Why do LED lights flicker? On a real site there are seven answers, and six of them are in the installation rather than the lamp. The symptom usually narrows the cause within a minute: a flicker that only appears below 20% is a dimmer-driver mismatch, a flicker that starts when the kettle goes on is a supply or neutral problem, and a flicker that only exists on video is a PWM frequency problem. This page names each cause, gives the number that decides it, and ends with a five-step order that finds the culprit without swapping every fixture in the room.
1. Flicker vs shimmer vs strobe: naming the problem
Four different faults get called “flicker”, each with a different fix. Naming yours saves a week.
Flicker is continuous modulation of light output at a steady frequency — usually 100 or 120 Hz mains ripple, sometimes higher. Shimmer is low-depth modulation you notice at low dim levels, typically under 10% output, where the driver is working near its stability limit. Strobe is deep, low-frequency modulation, often below 20 Hz, which reads as a pulse rather than a wobble. Stroboscopic effect is not flicker at all: the light is steady, but a moving object appears to jump, which is what happens under a high-frequency source at a short camera shutter.
There are also two faults that are not modulation. A dropout is the driver restarting, so the light blinks once every few seconds. A flash on switch-on is inrush charging the driver’s capacitor through the dimmer. Both look like flicker to a guest and neither is fixed by changing the driver’s PWM setting.
Three questions eliminate half the list before you touch anything: does it flicker at 100% or only when dimmed, does it happen with the switch off, and does it appear only on camera. Flicker at 1% with no camera points at the driver; a glow with the switch off points at leakage; a band on video points at frequency.
2. Cause 1: dimmer and driver mismatch
This is the most common cause and the easiest to prove. A phase-cut dimmer chops the mains waveform; the driver has to interpret what is left. Leading-edge (TRIAC) dimming was designed for a resistive filament, and it switches on mid-cycle into a capacitive load. The driver’s input capacitor charges in a short spike, the dimmer sees a current surge, and both ends oscillate. The result is a buzz you can hear and a flicker you can see, worst at the bottom of the travel.

Trailing-edge dimming switches off at the zero crossing and on at the start of the next half cycle, which suits a capacitive LED input far better. It is quieter and more stable, and it is what a retrofit should use. The catch is that the driver still has to be phase-cut capable: a trailing-edge dimmer on a driver with no phase-cut input will simply not dim, or will flicker at every level.
| Leading edge (TRIAC) | Trailing edge (MOSFET) | |
|---|---|---|
| Switching point | Mid-cycle, on | Zero crossing, off |
| Load it suits | Resistive, inductive | Capacitive, LED drivers |
| Typical artefacts | Buzz, inrush spike, dead band | Misfire at low load, heat |
| Minimum load | 10–20% of rating | 10–20%, often higher with LED |
| Depth reached in practice | 5–10% | 5–10% |
The test is cheap: put one fixture on a known-compatible trailing-edge dimmer on a bench. If the flicker goes, the dimmer was the problem. If it stays, move to cause 2.
3. Cause 2: minimum load and phantom load
A dimmer needs current through it to work. Below its minimum load it cannot hold the switching element stable, so it misfires at the zero crossing and the lamp stutters. The rule of thumb is 10–20% of rated dimmer load, and some designs want 25%. Four 9 W downlights is 36 W; on a 400 W dimmer that is 9%, which is under the floor and will flicker at low dim regardless of brand.
Phantom load is the mirror image, and shows up with the switch off. A two-wire dimmer has no neutral, so it powers its own electronics by leaking a small current through the lamp — typically under 0.5 mA. One lamp ignores it. Twenty lamps in parallel turn that leakage into a visible glow or a slow blink, because each driver’s input capacitor charges until it fires. The complaint — “the lights flash in the middle of the night” — is real and it is not a wiring fault.
Fixes, in order of cost: raise the load on the circuit, fit a dimmer with a neutral, or specify drivers with lower leakage. Adding one incandescent lamp used to be the standard trick; it still works, but it costs 60 W of heat forever and fails an energy model.
4. Cause 3: wiring and shared neutral issues
Shared neutrals produce the strangest symptoms. Two lighting circuits on one neutral is common in older buildings, and if the circuits are on different phases the neutral carries the sum. Voltage between the two circuit neutrals then appears across the driver’s input, and the light flickers or fails when the other circuit loads up.
The tell is timing. Flicker that starts when a lift or kitchen circuit operates is a supply-path problem, not a driver one, and the same applies when it vanishes on a temporary feed.
Long switch drops cause a related fault. A 30 m run of 1.5 mm² cable couples enough induced voltage onto the switched line to keep twenty high-impedance drivers half-awake, and a smart dimmer with no neutral behaves the same way. Check for that neutral at survey, not after the plaster is finished.
One safety note matters for hotel work: a borrowed neutral trips an RCD when the circuits are energised separately, and it is a fire risk in a way a flickering lamp is not. If the flicker comes with a tripping RCD, stop.
5. Cause 4: driver quality and ripple
Every LED driver rectifies mains, and unless the output is well filtered, the LED sees a DC level with 100 Hz (or 120 Hz) ripple on top. Cheap drivers save money on output capacitance, and the ripple depth climbs: a good driver holds percent flicker under 5% and a flicker index below 0.1, a poor one sits at 20–40%. The light looks fine at full output and turns obvious at 10%, where ripple is a larger fraction of a smaller average.
Ripple also shortens life: the die is driven above and below its average current thousands of times a minute, and the thermal cycling accumulates at the wire bonds. That is one reason a cheap retrofit fails at year three while the datasheet suggests ten.
6. Cause 5: PWM frequency and camera capture
If the light is fine to the eye and bands on video, the cause is PWM. A driver that dims by switching the LED on and off at 200 Hz is invisible to most people and unusable on camera, because a shutter at 1/50 s or 1/100 s samples the waveform at a different point in each frame. The result is rolling bands or a strobe effect in the edit.
Photographers and video crews have practical thresholds. Below 1 kHz, banding is likely at normal shutter speeds. Above 3 kHz, most cameras at 1/1000 s and most high-speed rigs are clean. Between the two, results depend on the camera and the shutter, which is why “camera safe” claims need a test rather than a logo.
Two workarounds exist and both cost something: slow the shutter, which the director may refuse, or change the driver for one above 3 kHz or a CCR type with no PWM at all. On a ballroom or press room, put the frequency in the fixture schedule and ask for a test clip at 1/1000 s before the event.
7. Cause 6: supply voltage and dimmer type
Drivers have a minimum operating voltage, and at low dim the margin gets thin. A circuit that sits at 207 V — within the ±10% band around 230 V — may be fine at full output and drop out at 5%, because the driver’s internal rail is already near its limit and the dimming stage pulls it lower. Long runs make it worse: 40 m of 2.5 mm² cable at 10 A loses about 2.5% of the supply, and that is before the joint resistance in a few old back boxes.
The dimmer type matters as much as the voltage. A dimmer rated for MLV (magnetic low voltage) loads has a different output stage from one rated for ELV (electronic) loads, and putting an LED driver on an MLV-rated dimmer is a mismatch even if the wattage adds up. Check the dimmer’s stated load type against the driver’s input type, not the total load alone.
Check the dimmer’s rating table for the word “LED”: a figure stated for incandescent is not a rating for this load. Measure at the fixture under load and again at 10% dim; more than a 5% drop from the board points at the cable.
8. Cause 7: incompatible lamps on one circuit
Mixing brands on one dimmer is the last cause and the one most often created by maintenance. A hotel replaces twelve failed lamps with a different brand because the original is discontinued; the new drivers have different input characteristics, different inrush and a different minimum load. The dimmer now sees a composite load it was never tested against, and the whole circuit flickers even though each lamp is individually fine.
The same happens when a circuit mixes LED with a leftover halogen, or when a phase-cut circuit has a 0–10 V fixture added by mistake: one driver oscillating pulls the supply down enough to make its neighbours oscillate, so the fault looks like it is everywhere. The fix is discipline, not electronics. Keep one driver family per dimmed circuit, record the model in the O&M manual, and replace failures with the specified type.
9. A five-step diagnostic order
Work in this order and you will not replace a driver that was never at fault.

Step 1 — reproduce and time it. Note the dim level, the time of day, and what else switched on. A flicker tied to another appliance is cause 3; a flicker tied to the dim level is cause 1, 2 or 5.
Step 2 — isolate the circuit. Put one fixture on one dimmer, on a bench or a temporary feed. If the fault disappears, the circuit is the problem. If it follows the fixture, the fixture is.
Step 3 — substitute the dimmer. A known-compatible trailing-edge dimmer costs less than a service visit and settles the mismatch question in minutes.
Step 4 — measure. Clamp the load current at 100% and 10%, measure voltage at the fixture, and record percent flicker with a meter sampling above 2 kHz. Numbers end the argument with the supplier.
Step 5 — escalate with evidence. Send the driver and dimmer models, the measured load, the flicker figures and a slow-motion clip. A supplier who sees that data replaces a driver; a supplier who sees “it flickers” asks you to check the wiring.
| Symptom | Most likely cause | First test |
|---|---|---|
| Flicker only below 20% dim | Dimmer-driver mismatch or minimum load | Bench test on a trailing-edge dimmer |
| Glow or blink with switch off | Leakage / phantom load | Measure leakage, count drivers on the circuit |
| Flicker when another circuit runs | Shared or borrowed neutral | Isolate and re-feed the circuit |
| Bands on video, fine to the eye | PWM frequency under 3 kHz | Slow-motion clip, then driver datasheet |
| Flicker at low dim on a long run | Voltage drop at the fixture | True-RMS measurement under load |
| One circuit, several brands | Mixed drivers on one dimmer | Standardise the driver model |
よくあるご質問
LEDのちらつきは危険ですか?
問題となるレベルでは、一部の人にとって危険です。3~70 Hzの変調は光過敏性発作と関連しており、15~20 Hzが最もリスクの高い帯域です。変調度の高い100 Hzのリップルは、勤務時間中の頭痛や眼精疲労との関連が指摘されています。変調度の低いちらつきは健康上の危険というより快適性の問題ですが、改善する価値はあります。
LEDライトが調光時だけちらつくのはなぜですか?
不具合の原因がランプではなく、調光経路にあるためです。最大出力時はドライバーが安定した状態で動作しますが、出力を下げると制御限界に近づきます。そのとき、調光器との不適合、最小負荷不足、または電圧降下があると、制御が不安定になる可能性があります。そのため、最初の試験ではベンチ上で10%と1%に設定して動作を確認します。
ドライバーの故障でちらつきが発生することはありますか?
はい。出力コンデンサーが劣化したドライバーは平滑化性能が低下するため、完全に故障する前にリップルが増加します。初期症状として、低い調光レベルでちらついたり、1秒に1回点滅したりすることがあります。以前は安定していた器具が、ほかに何も変えていないのにちらつき始めた場合は、配線より先にドライバーを疑ってください。
調光可能なLEDライトはすべてちらつきますか?
ドライバーが完全に滑らかな出力を実現することはできないため、どの製品にもある程度の変調はあります。違いは変調度と周波数です。優れたドライバーは、どの出力レベルでもフリッカー率を5%未満に抑えますが、性能の低い製品では10%設定時に20~40%に達します。箱に記載された「フリッカーフリー」は、測定による裏付けがない限り、単なるマーケティング上の主張です。
次のステップ
回路に不具合がある場合、最も早い解決方法は、実際に使用しているドライバーとの互換性を確認することです。 ドライバー・制御機器チーム が調光器の一覧、負荷数、 ダウンライトの設置計画 を確認し、設置前にちらつきが発生する組み合わせを特定します。これを基盤とする客室制御ロジックについては、 ホテル向け調光・スイッチングガイド でシーン設定とキーカード制御について解説しています。







