LED dimmer incompatibility
A Socratic walk-through of LED dimmer incompatibility — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does a new LED bulb flicker or buzz on the dimmer that ran the old one smoothly?
A dimmer ran an incandescent bulb for twenty years, smoothly from an ember to full brightness. Put an LED of the same fitting into the same socket and it stutters at the low end, refuses to light below halfway, jumps on abruptly, or the wall plate hums.
The natural reading is that the new bulb is the worse product. But notice which component is older. The dimmer was built against assumptions about what would be on the other end of the wire, and the bulb has stopped honouring them. The useful question is not "what is wrong with the LED" but "what was the dimmer quietly relying on?"
Reasoning it through
REASONING #Start with what almost everyone believes a dimmer does: reduce the voltage reaching the bulb. It does not. The ordinary household dimmer is a switch that opens and closes a hundred or a hundred and twenty times a second — twice per cycle of the mains, whose frequency and voltage differ by market. Each time the mains passes through zero it blocks; then, part-way through that half-cycle, a semiconductor switch called a TRIAC is triggered and passes what is left. Turning the dial moves the firing point. The peak voltage is untouched; the waveform is chopped.
Why did that ever look smooth? Because a filament is thermally slow. It cannot cool measurably in the few milliseconds it spends unpowered, so it averages the chopped supply into a steady glow. The dimmer was always crude and the bulb concealed it: the smoothness everyone credited to the dimmer was supplied by the lamp, which means the retrofit removed the very component that made the system work.
Now ask what else the TRIAC needed from a filament. Three things, and an LED supplies none of them.
First, current. A TRIAC, once triggered, conducts only while the current through it stays above a small holding current; below that it switches off until triggered again. A 60-watt lamp on a 230-volt supply draws 60 divided by 230, about a quarter of an amp; a 7-watt LED of similar brightness draws under a tenth of that. Many dimmers also steal a trickle through the load to power their own timing circuit, and a load that light does not supply it.
Second, a resistive character. A filament is a resistor: current follows voltage smoothly from the instant of firing. An LED bulb is a rectifier charging a capacitor, feeding an electronic driver. It draws nothing until the instantaneous mains voltage exceeds what its capacitor already holds — and then draws a sharp spike. So when the TRIAC fires, a step of a couple of hundred volts arrives at a capacitor, and that inrush rings through the small choke fitted inside the dimmer to suppress radio interference. A wound core carrying a sharp pulse moves slightly, and that is the buzz.
Third, proportionality. A filament's light tracks the power it is fed. An LED driver regulates its own output current, holding the light almost constant over part of the dial regardless of how much of the half-cycle it gets — then, below some threshold, losing the input it needs and shutting off. That is the dead zone at the bottom and the abrupt pop-on coming back up.
And the LED cannot hide any of it: it responds in microseconds, so every irregular half-cycle is a visible flash.
Is that account testable, or merely plausible? If the low-end flicker is a holding-current failure, adding load must cure it: put one incandescent lamp on the same dimmer alongside the LEDs, or fit a bypass module drawing the missing current. The refuting observation: if the flicker is unchanged with a 60-watt incandescent sharing the circuit, holding current is not the fault, and the trouble lies in the driver rather than the switch. A second test: fit a trailing-edge dimmer, which conducts from the zero crossing and switches off part-way through instead. That waveform starts gently into a capacitor and never produces the inrush step, so if it cures the buzz, the fault was the phase-cut type.
The analogy
THE ANALOGY #Think of a revolving door that keeps turning only while people keep walking through it. Give it a steady stream and it runs quietly. Send one small child every so often and it stalls between them, and someone must shove it again each time — so the door judders, audibly. The TRIAC is that door: it stays open on the traffic passing through it, and the LED does not send enough to keep it moving.
A revolving door stalls slowly and visibly, whereas the drop-out and retrigger happen a hundred times a second; and a door does not care what shape the people are, whereas much of the trouble here is the timing of the LED's current draw, not only its smallness.
Clarifying the model
THE MODEL #Two beliefs to correct, one of them the reason the fault is so puzzling.
"The dimmer lowers the voltage" is the folk mechanism, and it is false. If it were true, an LED would simply glow more dimly and there would be no puzzle. Everything strange here follows from the dimmer being a chopper rather than a regulator, and from the filament having been the smoothing element all along.
"LEDs are not really dimmable" is wrong more subtly. Dimmability is not a property of a bulb but of a pairing: the same LED can be flawless on one dimmer and unusable on the next, which is why manufacturers publish compatibility lists rather than a specification you could check yourself.
An honest caveat — which mechanism dominates in a given pairing is hard to establish from outside the wall, since holding current, inrush and driver regulation give overlapping symptoms, and published minimum-load figures vary so much between models that quoting one would mislead. The shape of the whole thing is path dependence, not error: a control was designed against a load the world has since stopped making.
A picture of it
THE PICTURE #How to readThis repurposes a packet-layout diagram, which normally shows which bits of a message sit in which positions, to show which slices of time are occupied by what. Read left to right across one half-cycle of the mains, divided into thirty-two equal slices, lasting about ten milliseconds before repeating. The dimmer blocks through the opening slices — that fraction is what the dial sets — then fires, and the slices just after are the inrush the choke turns into a hum. The final block is the pathology: too little current to stay latched, so the TRIAC lets go and the strip goes dark before the half-cycle ends, where with an incandescent it would be more conduction.
What became clearer
WHAT CLEARED #A dimmer does not dim. It chops the mains into fragments a hundred times a second and relies on the load to average them back into steady light, to draw enough current to keep its switch latched, and to behave like a resistor while doing it. A filament did all three without anyone noticing. An LED does none of them: too fast to average anything, drawing a tenth of the current, and presenting a capacitor that gulps rather than sips. The flicker and the buzz are not the new bulb failing — they are the old control's hidden assumptions becoming visible the moment the component covering for them was removed.
Where to go next
ONWARD #- Why some LED bulbs glow faintly after the switch is off, and what that says about leakage through a dimmer's own supply.
Key terms
TERMS #| Term | What it means |
|---|---|
| Phase-cut dimming | reducing light by blocking part of each mains half-cycle rather than by lowering the voltage. |
| TRIAC | a semiconductor switch that, once triggered, conducts until the current through it falls below its holding value. |
| Holding current | the minimum current that keeps a triggered TRIAC latched on. |
| Leading and trailing edge | dimmers blocking the start of each half-cycle versus its end; the latter suits capacitive electronic loads. |
Every term the collection defines is gathered in the glossary.