Switching frequency is the rate at which a power switch turns on and off per second, measured in hertz (Hz). In a power converter operating at 500 kHz, for example, the transistor flips between its on and off states 500,000 times every second. This single parameter shapes nearly every design trade-off in modern power electronics: the physical size of the converter, its efficiency, its heat output, and the quality of the voltage it delivers.
How Switching Frequency Works
Power converters don’t regulate voltage the way a dimmer switch does. Instead, they chop a DC input into rapid pulses using a semiconductor switch (typically a MOSFET), then smooth those pulses into a steady output with an inductor and capacitor. The number of on-off cycles per second is the switching frequency, and its inverse gives you the switching period. A converter running at 1 MHz has a switching period of 1 microsecond.
Within each cycle, the ratio of on-time to total period is called the duty cycle. By adjusting the duty cycle, the converter controls how much energy reaches the load. The switching frequency sets the canvas; the duty cycle paints the output voltage on it. Both together determine the ripple, regulation accuracy, and transient response of the supply.
This technique is called pulse width modulation (PWM). The key requirement is that the switching frequency be high enough that the load never “notices” the individual pulses. For an LED dimmed to 20% brightness, switching at 1 Hz would produce visible flicker. At 100 Hz or above, the eye perceives only a slightly dimmer light. Power supplies aimed at sensitive electronics need frequencies orders of magnitude higher still.
Why Higher Frequency Shrinks Components
The inductor and capacitor in a power converter act as an energy buffer, smoothing the pulsed waveform into a near-constant output. At low switching frequencies, each pulse lasts longer, so the buffer must store more energy per cycle. That demands physically larger inductors and capacitors. Raise the frequency and each pulse shortens, reducing the energy swing the filter components need to handle.
The relationship is direct. Output current ripple in a buck converter is proportional to 1/(L × f_sw): double the frequency and you can halve the inductance for the same ripple. Output voltage ripple follows a similar pattern, also scaling inversely with frequency. Texas Instruments notes that engineers choosing a higher switching frequency can select smaller filter inductors, less output capacitance, and even smaller input capacitors, shrinking the overall PCB footprint. At 1000 kHz versus 600 kHz, for instance, the faster loop response from smaller passive components also improves how quickly the converter reacts to sudden load changes.
This is why the trend across the industry has been toward higher frequencies. Smaller converters fit inside phones, laptops, and server racks where board space is expensive.
The Efficiency Trade-Off
Every time a transistor switches, it loses a small amount of energy. The average power burned in switching equals the energy lost per transition multiplied by how many transitions happen each second. Double the frequency, double the switching loss.
The loss comes from several places. During each turn-on and turn-off, there’s a brief moment when both voltage across the transistor and current through it are nonzero simultaneously. That overlap dissipates heat. The transistor’s internal capacitances also store energy that gets dumped as heat each cycle. And driving the gate of the transistor itself requires energy proportional to its gate charge times the switching frequency.
Below roughly 500 kHz, conduction loss (the heat from current flowing through the switch’s on-resistance) tends to dominate. Above 500 kHz, switching loss often becomes the larger term. This is the fundamental ceiling on switching frequency: at some point, the efficiency penalty from switching losses outweighs the size savings from smaller passives. Every converter design lands somewhere on this curve, balancing size against heat.
Heat and Thermal Limits
Switching losses convert directly into heat inside the transistor. At high frequencies, the extended time spent in transitional states increases the drain-to-source resistance, generating more heat per cycle. Parasitic elements in the circuit (stray inductances, capacitances in the package) worsen the problem by stretching switching transitions and adding ringing.
The thermal consequences are real. A MOSFET that runs cool at 200 kHz may need a heatsink or forced airflow at 1 MHz. In compact designs where airflow is limited, thermal constraints can force engineers to back off on frequency even when the semiconductor could theoretically switch faster. Reliable long-term operation depends on keeping junction temperatures well within rated limits, so thermal management and switching frequency selection go hand in hand.
Typical Frequency Ranges by Application
Different applications land at very different points on the size-versus-efficiency spectrum:
- High-voltage isolated supplies (AC adapters, LED drivers): typically 20 kHz to 150 kHz. These designs use silicon transistors and must stay below certain thresholds to meet electromagnetic interference (EMI) standards. Many are set below 75 kHz specifically for EMI compliance.
- On-board DC-DC converters (powering processors, memory, and board-level circuits): usually 200 kHz to 1 MHz. This is the workhorse range for most switch-mode power supplies inside consumer electronics and servers.
- Automotive power supplies: often targeted between 400 kHz and 500 kHz, or above 1.6 MHz, to avoid frequency bands regulated under automotive EMI standards like CISPR 25, which sets strict limits on emissions below 350 kHz and between 525 kHz and 1610 kHz.
- High-density point-of-load converters: the current mainstream is pushing between 3 MHz and 4 MHz, where the inductor and capacitor can be tiny enough to fit directly next to a processor.
How GaN Is Pushing the Limits
Traditional silicon MOSFETs hit practical frequency limits because their internal capacitances and switching characteristics produce too much loss at very high speeds. Gallium nitride (GaN) transistors change the equation. GaN devices switch faster and with less energy lost per transition, thanks to lower parasitic capacitances and the material’s ability to handle high voltages with smaller, faster structures.
In real-world designs, GaN transistors enable switching frequencies up to 500 kHz in multi-kilowatt converters (3.3 kW to 8 kW range), achieving power densities near 100 watts per cubic inch. That’s well beyond what silicon MOSFETs deliver at the same power levels. At lower power levels, GaN pushes into the multi-megahertz range. The result is converters that are dramatically smaller and lighter, which is why GaN chargers for laptops and phones have become noticeably more compact than their silicon predecessors.
EMI and Frequency Selection
A switching converter is, by nature, a source of electromagnetic noise. Every time the transistor flips, it generates harmonics that radiate from PCB traces and conducted noise that rides back onto the input power line. The fundamental frequency and its harmonics must fall within limits set by regulatory standards.
This is why frequency selection isn’t purely an engineering optimization problem. Designers often choose specific frequencies to dodge regulated emission bands. In automotive applications, the gap between 350 kHz and 525 kHz is a common target precisely because it falls between two regulated windows. In consumer electronics, staying below 150 kHz keeps the fundamental out of the conducted emissions test range defined by standards bodies. Higher-frequency designs above 1.6 MHz push harmonics into ranges where they’re easier to filter or where limits are more lenient.
The practical takeaway: switching frequency determines not just size and efficiency, but whether a product can pass the emissions testing required to ship.