Switching Power Supplies

Pick up your phone charger. Look at a laptop power brick. Examine the wall adapter for your Wi-Fi router. Every one of these is a switching power supply — a device that converts electrical energy with dramatically higher efficiency than a traditional linear supply by rapidly switching a transistor on and off rather than dissipating excess voltage as heat. Switching power supplies (also called switch-mode power supplies, or SMPS) now dominate consumer electronics because they are small, light, efficient, and inexpensive to manufacture. Most modern computers, televisions and audio amplifiers use them internally as well.

For anyone building or repairing electronics, switching power supplies are a double-edged sword. Their efficiency advantages are real and valuable. But every SMPS generates broadband electrical noise as a direct consequence of its operating principle, and a cheap or poorly designed SMPS can put audible hiss and whine into an audio system, add noise to sensitive measurements, and interfere with nearby radio, TV and wireless equipment. Understanding how switching supplies work is essential both for using them intelligently and for keeping their noise out of your circuits.

Why Switching Instead of Linear?

A linear voltage regulator wastes energy as heat. The pass transistor sits between the input and output, and it must drop a voltage equal to (V_in − V_out) while passing the full load current. Every watt of power dropped across the transistor becomes heat. If you are regulating from 18 V down to 13.8 V at 10 A, the transistor dissipates (18 − 13.8) × 10 = 42 W. That 42 W must be removed by a heatsink, which requires a large chunk of aluminum, good airflow, and accounts for significant size and weight. The efficiency of this linear supply is only (13.8 × 10) / (18 × 10) = 76.7%.

Now consider a car alternator charging at 13.8 V, supplying 10 A: it delivers 138 W to the load. The linear supply takes 180 W from the transformer, wastes 42 W as heat, and delivers 138 W — efficiency 76.7%. A switching supply achieving 90% efficiency takes only 153 W from the transformer and wastes only 15 W as heat. The heatsink can be much smaller. The transformer can be much smaller. The entire supply is lighter and cheaper to manufacture.

At higher input voltages, the advantage becomes even more dramatic. A 5 V computer supply derived from 120 V AC line voltage through a linear regulator would have an efficiency of roughly 5/170 = 3% — totally impractical. Switching supplies handle this ratio efficiently. This is why switching topologies are used in virtually every AC-to-DC adapter manufactured today.

The Fundamental Principle

The key insight behind all switching power supplies is the behaviour of an inductor when a voltage is applied and then removed. When you connect a voltage across an inductor, current rises linearly. The inductor stores energy in its magnetic field. When you disconnect that voltage, the inductor's magnetic field collapses and it generates a voltage to maintain current flow — the "inductive kick." This energy can be directed into a load instead of being wasted.

The switching supply uses a transistor (usually a MOSFET) to rapidly connect and disconnect the input voltage from the inductor. During the on-time, energy flows from the input into the inductor's magnetic field. During the off-time, the inductor releases its stored energy to the output. The output capacitor smooths the current pulses into a steady DC voltage. The ratio of on-time to the total switching period — the duty cycle D — controls how much energy is transferred per cycle and therefore sets the output voltage.

The higher the switching frequency, the smaller the inductors and capacitors needed to store and smooth the energy pulses. This is why higher switching frequencies allow smaller, lighter power supplies. Modern switching supplies typically operate at frequencies between 20 kHz (just above audio range, to avoid audible whine) and 1 MHz or higher (ultracompact laptop adapters). Higher frequencies mean more switching transitions per second, which means more high-frequency interference generated per unit time.

SMPS block diagram: the PWM controller adjusts the duty cycle in response to output feedback, maintaining constant output voltage regardless of input variations or load changes. Every switching transition creates high-frequency noise.

View Larger

The Buck (Step-Down) Converter

The buck converter is the simplest and most common switching topology. It always produces an output voltage lower than its input voltage — hence "step down." You will find buck converters in computer motherboard voltage regulators (converting 12 V to 1.2 V for the CPU), in solar charge controllers, in USB charging adapters, and in many battery-operated gadgets.

Circuit Elements and Operation

The buck converter uses four passive components plus the switching transistor: a series switch (MOSFET Q), a freewheeling diode D, an inductor L, and an output filter capacitor C. The load R_L connects across the capacitor.

Switch ON phase: When the MOSFET turns on, current flows from V_in through the inductor to the output capacitor and load. The inductor current ramps up linearly because the voltage across it is (V_in − V_out). Energy is stored in the inductor's magnetic field. The freewheeling diode is reverse-biased and carries no current.

Switch OFF phase: When the MOSFET turns off, the inductor cannot allow its current to stop instantaneously. It generates a reverse voltage to maintain current flow, forward-biasing the freewheeling diode. Current now flows from the inductor through the freewheeling diode, continuing to supply the output capacitor and load. The inductor current ramps down as energy is released. This phase ends when the MOSFET turns on again.

The average output voltage is determined by the duty cycle:

The PWM controller measures the output voltage with a feedback divider and automatically adjusts D to compensate for changes in V_in or load current. If the output tries to drop below 13.8 V, the controller increases D to pass more energy per cycle. If the output tries to rise, D is reduced. This feedback loop is what makes the output regulation precise.

Buck converter operation: during switch-on, energy builds in the inductor. During switch-off, the inductor delivers stored energy to the output through the freewheeling diode. Output voltage = duty cycle × input voltage.

View Larger

The Boost (Step-Up) Converter

The boost converter does the opposite of the buck — it produces an output voltage higher than its input voltage. This seems counterintuitive, but it follows directly from the inductor's energy-storage behaviour.

How Boost Conversion Works

Switch ON phase: The MOSFET switch connects the inductor directly from V_in to ground. Current builds in the inductor from the input source. The output diode is reverse-biased, so the output capacitor supplies the load on its own during this phase.

Switch OFF phase: The MOSFET turns off. The inductor now has current flowing through it and nowhere to send it — so it generates a voltage in the same direction as V_in. The total voltage at the top of the inductor becomes V_in + V_L, which can greatly exceed V_in. This higher voltage forward-biases the output diode, driving current into the output capacitor and load at the boosted voltage.

The boost converter's output current is lower than its input current (conservation of energy: if voltage doubles and efficiency is 100%, current halves). At 90% efficiency boosting 12 V at 5 A to 24 V: output current ≈ 12 × 5 × 0.90 / 24 ≈ 2.25 A. This must be considered when designing boost converters for loads requiring significant current.

The Flyback Converter (Isolated)

The buck and boost converters share a direct electrical connection between input and output — they are non-isolated. This is fine for battery-to-battery or DC-to-DC applications, but when converting from AC line voltage to low-voltage DC (as in a phone charger or laptop adapter), isolation between the high-voltage AC input and the low-voltage DC output is essential for safety and to meet regulatory requirements.

The flyback converter achieves isolation by using a transformer instead of a simple inductor. The operating principle is similar to a buck-boost converter: during the switch-on phase, energy is stored in the transformer's magnetizing inductance (like an inductor). During the switch-off phase, the stored energy is transferred through the transformer to the secondary winding and thence to the output through a diode and capacitor. The transformer turns ratio sets the relationship between input and output voltages.

Because there is no direct electrical connection between primary and secondary, the output is completely isolated from the AC input. This is the safety isolation that protects you from electrical shock when you touch the output terminals of a power adapter. Regulatory agencies (UL, IEC, CE) require this isolation in any consumer product that derives low-voltage DC from line voltage AC. The flyback converter is the dominant topology in phone chargers, laptop adapters, and all AC-DC wall warts.

A key practical point: in a flyback converter, the voltage across the switch at turn-off is V_in plus a voltage spike caused by the transformer's leakage inductance. This spike can easily reach 400–600 V even in supplies designed for 120 V AC input. The MOSFET switch must be rated for this voltage — typically 600 V for 120 V AC applications and 900 V for universal 85–265 V AC input designs.

PWM Control and Voltage Regulation

All three topologies above require a control circuit that adjusts the duty cycle in response to output voltage changes. This is accomplished by a pulse-width modulation (PWM) controller IC. The operation is straightforward:

1. A feedback voltage divider samples the output voltage and feeds it to a comparator inside the PWM controller.
2. The comparator compares the feedback voltage to a precise internal reference (typically 1.25 V or 2.5 V).
3. If the output is too low, the comparator increases the duty cycle — more on-time, more energy per cycle, output rises.
4. If the output is too high, the duty cycle decreases — less on-time, output falls.
5. The system reaches equilibrium when the error is zero — output voltage equals the setpoint.

The response speed of this feedback loop determines the transient response of the supply — how quickly it recovers when the load changes suddenly. A fast loop (high bandwidth) recovers in microseconds. A slow loop may allow the output to droop for milliseconds, which can cause problems for processors or sensitive analog circuits that need stable supply voltage during transients.

Common PWM controller ICs include the UC3842, UC3844, SG3525, and TL494, all of which have been in continuous production since the 1980s. Modern compact designs increasingly use single-chip solutions where the controller, MOSFET driver, and sometimes the MOSFET itself are integrated into one package.

Typical switching frequencies range from 20 kHz (the limit of human hearing — below 20 kHz, the inductor can audibly whine) to 1 MHz and beyond. In laptop adapters, frequencies of 65–130 kHz are common. In very compact USB-C chargers, frequencies up to 1 MHz or more allow the inductors to shrink to a few millimeters in height.

Efficiency Advantage: Worked Numbers

The contrast between linear and switching efficiency is most dramatic in high-power, wide input-output voltage difference applications. Here is a direct comparison for a 13.8 V, 10 A supply:

The 15 W difference between linear (42 W loss) and SMPS (15 W loss) is 27 W — permanently. Over an eight-hour working day: 27 × 8 = 216 Wh = 0.216 kWh saved. Over a year of daily use: roughly 79 kWh saved. At US electricity prices around $0.15/kWh, the SMPS saves about $12 per year in electricity — and produces a much cooler workbench.

For portable and battery-powered equipment, the efficiency difference is even more critical. Running a 20 W load from a 12 V LiFePO4 battery: a 90% efficient switching supply gives you 90% of your battery capacity, while a 70% efficient linear supply gives you only 70%. For a 30 Ah battery, that difference is 6 Ah of usable capacity — several extra hours of run time for a camping light, a portable PA system or a remote sensor station.

EMI — The Price of Switching

Every switching transition in an SMPS creates a brief but intense electromagnetic disturbance. When the MOSFET switch turns on or off, the current and voltage change at rates of millions of amperes per second. These rapid di/dt and dv/dt transitions generate broadband high-frequency energy that radiates from the wiring and components inside the supply. This energy appears across a wide frequency range — from the fundamental switching frequency all the way up through hundreds of megahertz — because fast square-wave transitions have harmonics extending to very high frequencies.

Two Types of Interference

Conducted interference travels along the power leads connecting the SMPS to your equipment. The noise generated inside the supply flows out through the DC output leads and into anything connected to them — an audio amplifier, a microcontroller board, an analog sensor front end. It also flows back up the AC input lead and into the electrical wiring of your house, where it can reach any other equipment plugged into the same circuit.

Radiated interference is emitted directly from the SMPS as electromagnetic waves. The switching waveforms on the internal wiring and PCB traces act as small transmitting aerials. Every loop of wire carrying pulsing current radiates. The high-frequency harmonics can couple into nearby cables, high-impedance inputs, and the radio, TV and wireless receivers around the house.

The Frequency Pattern

The interference from a switching supply appears as a series of noise peaks spaced exactly at the switching frequency. A 100 kHz switching supply generates peaks at 100 kHz, 200 kHz, 300 kHz, and so on — 300 potential interference peaks between 0 and 30 MHz alone. The amplitude of the harmonics generally decreases with frequency, but the pattern can extend surprisingly high — a poorly filtered 100 kHz SMPS can still produce measurable noise at 21 MHz (the 210th harmonic) and beyond.

A switching supply at 500 kHz produces harmonics at 500 kHz intervals — 60 peaks across 0–30 MHz. Higher switching frequencies allow smaller inductors and capacitors, but they also push more energy into the higher frequencies where wiring radiates most efficiently, so good layout and filtering matter even more.

Switching noise does not only show up as radio interference. If the supply's output ripple or its harmonics fall within the audio range, or mix down into it, you hear them as a whine or hiss from an amplifier. On an oscilloscope, they appear as fast spikes riding on the DC output at every switching edge. In a measurement circuit, they appear as jitter in the readings.

EMI Filtering and EMC Compliance

Switching supplies sold in most markets must meet electromagnetic compatibility (EMC) limits for unintentional emitters — for example FCC Part 15 Class B in the United States and CISPR 32 / EN 55032 Class B in Europe. These limit the conducted emissions on the AC input leads and the radiated emissions from the device. The tests use a calibrated measuring receiver and a defined test setup. Passing Class B means the supply is unlikely to cause harmful interference in a normal residential environment.

However, compliance limits are a compromise. A supply that passes with a comfortable margin may still produce noise that is clearly visible on sensitive measuring equipment or audible in a high-gain audio circuit on the same bench. Some supplies that bear compliance marks are counterfeits or do not actually comply — enforcement against imported consumer goods is limited.

Quality SMPS designs include multiple stages of EMI filtering: common-mode chokes on the AC input leads, X-capacitors across the AC input, Y-capacitors from each AC line to the chassis ground, shielded transformers, and additional output filtering. A properly designed and manufactured SMPS can be nearly as quiet as a linear supply — the supply itself is not inherently noisy, but cheap, cut-down designs omit the filtering to reduce cost.

Identifying SMPS Noise on Your Bench

If a circuit has unexplained hiss, whine or spiky readings, or a nearby radio picks up a pattern of equally-spaced noise peaks, the systematic approach to identifying the source is straightforward:

1. Look at the supply rail on an oscilloscope with the probe's short ground spring rather than the long ground lead, which picks up noise of its own. Switching noise shows as sharp spikes at the switching frequency, often with ringing after each edge. Note the spike amplitude and spacing.
2. Unplug switching supplies one at a time and observe whether the noise drops. Start with anything near the circuit: USB chargers, laptop adapters, LED light drivers (many LED lights use SMPS internally), monitor power bricks, phone chargers, router power adapters. Unplug each one and observe the noise level. You will quickly identify the culprit or culprits.
3. Swap the circuit's own supply for a battery or a known-quiet linear supply. If the noise disappears, the circuit's supply is the problem; if it remains, the noise is reaching the circuit from somewhere else — through signal cables, a shared ground, or radiation.
4. Measure the spike spacing. If spikes are 10 µs apart, the switching frequency is 100 kHz; if 15 µs apart, about 65 kHz. A portable AM radio tuned between stations and held near each suspect is another quick way to find a noisy supply — the buzz rises sharply as you get close.

Once you have identified the source, your options are: replace it with a quieter supply, move it physically farther from the sensitive circuit and its cables, add ferrite clamp-on cores to its leads to reduce conducted noise, add an LC filter on its output, or use a shielded enclosure. Adding a ferrite clamp (Fair-Rite type 31 material, or the split-core snap-on type) with multiple turns of the cable through it is often highly effective and costs only a few dollars.

Choosing Between Linear and SMPS

There is no single correct answer — the right choice depends on the application, power level, and how much you value efficiency versus noise performance.

Recommendation for noise-sensitive work: For audio preamplifiers, precision measurement, sensor front ends and anything you are trying to debug for noise, a quality linear bench supply is the right choice. It is heavy and inefficient but it is essentially silent. A well-filtered SMPS from a reputable manufacturer, followed by a linear post-regulator or an LC filter, is a good compromise when efficiency also matters.

Recommendation for portable or high-power use: A quality SMPS makes more sense for portable, battery-powered and high-power equipment because of weight and efficiency. Choose one from an established manufacturer with genuine EMC compliance, and avoid generic open-frame units or cheap imports from unknown makers.

For laptop, phone, and peripheral power around the bench: Physically separate chargers and adapters from sensitive circuits and their signal cables as much as possible. Add ferrite clamp-on cores to power leads near the work area. If you can, feed the bench from an AC circuit that does not also feed computer equipment.