What Is PWM? Duty Cycle and Frequency Explained

Swipe sideways to compare the decision conditions.

This guide uses ideal two-level, active-high signals unless stated otherwise. The worked calculations are teaching examples, not recommended settings or measured customer results. Analog Devices’ PWM learning activity provides a useful introduction to the underlying waveform.

How do duty cycle, frequency and pulse width relate?

One period, T, contains a high interval and a low interval. For an active-high input, the high interval is the active pulse. Keep all time quantities in the same unit when calculating duty.

T = t_HIGH + t_LOW
D = t_HIGH / T  ·  Duty (%) = 100 × D
f = 1 / T  ·  t_HIGH = D / f

D is a fraction from 0 to 1. Frequency f is measured in hertz, or cycles per second. A 1 kHz signal has a 1 ms period; at 50% duty, it stays high for 0.5 ms and low for 0.5 ms.

Swipe the table sideways to compare all timing values.

How to use the table: compare the two 50% rows. The fraction is unchanged, but doubling frequency halves the pulse width. That matters when a receiver cannot reproduce very short pulses.

How do polarity and 0% or 100% endpoints change the reading?

For an active-low input, the LOW interval may be the commanded active time. A scope’s positive-duty reading can therefore be the complement of the device’s control duty.

Ideal 0% is continuously LOW; ideal 100% is continuously HIGH. Neither has repeating visible edges, so a frequency counter may report no valid frequency even if the timer is running. Real peripherals can use special endpoint settings, and some power stages require minimum off time. Verify continuous-on support instead of assuming it.

What does the PWM signal control after it leaves the source?

A microcontroller commonly generates PWM with a timer and compare value. Analog PWM can instead compare a control voltage with a repeating ramp. In either case, the resulting signal is a command; the receiving circuit decides what happens next.

1. CONTROLLERSets the pulse timingThe output pin supplies a logic-level waveform within its electrical limits.
2. RECEIVER / DRIVERInterprets or amplifies itA compatible input, gate driver or power stage handles the next step.
3. LOADProduces the useful resultCurrent, light, movement or heat depends on the complete circuit.

A motor winding and its recirculation paths shape current. A filter smooths voltage. A heater responds through thermal inertia. A dedicated fan-control input can decode duty without having its supply voltage chopped.

Switching can be efficient because a transistor spends much of its time conducting with a small voltage drop or blocking with little current. Conduction, transitions and gate drive still cause losses. PWM is a control method, not a promise of lossless operation.

Does 50% PWM mean half voltage or half power?

It means half the cycle is active. What that implies for voltage or power depends on where you measure and what the load is.

When is average voltage the useful PWM quantity?

For fixed high and low levels, the time-average voltage is:

V_AVG = D × V_H + (1 − D) × V_L

A 0-to-5 V waveform at 40% duty averages 2 V. It still switches between 0 V and 5 V. To obtain an approximate analog level, a low-pass filter must suppress the carrier.

The Analog Devices ADALM1000 activity demonstrates reconstruction with an RC filter. Stronger smoothing trades lower ripple for slower response; loading can also shift the result. Check ripple, settling, output impedance and startup behavior. Use buffering or a dedicated DAC when the application requires it.

When does RMS voltage determine PWM heating?

For ideal unfiltered PWM switching directly between 0 V and V across a constant resistance R:

V_RMS = V × √D
P_AVG = D × V² / R

Root-mean-square (RMS) voltage describes the equivalent DC heating effect in that resistor. The expression comes from averaging the squared voltage, not squaring the average voltage.

This calculation is not a heater design recommendation. Resistance changes with temperature, pulse ratings and thermal conditions matter, and inductive motors, current-regulated LEDs and filtered converter outputs require different waveform analysis.

How do you choose a usable PWM frequency?

Start with the receiving device’s allowed range. Then check the load response, minimum pulse times, switching losses, acoustic behavior and electromagnetic interference (EMI). There is no best number for every PWM application.

What changes when PWM frequency increases?

It can reduce some current ripple or ease filtering, but more transitions per second can increase switching-loss power. A fixed delay or dead time also occupies a larger fraction of a shorter cycle.

TI’s motor-control guidance describes the balance between current harmonics, acoustic noise, switching losses and dead-time distortion. Moving a carrier out of an audible range does not guarantee a silent mechanism.

Dead time is a non-overlap interval between complementary power switches, used where required to prevent simultaneous conduction. Do not remove it just to regain duty range. Follow the driver’s timing requirements and verify the effective waveform.

What changes when PWM frequency decreases?

Longer periods give a slower driver more time and, at the same timer tick rate, provide more timing increments per cycle. But they can bring greater ripple, visible artifacts or audible effects. Test the lowest intended command, startup and changing load; a plausible multimeter reading alone does not establish good operation.

How do LEDs, motors, servos and converters interpret PWM differently?

How should LED PWM dimming be validated?

In a suitable constant-current LED driver, PWM dimming turns regulated current on and off. Average current can approximately track duty if each pulse reaches the intended current and transition effects are small. Perceived brightness and camera appearance need separate checks.

At low duty, turn-on delay and current settling can consume much of the pulse. Check actual LED current, optical artifacts and camera banding in the intended use. A universal “flicker-free” frequency would ignore the driver, waveform and observer or camera conditions.

Why does 50% duty not guarantee 50% motor or fan speed?

Motor current depends on the winding, supply and drive circuit. Speed also depends on load, friction, back electromotive force and feedback. If speed must remain stable under changing load, evaluate closed-loop control.

Check what OFF means: a driver may brake, coast or recirculate current differently. For a fan with a dedicated PWM input, follow that interface specification. Chopping supply power is not automatically equivalent to using the control input.

Why must a hobby servo be specified by pulse width?

Pololu’s detailed servo-interface explanation identifies pulse duration as the central command quantity. About 1.5 ms is a common neutral reference, but exact limits and calibration vary.

Illustrative comparison: 1.5 ms repeated every 20 ms is 7.5% duty. Keep 7.5% duty but shorten the period to 10 ms, and the pulse becomes 0.75 ms—a different command, potentially outside the allowed range.

For a servo, specify pulse duration and permitted update rate, not a duty percentage alone. Do not assume a particular pulse always means the same angle across different servo models.

What must be checked for converters and heaters?

A switching converter’s output depends on its topology, operating mode and feedback loop. Do not apply one converter’s duty relationship to every design. Thermal loads may tolerate slower switching, but the switch technology still matters: a mechanical relay is not a substitute for a transistor switching at kilohertz rates.

How does PWM frequency limit timer resolution?

A timer can place edges only at available clock instants. For a simple edge-aligned counter with an effective tick rate f_TICK and N timing intervals per cycle:

f_PWM = f_TICK / N
Duty increment ≈ 1 / N

N is a count of intervals, not a universal register value. Prescalers, counting modes and endpoint behavior vary. Microchip AN539 explains the frequency-resolution trade-off for its example device; use your controller’s manual for implementation.

Also check shared timer resources, synchronized updates, complementary outputs and fault handling. Edge-aligned and center-aligned modes place edges differently and can require different frequency formulas. Microchip’s PWM peripheral overview illustrates why a nominal bit-depth label is only part of the selection.

How should you measure and troubleshoot a PWM signal?

An oscilloscope separates pulse width, period, amplitude and edge behavior. A meter’s DC reading may resemble the average, but it cannot by itself show missed pulses, ringing or a receiver that never recognizes HIGH.

Start at the receiving pin. Confirm high and low levels, period and active pulse duration. Then compare the observed duty with the command, accounting for polarity.

Check the driver output and load current next. A correct-looking logic input cannot prove that short pulses survive propagation delay or that the load receives the intended current.

Measurement safety: never defeat an oscilloscope’s protective earth to probe a floating switch node. Use an appropriately rated differential or isolated measurement system within its voltage, common-mode, category and bandwidth limits. High-voltage and motor-drive measurements require suitable training and procedures. See Tektronix’s probe primer.

Swipe the troubleshooting table sideways for the next check.

Use symptoms as starting points, not diagnoses. Record supply voltage, load, frequency, duty, temperature and measurement location. Include startup, shutdown and duty endpoints. A no-load waveform may not explain a failure during motor startup.

What must engineers and buyers specify before ordering a PWM-capable part?

For a controller, driver, fan or LED module, “supports PWM” is only the beginning. Put the following requirements beside the full manufacturer part number when reviewing a quotation or alternative.

1. Which kind of PWM interface?A logic command, a switched power output and a pulse-width servo interface are not equivalent.
2. Which electrical levels and polarity?Confirm HIGH/LOW thresholds, input tolerance, reference connection and input structure.
3. Which usable timing range?Specify frequency, minimum active and inactive times, required duty range, and 0%/100% behavior.
4. Which load and power limits?State the supply, current, voltage, thermal conditions and required current recirculation or protection.
5. Which resolution and update behavior?Check effective duty steps at the chosen frequency, shared timers and synchronized changes.
6. Which startup and fault response?Agree on enable behavior, overcurrent protection, shutdown and the safe state if the command disappears.

A higher maximum frequency or larger advertised resolution does not establish compatibility. Obtain engineering approval before treating an alternative as a drop-in replacement. YURUNOX’s purchasing process and quality-assurance information help frame the separate sourcing and documentation requirements.