Screen Flicker and PWM Dimming: Who It Affects and How to Check Your Screen

Turn a phone’s brightness all the way down in a dark room and it just looks dimmer. On many phones, though, something else has started: the screen is now switching itself fully on and off hundreds of times a second. Most people never notice. Some say it gives them headaches or sore eyes within minutes, and a few can see it directly. This guide explains what is going on, what the research does and doesn’t show, and how to check your own screen.

It is general information, not medical advice. If you get headaches, eye pain or visual disturbances that don’t settle when you stop using a screen, see an optometrist, an ophthalmologist or your doctor.

What PWM dimming is

There are two ways to make a light dimmer. You can send less current through it, or you can switch it fully on and off very fast and change how long it stays on in each cycle. The second method is pulse-width modulation, PWM. Your visual system averages the pulses, so a light that is on 30% of the time looks dimmer than one that is on 90% of the time, even though each pulse is just as bright.

Two numbers describe it. The duty cycle is the share of each cycle the light is on. The PWM frequency is how many cycles happen per second, in hertz. It is not your refresh rate: a 120 Hz phone redraws the picture 120 times a second, and may pulse its pixels at 240, 480 or a couple of thousand times a second.

LCD monitors and laptops dim their backlight, and some still do it with PWM; others dim by current and are sold as “flicker-free”. OLED screens have no backlight. Every pixel is its own light, and most OLED phones use PWM over at least part of their brightness range.

Why OLED phones flicker most at low brightness

OLED pixels are hard to dim by current alone without the colors shifting, so phones pulse the panel instead and keep each pulse at a current the pixels handle well. The trouble is what the slider does to the duty cycle. DXOMARK, which films phone screens with high-speed cameras, describes minimum brightness as translating to around a 10% duty cycle: the screen is dark for most of every cycle, and the off interval is far longer than at high brightness.

That combination, full on to full off with long dark gaps, is the one that matters for comfort. A phone that is steady at 80% brightness can be pulsing hard at 10%, which is why complaints cluster around reading in bed with the lights off.

Who is sensitive

The best evidence that flicker causes symptoms comes from lighting, not phones. In a double-blind crossover study in real offices, Arnold Wilkins and colleagues replaced fluorescent lighting that pulsed at 100 Hz with 43 to 49% modulation with high-frequency ballasts that cut the modulation below 7%. Headaches and eyestrain were more than halved. The IEEE flicker standard, summing up this and later work, puts the effect down to a minority of people who are particularly affected, rather than to everyone a little.

Flicker too fast to see as flicker can still be seen indirectly. When your eyes jump from one point to another, a pulsing light lands on the retina as a row of separate copies, a “phantom array”. This is why PWM-driven LED car taillights, which pulse at a few hundred hertz, leave a dotted trail when you glance past them. Roberts and Wilkins found that observers could detect the effect at frequencies averaging about 2,000 Hz, falling to chance only around 3,000 Hz. A later study with a small, bright light put the average threshold near 6,000 Hz, and a few people saw it at 11,000 Hz. Those people also reported more visual discomfort in daily life.

For phones specifically, the evidence is thin:

A 2023 conference paper by Wu and colleagues compared OLED phones dimmed with PWM at 360 to 1,920 Hz, and with DC dimming. After about 35 minutes of viewing in the dark, it found no statistically significant difference in visual fatigue, measured by questionnaire or objectively. It is one short study whose summary doesn’t give the sample size.
A 2025 study in the Journal of the Optical Society of America A measured visibility, not symptoms. The stroboscopic effect of OLED PWM was visible at 480 and 960 Hz when the screen was below 40 cd/m², and not visible at 1,920 or 2,160 Hz.

The honest summary: flicker can cause headaches and eyestrain in sensitive people, and phone PWM at low brightness is real, measurable and visible to some. Nobody has yet shown in a solid trial that phone PWM causes symptoms, or how many people it affects. If you notice it, the lack of a big study doesn’t make your symptoms imaginary, and the fixes below are cheap to try.

The IEEE 1789 risk limits

IEEE Std 1789-2015 is the reference most people quote for “how much flicker is too much”. It was written for LED lighting, not screens, so using it for displays is a reasonable extrapolation rather than its intended use. It measures flicker as percent modulation, 100 × (max − min) / (max + min). A light that switches fully off between pulses, as an OLED does at low brightness, is at 100%.

From 90 Hz upwards, the standard draws two lines:

Region	Limit on modulation	Frequency where 100% modulation is within it
Low risk	below 0.08 × frequency, up to 1,250 Hz	1,250 Hz and above
No observable effect	below 0.0333 × frequency, up to 3,000 Hz	3,000 Hz and above

Below 90 Hz the limits are stricter: 0.025 × frequency for low risk and 0.01 × frequency for no observable effect. A worked example from the US Department of Energy’s guide to the standard: at 120 Hz, the low-risk limit is about 10% modulation.

So for a phone that switches fully off between pulses, the standard would put 480 Hz PWM outside the low-risk region, 1,250 Hz at the start of it, and about 3,000 Hz where no effect is expected. The standard has critics. A 2023 review in Lighting Research & Technology calls it controversial: too strict in places, not strict enough above roughly 400 Hz, where phantom arrays are most visible, and silent on duty cycle. Treat the limits as a sensible yardstick, not a medical threshold.

How to check your own screen

A web page cannot measure PWM. The browser has no access to how the panel is driven, and the pulses are far faster than anything a page can draw. Checking for PWM takes a camera, or a review that measured your model.

The camera trick

A phone camera reads its sensor line by line, so a light that pulses during the readout shows up as dark bands.

Fill the screen you are testing with white. The dead pixel test gives you a full-screen white field.
Set its brightness to where you use it at night.
Point a second phone’s camera at it. In a pro or manual mode, set a fast shutter speed, 1/1,000 s or faster, or record in slow motion.
Look for dark horizontal bands, or bands that roll in the slow-motion clip. Then raise the brightness a step at a time. Where the bands fade out is roughly where the phone stops using PWM.

DXOMARK shows the same banding with a 1,500 frames-per-second camera, and the Department of Energy’s guide notes that digital cameras sometimes pick up flicker. Two caveats: a phone camera shows that PWM is there and roughly where it stops, not its frequency or depth, and no bands is not proof of no flicker, because a high frequency or a slow shutter can hide it. For numbers, look for a review that publishes PWM measurements for your model.

Without a camera, the same guide suggests the phantom-array check: in a dark room, put a small bright dot on a black background, or wave a pen in front of the screen, and flick your eyes quickly past it. A trail of separate copies instead of a smooth smear suggests flicker.

Our flashing block and refresh rate test

The Flashing block on our ghosting test (mode 3) checks a different kind of flicker: whether the screen shows every frame it is sent. It turns a square black and white on alternate frames over a gray background. On a healthy screen at a steady rate, your eyes blend it into steady gray. Visible pulsing or a slow beat points to dropped or repeated frames, or a refresh rate that doesn’t match what the system is sending, and the refresh rate test tells you that rate and how steady it is. Neither tool measures PWM. Both can look perfect on a screen that pulses at 240 Hz.

The Flashing block is itself a strong flicker stimulus. Read the safety note at the end of this page before you use it.

What you can do

Look for a dimming setting. Some Android phones have an option called DC dimming, DC-like dimming, anti-flicker or high-frequency PWM dimming, usually under display or eye-comfort settings. Names and behavior vary by brand, and some cost a little color accuracy at the lowest brightness.
Keep the panel brighter and dim the picture instead. iPhone’s Reduce White Point (Accessibility, Display & Text Size) and Android’s Extra dim lower what the screen shows while the panel brightness stays higher. Whether that avoids PWM depends on the phone, so check with the camera trick.
Turn a light on. In a dark room you push the slider to the floor, which is exactly where PWM is deepest. A lamp lets you run the screen brighter without glare.
Check before you buy. If you know you are sensitive, look for reviews that measure PWM frequency and the brightness where it starts, and favor screens that stay flicker-free across the whole range or pulse at several thousand hertz.
On monitors, look for backlights dimmed by current. An LCD with PWM flickers harder at low brightness, just like a phone.

If your eyes still ache after you have ruled flicker out, the usual causes are more ordinary: see digital eye strain: what helps. And if you dim your phone for sleep, blue light, night mode and sleep explains why brightness matters more than color.

A separate risk: photosensitive epilepsy

PWM at hundreds of hertz is not what triggers photosensitive seizures. That risk comes from visible flashing. An Epilepsy Foundation of America review found 15 to 25 flashes a second most provocative, within a range of 1 to 65, and estimated that light-triggered seizures occur in about 1 in 4,000 people aged 5 to 24. Epilepsy Action estimates that 3 to 5 in every 100 people with epilepsy are photosensitive. Web accessibility guidelines (WCAG 2.3.1) say content should not flash more than three times in any one second unless the flashes are small or dim enough.

Some of our tests flash far faster than that:

The Flashing block on the ghosting test turns a large square black and white on every frame: 30 flashes a second on a 60 Hz screen, more on faster ones. It holds still behind a warning until you press “Start flashing test”. Don’t start it if you, or anyone who can see the screen, has photosensitive epilepsy.
In the effects gallery, Hypernova, Flux Core and Demoparty flash, and ask before they play. The benchmark’s Hypernova finale has a “Reduce flashing” option.
The burn-in test cycles full-screen colors every 3 seconds by default; at its fastest setting it changes the whole screen four times a second.

If a flashing image catches you without warning, Epilepsy Action’s advice is to cover one eye with your hand and turn away, rather than closing your eyes. Then close the test with Esc or the close button.