LED Strip Flicker: PWM Frequency and Camera Banding

Key points

• Percent flicker below 10% at any frequency is the practical camera-safe threshold; above 30% bands appear at shutter speeds shorter than 1/100 s.
• IEEE 1789 sets two thresholds: no observable effect below 90 Hz requires flicker percent under 0.25% per 100 Hz step, and the low-risk limit above 1 kHz is generous — controller frequency, not the standard, is what saves you.
• A strip that flickers at 100% brightness usually has a power supply problem, not a PWM problem: check for 100/120 Hz ripple from a failing or oversized driver first.
• Specifying the PWM frequency in the RFQ is the buyer's job — most cheap dimmers ship at 400–800 Hz and fail on camera.

The short answer

Flicker on camera is a duty-cycle artefact: a PWM controller switches the strip fully on and fully off hundreds or thousands of times per second, and the human eye averages this out while a rolling shutter camera does not. The decision rule is simple — keep percent flicker below 10% for any filming, and keep PWM frequency at 2 kHz or above so the flicker that remains integrates away at every realistic shutter speed.

Criterion	Cheap dimmer (400–800 Hz PWM)	Broadcast-safe (≥ 2 kHz PWM, CC dimming)
Percent flicker	30–100%	< 10%
Visible at 30 fps	Rolling bands below 1/200 s	Clean
Slow motion (120 fps+)	Unusable	Clean to 1/4000 s
Dimming smoothness at low levels	Steps or jumps	Linear
Typical unit cost	$2–5	$15–40

The row most people get wrong is the first one. Buyers compare prices and assume all dimmers are equivalent because the strip lights up — the failure only shows up later, on the footage, where it costs a reshoot.

How the number is calculated

Percent flicker (IEEE 1789 terminology) is measured from a photodiode waveform:

percent flicker = 100 × (V_max − V_min) / (V_max + V_min)

A pure square-wave PWM at 50% duty has percent flicker of 100%, regardless of frequency — the frequency determines whether the camera resolves that flicker into bands. The camera resolution condition depends on shutter time versus PWM period:

t_shutter < T_PWM  →  bands appear (sampled mid-cycle)
t_shutter ≥ 2 × T_PWM  →  flicker integrates out

At 25 fps the fastest standard shutter is 1/2000 s. For that shutter to integrate cleanly, the PWM period must be at most 0.5 ms, i.e. 2 kHz. At 1/50 s (the 180° default), even 400 Hz integrates two full cycles — which is why flicker often “passes” on the office phone camera and fails on the real shoot.

Worked example: a signage lightbox filmed at 30 fps, 1/500 s shutter, on a 600 Hz dimmer. PWM period is 1.67 ms; the shutter samples only 30% of a cycle. Whether the frame catches the on-phase or off-phase is arbitrary per scan line — that is the rolling band. Raising the same dimmer to 2.4 kHz puts 4 complete cycles inside 1/500 s and the banding disappears with no other change.

When to use it — and when not to

Use PWM dimming at ≥ 2 kHz when the installation is filmed, appears in camera-framed retail spaces, or is dimmed across a wide range — PWM keeps colour temperature constant from 100% down to about 5%, because the LED current during the on-phase never changes.

Do not use PWM dimming when a cheap 400–800 Hz controller is the only option and footage matters — specify constant-current (CC) or DC (analogue) dimming instead. CC dimming varies the actual current, so percent flicker is near zero at any shutter speed, but colour drifts slightly below 20% brightness and each channel costs more to drive. For architectural projects with no cameras, 1–2 kHz PWM with percent flicker under 30% is adequate and cheaper; do not oversell broadcast specs into a warehouse aisle.

One more case PWM cannot fix: flicker at 100% brightness. At full duty there is no switching, so ripple at 100 Hz (50 Hz mains, EU/CN) or 120 Hz (60 Hz mains, US) comes from the driver — usually overload, ageing electrolytics, or missing PFC. Fix the supply; no controller change will help.

What to put in the specification

• PWM frequency per channel: ≥ 2 kHz (state the number; “flicker-free” without a frequency is not a spec)
• Percent flicker at 50% duty: ≤ 10%, measured per IEEE 1789 or ASSIST Metric
• Dimming curve: linear or gamma-corrected, smooth from 1% to 100% without visible steps
• Driver ripple: ≤ 5% peak-to-peak at rated load (this is the 100/120 Hz term)
• For slow-motion or high-speed work: ≥ 4 kHz PWM, or CC dimming, stated explicitly in the RFQ

Scope limit

This page covers constant-voltage strip dimmed by PWM or CC methods and the camera-visible consequences. It does not cover mains-voltage (AC COB) strip, where the 50/60 Hz rectified ripple is fixed by the product’s internal design — if that is the platform, ask the manufacturer for a driver-integrated flicker report instead. It also does not cover DMX/0–10 V architectural control topologies; for those, the flicker behaviour lives in the dimming rack spec, not the strip.

Verifying it on the sample

Three checks on the incoming sample, no lab needed:

1. Phone slow-motion test at 240 fps — 240 fps resolves up to roughly 1 kHz of PWM structure. Bands at 240 fps mean the frequency is below 1 kHz; a clean 240 fps feed with visible bands only at 1/1000 s still passes for normal video.
2. Percent flicker at 50% duty — point the photodiode of any lux meter with a fast output (or an oscilloscope plus photodiode) at the strip: (max − min)/(max + min) must be ≤ 10%.
3. Full-brightness ripple — dimmer bypassed, measure again. Any residual flicker here is the driver’s 100/120 Hz ripple; pass criterion ≤ 5% p-p. Reject the reel only if the strip itself adds flicker the driver does not — which, on a constant-voltage strip, it never does.

This article describes general manufacturing practice for machined and fabricated components. It is not a substitute for a drawing-specific engineering review — tolerances, material selection and process capability depend on the individual part.