can:

  • Describe the switching cycle as energy stored in the inductor and released to the output.
  • Explain why a switching supply runs cool where a linear regulator runs hot.
  • Read pulse-width modulation and relate the duty fraction to the regulated output.
  • Recognize a synchronous rectifier on a board and name the failure modes it introduces.
  • Reason about a control loop's health from the output's ripple and transient response.

Glossary Additions

New terms introduced in this section:

  • synchronous rectifier — the second, actively switched MOSFET that a modern switching supply uses in place of the passive catch diode to carry the inductor's off-time current, switched on in complement to the main switch so that the current returns through its low on-resistance rather than a diode's forward-voltage drop, recovering most of the energy that drop would waste and raising efficiency at the high currents modern rails carry. Its repair significance is that it adds failure modes the simple diode circuit did not have: the main switch and the synchronous rectifier must never conduct at once, which would short the input to ground in a destructive shoot-through, so a small dead-time separates them, and a gate-drive fault, a failed synchronous MOSFET, or lost dead-time is a distinctly modern way for a converter to run hot, fail, or destroy itself.
  • pulse-width modulation — the control method by which a switching supply sets its output: the controller varies the width of the on-pulse in each switching cycle — the fraction of the cycle the main switch spends conducting, its duty fraction — because a wider on-pulse banks and delivers more energy per cycle and so raises the output, while a narrower one lowers it. It is the throttle every switching converter is steered by, adjusted continuously rather than set once, and in a regulated supply it is driven by the feedback control loop that senses the output and widens or narrows the pulse to hold that output steady against changes in input voltage and load current.
  • control loop — the feedback system that holds a switching supply's output steady: it senses the actual output through a divider, compares it to an internal reference, and adjusts the pulse-width modulation to correct any error, doing so continuously and far faster than any load change. Its design is a balance between speed and stability — a loop too slow or poorly compensated lets the output sag on a load step or oscillate, while a well-designed loop holds the output flat through changes in input and load. Because the loop is distributed across the controller and its feedback network rather than living in one component, its health cannot be read from a single part but shows instead in the output's ripple and its response to a load transient, which are the signals a technician reads to judge whether the regulation itself, rather than the power stage, is at fault.

Suggested Next Sections

Must read next:

  • PMIC Architecture in Modern Devices — Section 3.2 takes the single switching rail of this section and shows how modern devices integrate many of them: the power-management IC that packs multiple regulated rails, sequencing, and protection into one chip, and how to read its architecture on a board.

Recommended: