Do LED Lights Cause Headaches, and How Do I Test Mine?

Usually flicker, and usually the driver rather than the diode. The reason people ask do LED lights cause headaches is that the light in the room is pulsing 100 to 120 times a second, deep enough for the visual system to register and fast enough that nobody sees it. The pulsing comes from the power supply built into the bulb, not from the LED chip, which is why two bulbs of the same wattage and the same color can feel completely different in the same fitting.

Which makes this a wiring problem before it is a health one. Film the bulb on your phone, work out whether the fault sits in the bulb or the dimmer, then check whether the fixture will let you change it at all. A screw socket lets you replace the driver by replacing the bulb; a sealed backlit panel does not. The Houseland macaron sconce is the first kind, with an E27 socket that standard US E26 LED bulbs fit and the bulb left as a separate choice at checkout.

Why do LED lights cause headaches when the room looks perfectly steady?

Because an LED cannot use mains power as it arrives. Mains alternates 50 to 60 times a second and an LED needs direct current, so every bulb carries electronics to rectify and smooth it. Waveform Lighting's flicker guide describes what a cheap version of that circuit does: the light collapses at each valley in the rectified signal, twice per cycle, so the bulb swings between bright and dim 100 to 120 times a second.

How far it falls in each swing is the number that matters, and the trade calls it percent flicker: the gap between the brightest and dimmest point of one cycle. A filament bulb sits near 10 to 20 percent, because the wire stays hot through the valley. An LED has no heat to coast on, so a weak driver takes the light most of the way to nothing and back a hundred times a second.

The problem is recognised enough to have a standard of its own. IEEE 1789-2015 is titled Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers, and DigiKey's engineering summary lists the effects as eyestrain, blurred vision and impaired performance on sight-related tasks, says flicker triggers migraines in about 10 percent of the population, and names children and people with migraine, autism or epilepsy as most at risk. Seizures, it adds, are not triggered above 70 Hz, which is why a 100 Hz bulb can be ordinary, legal and still ruining your evenings.

How do I test my own bulbs in two minutes?

Record two or three seconds of the lit bulb in your phone's slow-motion mode and watch the playback. Bands or strobing on a slow-motion clip mean real modulation, and the harder the bands are to miss, the worse the bulb is. Waveform Lighting recommends this instead of an oscilloscope or a light meter, which start at several hundred dollars.

1. Wait until dark and switch on one fixture at a time. Several sources average each other out and hide the offender.
2. Open the camera, switch to slow motion, point it at the lit bulb an arm's length away, and record 2 to 3 seconds.
3. Shoot video, not a still. Phone software now cleans the dark lines out of photographs, which helps photography and ruins diagnosis.
4. Film a battery flashlight as a control. It runs on DC with no mains ripple, so if that clip bands too, you are photographing your camera rather than your lighting.
5. Spin a pen between finger and thumb under the light. Steady light smears it into one blur; a pulsing light freezes it into a fan of separate copies.
6. Read for ten minutes with the lamp off to one side. The edge of your vision picks up flicker better than the centre, so a light can feel wrong while you look away.

The control clip is the step everyone skips, and the one that stops you buying a bulb you did not need: phone sensors read a scene line by line, so some settings band light that is genuinely quiet.

Flicker vs glare vs color temperature: which one is doing it?

Flicker is the one that survives the slow-motion clip, and the other two can be ruled out in one evening with your hand and a second lamp.

Glare and flicker get mixed up because both worsen as the room around the light darkens. The tell is that glare stops the moment you break the sight line to the bulb, while flicker does not care where you look. If a hand over the bulb fixes it, you need a shade rather than new electronics.

What do the flicker numbers on a spec sheet actually mean?

They are a frequency in hertz and a percent flicker, and IEEE 1789 ties the two together instead of setting one flat limit. DigiKey's summary of the standard gives the low-risk line as percent flicker below 0.08 times the frequency between 90 Hz and 1250 Hz, and unrestricted above 1250 Hz. The no-effect line is stricter: 0.0333 times the frequency between 90 Hz and 3000 Hz. Under 90 Hz both tighten to 0.025 and 0.01 times frequency.

Put a normal bulb through that. At 100 Hz the low-risk ceiling lands on 8 percent flicker and the no-effect ceiling near 3 percent, and at 120 Hz they move to about 9.6 and 4 percent. Waveform quotes the same 8 and 3 percent for a 100 Hz bulb, which checks the arithmetic. Three things follow that no box tells you:

• Beating a filament bulb is not the bar. Waveform puts an incandescent at 10 to 20 percent flicker and a flicker index of 0.04, so "as steady as a traditional bulb" is a claim already above the 8 percent low-risk line.
• Frequency buys forgiveness. A driver working above 1250 Hz falls outside the low-risk restriction, which is how a well-built LED can be deeply modulated and still bother nobody.
• A missing number is an answer. Waveform reads a controller with no published frequency as a sign it runs at a few hundred hertz at best, and the same goes for bulbs that print lumens, kelvin and watts but never percent flicker.

Which fixtures can you fix by changing the bulb?

Any fixture with a socket in it. Waveform is blunt here: with a screw-in E26 or E27 bulb the flicker lives in the bulb's own electronics, so curing it means replacing the bulb, which is the cheap end of this problem.

• Screw sockets and G9 capsules. The driver is inside the thing you are holding, so a better bulb is a better driver. The Houseland macaron sconce runs an E27 socket that standard US E26 LED bulbs fit, and the bulb is a separate line at checkout: 12W warm, 12W white, or none, if you already know what you want to sit under.
• Integrated panels and strip art. Emitter and driver are sealed into the piece, so there is nothing to unscrew and your only levers are the level you run it at and where you hang it. That is the real difference between a backlit panel and a run of LED strip.
• Anything on a remote or a wall dimmer. Test it at full output first. Clean at maximum and bad near the bottom means the driver has run out of headroom, the behaviour described in flicker at the lowest brightness step.

This is the part we can be exact about, because it is a socket question. The sconces in the Houseland wall lamps collection are socketed rather than integrated, E27 on the macaron design and G9 on the pull-chain globes, so the driver you live with is one you chose.

From the shop

Macaron Bedside Wall Sconce, Adjustable Wood Base Wall Lamp, 3 Shapes 13 Colors

An E27 socket that standard US E26 LED bulbs fit, on a head that swivels about 90 degrees, with the bulb chosen separately: 12W warm, 12W white, or none. Painted iron shade on a solid rubberwood base, in three shapes and 13 colors.

View the macaron bedside sconce — 13 colourways & current price

What if a new bulb doesn't fix it?

Suspect the dimmer next, and test with it out of the circuit. Waveform's guide points out that dimmers and remote controllers work by pulse-width modulation, switching the LED fully on and off and varying the ratio, so the controller can be the whole source of the pulsing while the bulb is innocent.

The figure to hunt for is the PWM frequency. Waveform recommends 25 kHz or higher for a genuinely quiet result, notes that IEEE 1789 treats 3000 Hz as fast enough to mitigate the effect, and gives a second reason to clear 20 kHz: hearing tops out near 20,000 Hz, so a supply switching above that cannot whine at you either. A controller that publishes no PWM frequency is usually running at a few hundred hertz. If your light is on a remote, settle it at one level and leave it, which is the argument in dimming from the remote.

And if a low-flicker bulb at full output with no dimmer in the loop still leaves you with a headache, the light is probably not the cause. This depends on your wiring, your bulbs and your own sensitivity, so change one variable at a time. Glare, a bright screen in a dark room and ordinary migraine triggers produce the same evening ache, and none are cured by buying lighting.

What I'd do

Fix one room rather than the house, and change one thing at a time. Take the room where the headaches happen and work down this list:

1. After dark, film each light in that room in slow motion, one at a time, plus a battery flashlight as a control.
2. Write down which fixtures hold a bulb you can unscrew. Those are the cheap fixes.
3. Buy one replacement bulb that publishes a percent flicker figure under 8 percent at 100 Hz, and fit it where you sit most.
4. Take the dimmer out of the loop and run that light at one fixed level.
5. Give it two weeks before judging. Triggers stack up over days, and one quiet evening proves nothing.
6. If nothing shifts, stop buying bulbs. Look at glare, screens and the rest of the evening, and take a persistent headache to a doctor rather than a hardware shop.

Shop wall lamps from Houseland

-20%Globe Wall Sconce with Switch, Pull Chain Glass Wall Lamp, 4 Shapes 4 Colorsfrom $104.46$130.58-20%Bear Nursery Wall Sconce, Adjustable Iron Wall Lamp with Wood Base, 8 Colorsfrom $97.89$122.36-20%Vanity Sconce with Exposed Globe Bulb, Adjustable Iron Wall Lamp, 5 Colorsfrom $72.22$90.28-20%Fluted Resin Hallway Sconce, Up Down Wall Light in 4 Colors, Two G9 Bulbsfrom $83.79$104.74Wood Flush Mount Ceiling Light, Walnut or Natural, Slim or Tall LED Fixturefrom $99.95

See all wall lamps →