Flicker Depth: The LED Candle Spec Buyers Should Test
Ask a supplier which flicker mode looks most like a real flame, and you will get a demo video, a thumbs up, and a number you can never put on a spec sheet. I have watched buyers argue about realistic flame on phone calls for an hour, because everyone is rating a flicker effect they cannot measure. The measurable half of it has a name: flicker depth.
Flicker depth is a percentage, not a mode count
The mode count on a remote — 5 flicker levels, 3 flame effects — tells you how many software states the LED driver holds. It says nothing about which of those states is a candle. Flicker depth is the distance between the flame’s peak brightness and its dim floor, measured as a percentage of peak. A real flame in a still room sits somewhere around 55–70% depth, meaning it dips to roughly 30–45% of its brightest moment and comes back, with the dip itself varying from one cycle to the next.
Our test rig is a photodiode aimed at the wick-tip LED and a logging meter. A cheap single-LED flameless candle with a fixed 2 Hz blink lands at 95% depth — it goes near-dark and back, every second, like a lighthouse that forgot its job. That is not a flame, and the eye rejects it in about two seconds. When a vendor’s flame mode logs a steady 980-lumen line with a 2% wobble on top, that is the opposite failure: too smooth to read as fire. Both extremes are visible defects, and both are cheap to catch before you order 20,000 units.
The dip shape matters more than the dip size
Here is what the mode count never tells you: two candles at the same 60% depth can look completely different, because the depth is only the vertical axis. The horizontal behavior — how long each dip lasts, how the interval between dips varies, whether the dim floor ever holds — is what separates a believable flame from a blinking light. A real flame’s dips are irregular: 0.3 seconds, then 0.9, then 0.4, and the interval between them drifts too. The flicker modes on a stock driver are almost always evenly spaced, and the evenness is the tell.
When we qualify a driver for a wax-coated LED candle — the kind buyers mistake for real wax on a store shelf — we reject any algorithm that lets the dim floor sit below 25% of peak for more than a third of a second. Below that floor, the flame reads as off and back on, and the illusion dies even though every spec on the sheet says pass. Ask your led candle manufacturer for the dim-floor percentage and the dip-duration range, not the mode count. Three questions that return numbers beat one demo video.
What to actually put on your spec sheet
Three lines, none of them about modes: flicker depth at 55–70% of peak, dim floor no lower than 25% of peak, and dip interval variance above 40% cycle-to-cycle. Add the flicker frequency range if you are lighting a wedding reception or a restaurant table, because a 4 Hz flicker across 300 units on a floor looks like a strobe, and a 1.2 Hz flicker on a single centerpiece reads as a dying pilot light. We tune the range by application, and the range is in the algorithm file, not in the mode list.
One last note from the inspection side: flicker depth drifts with supply voltage. A candle that measures 65% depth on a fresh cell can read 40% at the end of its runtime, because the LED dims while the driver’s dim floor does not move. So the spec has a second axis — measure depth at 100% charge and at the last hour of a full cycle. The gap between those two numbers is the flicker spec your customers will actually experience at 11 pm on a Friday, not the one they saw at the booth. That is the number worth arguing over.