Pwm

PWM, short for pulse-width modulation, is a technique for controlling average power delivered to a load by rapidly switching a signal fully on and fully off, varying only the fraction of time it spends on during each cycle — the duty cycle — rather than varying an analog voltage level the way a linear control approach would. Because the switch itself is either completely on or completely off at any instant, never sitting in a lossy, power-wasting in-between state, PWM delivers the efficiency advantage that makes it the default control technique anywhere a circuit needs to adjust delivered power, brightness, or voltage without burning the difference as heat. How PWM actually encodes a desired average power level into a switched digital waveform looks like this:

Desired average output level determined
A target brightness, speed, voltage, or power setting
Switching cycle, repeated at a fixed frequencyDuty cycle directly encodes the target average level
Signal driven fully on for a set fraction of the period
Longer on-time means a higher average delivered level
Signal driven fully off for the remaining fraction
Shorter on-time means a lower average delivered level
↻ next PWM period
Load's own inertia averages the switched signal
A motor's mechanical inertia, an inductor's current, or the eye's persistence

A few things are worth knowing about why PWM's on-off switching ends up looking continuous to whatever it's actually controlling:

PWM's entire trick depends on the load being unable to respond fast enough to notice the individual on-off transitions, only the resulting average — which is exactly why PWM frequency has to be chosen specifically to outrun whatever physical process is on the receiving end. A DC motor's mechanical rotor has real inertia and simply cannot speed up and slow down fast enough to track individual PWM pulses switching thousands of times per second, so it settles into a steady rotation speed set by the average power those pulses deliver. The human eye similarly can't perceive individual light pulses above roughly fifty to a hundred hertz, which is why PWM-dimmed LED lighting appears as a smooth, continuous brightness level rather than a visibly flickering light, provided the switching frequency is set comfortably above that perceptual threshold. If the PWM frequency is set too low relative to whatever's being controlled, the load does start to respond to the individual pulses rather than their average — audible motor whine, visible flicker, or mechanical juddering are all symptoms of a PWM frequency that's too close to whatever the controlled system can actually perceive or respond to.

Duty Cycle Sets the Average, Not the Instant three duty cycles, same peak voltage, very different delivered power 25% duty cycle low average power 50% duty cycle medium average power 75% duty cycle high average power same switching frequency, same on/off voltage — only the fraction of time spent high changes

Motor speed control is one of the most widespread real-world uses of PWM, because the technique lets a drive circuit adjust a motor's average delivered power with almost none of the wasted heat a linear, resistor-based speed control would generate. Driving a motor through a fully-on, fully-off switch and varying only the duty cycle delivers whatever average power corresponds to that duty cycle while the switching transistor itself dissipates very little power, since it's never sitting in the lossy, partially-on state a linear control circuit would require to throttle the motor's voltage directly. This is exactly why PWM dominates modern motor control, from cooling fans and electric vehicle drive motors to the tiny motors inside hard disk drives and camera autofocus mechanisms — the efficiency gap between switching and linear control widens dramatically as motor power increases, making PWM control essentially mandatory for anything beyond the smallest, lowest-power motors.

PWM is also the control signal that sets the switching duty cycle inside a buck or boost converter, which means PWM isn't just one specific application alongside switching power supplies — it's the literal mechanism that makes switching power conversion work at all. A buck converter's feedback loop continuously adjusts its switch's duty cycle to hold the output voltage on target, and that duty-cycle-controlled switching signal is PWM by definition — the same technique, applied to the same switch-and-inductor hardware, just with the specific goal of regulating an output voltage rather than directly setting a motor's speed or an LED's brightness. Understanding PWM as a general power-control technique, rather than a feature specific to any one application, makes clear why the same underlying concept shows up identically in motor drivers, LED dimmers, audio class-D amplifiers, and every switching voltage regulator — they're all the same core trick aimed at different loads.

Converting a PWM signal back into a genuinely smooth analog voltage, rather than relying on a load's own physical inertia to do the averaging, requires an explicit low-pass filter that removes the high-frequency switching content while passing the slower-changing average through. While a motor's mechanical inertia or the eye's visual persistence provides "free" averaging for those specific loads, applications needing an actual clean analog voltage — using PWM as a cheap, simple digital-to-analog converter, for instance — need a real resistor-capacitor or inductor-capacitor low-pass filter stage after the PWM output, smoothing the switched square wave into the genuinely continuous voltage the duty cycle represents. This filtered-PWM approach is a legitimate, commonly used low-cost alternative to a dedicated digital-to-analog converter chip in applications where speed and precision requirements are modest enough that a simple filter and a microcontroller's built-in PWM output can do the job adequately.

Read PWM as power control through timing rather than through voltage level: a signal switched fully on and fully off, with only the fraction of time spent on carrying the actual information, works precisely because whatever it's driving — a motor's inertia, an eye's persistence, a switching converter's output capacitor — naturally averages those rapid transitions into the smooth, continuous behavior the application actually needs, all while the switch itself wastes almost none of the power a linear, voltage-dropping approach would lose as heat.

Take pwm further

Ask the copilot about this term, or have our engineers assess it against your process.