When Should You Replace an LED Grow Light?

The box said 50,000 hours. My meter said otherwise.

A four year old bar light over my Ficus lyrata still looked bright to my eye, still ran cool, still had every diode lit. At 40 cm it was pushing 78 µmol/m²/s. When that same fixture was new I logged 121 at the same distance, same spot, same meter. It lost 36 percent of its output and never once looked broken.

That is the whole problem with LED grow lights. They rarely die. They fade, quietly, while you keep giving the plant the same twelve hours and wondering why the new leaves are getting smaller.

Do LED Grow Lights Really Last 50,000 Hours?

The number is real. It is also answering a different question than the one you are asking.

Manufacturers quote L70, the point at which a diode still emits 70 percent of its original light. So a 50,000 hour L70 rating means the LED itself is expected to keep 70 percent output that far out. It does not mean the fixture is good for 50,000 hours of growing.

Three things break that promise indoors:

• The driver, not the diode. The power supply is the weak link in budget fixtures. Electrolytic capacitors in a driver running at 60 °C have a service life closer to 15,000 hours. When output sags across the whole panel at once, suspect the driver.
• Heat at the board. L70 figures assume a case temperature around 25 °C. A sealed plastic bar tucked under a shelf with no airflow can sit at 55 to 65 °C, and every extra 10 °C roughly halves the useful life.
• Dirt. A film of dust and kitchen grease on the diffuser cost me 9 percent output. A wipe with isopropyl brought most of it back.

My rough rule after metering a dozen fixtures over five years: divide the box claim by three for a cheap unbranded light, by two for a mid-range one with a metal heatsink.

What Four Ageing Fixtures Actually Measured

I meter every light the day it arrives and write the figure on masking tape stuck to the cable. Cheapest habit I own.

• 35 W clip-on bar, 4 years, ~17,500 hours. New: 121 µmol/m²/s at 40 cm. Now: 78. Down 36 percent.
• 24 W screw-in bulb, 2 years, ~8,700 hours. New: 96 at 30 cm. Now: 88. Down 8 percent.
• 100 W panel with an aluminium heatsink, 5 years, ~20,000 hours. New: 310 at 45 cm. Now: 254. Down 18 percent.
• 18 W strip light, 3 years, ~13,000 hours. New: 54 at 25 cm. Now: 31. Down 43 percent, and two diodes had gone dark.

The panel with the real heatsink is the standout: more hours than the clip-on bar, half the loss. Build quality shows up as retained output, the same argument I make in LED versus fluorescent grow lights.

None of them failed. Even the 43 percent case still switched on every morning.

How Do You Tell a Fading Light From a Fading Plant?

You measure, because your eye is useless here. Human vision adapts logarithmically, so a fixture can drop a third of its output and look identical.

Three ways to check, in order of how much I trust them:

A PAR meter. Best answer, around 60 to 150 for a usable one. Take the reading at leaf height, mark the exact spot with tape, and repeat it every six months in the same place.

A lux meter or phone app. Not accurate in absolute terms for LED spectra, but fine for tracking change on one fixture. If the same app in the same spot reads 8,400 lux this year against 12,000 last year, that gap is real even if neither figure is. Both methods are compared in PAR meter versus lux app.

Plant symptoms. Slowest and most ambiguous signal. New leaves smaller than old ones, longer gaps between nodes, an Epipremnum aureum reaching sideways off the shelf, an Echeveria elegans stretching and losing its rosette shape. By then the plant has been short of light for weeks, and the fix list is in etiolation from too little light.

One trap worth naming. Check the distance before you blame the fixture. Plants grow toward the light, so a pot that sat 40 cm below the bar last spring might be 22 cm below it now. Height maths is in how far to hang grow lights.

The Four Numbers That Say Replace

Here is my actual decision list. Any one of these and the fixture goes.

1. Output down 30 percent or more from baseline. Below that I lower the light or extend the photoperiod by an hour. Past 30 percent, compensation stops working.
2. More than 5 percent of diodes dark. Dead diodes on a series string load the survivors and the failure rate accelerates.
3. Visible flicker, buzzing, or a hot spot on the housing. Driver end of life. Replace it now rather than after the smell.
4. Yellowed or crazed diffuser plastic. UV and heat cloud the lens, and no amount of cleaning fixes it.

Two things are not on the list: age in years on its own, and a lamp looking dimmer than its neighbour. Hours at temperature matter, calendar years do not, and comparing two different fixtures tells you nothing about either.

When you do replace, buy the heatsink rather than the wattage. A 60 W fixture with finned aluminium will out-live a 100 W plastic bar and hold its output better across year three and four.

Getting More Hours Out of the One You Have

Three habits that measurably extend a fixture, ranked by what they gave me.

Give it 3 cm of air above the housing. My worst case, the 18 W strip, was screwed flat against a wooden shelf with no gap. Case temperature 61 °C. The panel that aged best hung in free air at 34 °C.

Clean the lens every three months. Isopropyl on a microfibre cloth, cold fixture. The 9 percent I recovered is more than a year of normal decay.

Run a hard 12 hour photoperiod. Fewer hours is fewer hours of decay, and one clean switch per day is kinder to the driver than repeated cycling. See the grow light timer guide.

An LED grow light gives you no failure event to react to. It gets weaker while your plant gets leggier, both slowly enough that neither registers. Write the day-one reading on the cable, re-check every six months, act at 30 percent down. Not sure how much light your shelf gets to begin with? Start with our free light calculator.