Flicker is the rapid, repeated variation of light output, and LED systems can produce more of it than the incandescent lamps they replaced. The LED itself responds to current changes within microseconds, so whatever ripple the driver puts on the output appears directly in the light. That sensitivity makes flicker a driver quality question, and IEEE 1789 is the standard that turns it into numbers.
Where Flicker Comes From
Three sources dominate. Mains ripple: a driver with insufficient output filtering passes the 100Hz or 120Hz rectified line frequency into the LED current. Dimming method: PWM dimming at low frequencies or low duty cycles makes the light switch fully on and off, and the flicker depth follows the dimming depth. Interaction effects: some drivers misbehave with certain loads or at the edges of their range, producing oscillation rather than smooth regulation.
Incandescent lamps had thermal inertia that smoothed their output; LEDs have none. The same driver topology that was tolerable behind a glowing filament becomes visible and measurable behind a diode.
The Two Numbers
Percent flicker, also called modulation depth, is the difference between maximum and minimum output divided by their sum, expressed as a percentage. It describes how deep the variation goes. Flicker index, defined in the older IES handbook method, compares the area above the average output to the total area under one period. It describes the waveform shape, penalizing sharp on-off patterns more than gentle sinusoidal ripple.
| Metric | Formula basis | Strengths | Blind spot |
|---|---|---|---|
| Percent flicker | (max-min)/(max+min) | Simple, instrument-friendly | Ignores waveform shape |
| Flicker index | Area ratio per period | Captures shape and duty | Harder to measure, no frequency weight |
| PstLM | Short-term perception model | Standardized in IEC TR 61547-1 | Complex, needs full waveform |
Neither percent flicker nor flicker index accounts for frequency on its own, and frequency is where the harm lives.
What IEEE 1789 Says
The standard plots allowable percent flicker against frequency. Above roughly 1250Hz, flicker at any depth falls in the no-effect region. Between about 90Hz and 1250Hz, a low-risk boundary allows deeper modulation as frequency rises: roughly a percent-flicker limit near 8 percent at 100Hz, rising linearly with frequency in the low-risk region. Below 90Hz the risk boundaries tighten sharply, and visible flicker becomes the dominant concern.
Practical driver conclusions follow directly. Output ripple at twice the line frequency must be filtered or the depth constrained. PWM dimming should run above the low-risk frequencies, with 1000Hz and above a common specification point, and never below a few hundred hertz for occupied spaces. Camera-facing installations add their own constraint, since PWM beat frequencies produce banding in video regardless of human perception.
Writing It Into a Spec
A workable procurement line states three things: percent flicker below a defined value at rated output, the same limit holding through the dimming range, and PWM frequency at or above a defined floor. Measuring requires only a photodetector with adequate bandwidth and an oscilloscope; verify at multiple dim levels, because many drivers pass at full output and fail at ten percent.
FAQ
Is flicker at 100Hz always bad? Not always, but depth matters. A few percent of shallow 100Hz ripple sits in the low-risk region; full-depth modulation at 100Hz does not.
Do all dimming protocols flicker? PWM flickers by design unless the frequency is high enough. Analog 0-10V and current-level dimming vary amplitude instead, shifting the flicker question to driver ripple quality.
Can flicker be fixed after installation? Only by changing the driver or the dimming settings. The LED emits what it is fed; there is no optical fix.
