20,000 Hours or 50,000? How to Calculate the Real Life of an LED Driver

|How to|11/02/2026|11 min|
Table of contents
Share Post

Every second datasheet in this industry claims 50,000 hours. Some claim 100,000. Almost none of them tell you at what case temperature, at what load, and whether the number refers to the driver or to one component inside it. When a buyer asks us how long our drivers last, the honest answer takes about four minutes of arithmetic and one measurement they probably have not taken.

That answer, for a 400 W indoor unit running at full load in a closed cabinet at 45 °C, is about 20,000 hours — which is the figure on our datasheet, and which works out to roughly 4.5 years at 12 hours a day. Not 50,000. The difference between 20,000 and 50,000 is not marketing spin on our side or on anyone else’s; it is a temperature difference of about 13 °C in a place nobody measures.

This article is the calculation. It is the one we run internally before quoting a warranty, and it is the one you should run before accepting a five-year guarantee from a supplier who has never asked where the driver will be mounted.

Only one component sets the life, and it is not the one you think

A switching driver contains maybe 180 components. The MOSFETs are rated for 100,000 hours of operation at their junction temperature. The controller IC is a commodity part with a failure rate measured in FITs. The PCB, if it is a decent FR-4 with proper creepage, outlives all of them.

The electrolytic bulk capacitor on the input, and the output capacitors on the secondary, are the limiting items. They contain a liquid electrolyte that evaporates slowly through the rubber seal, and the evaporation rate doubles for every 10 °C rise in core temperature. That is not a rule of thumb invented by marketing departments; it is the Arrhenius relationship, and the industry-standard form is:

L = L₀ × 2^((T_max − T_c) / 10)

Where L₀ is the rated life at the rated maximum temperature, T_max is that maximum (usually 105 °C for the capacitors used in power supplies), and T_c is the actual core or case temperature in service.

Take a capacitor rated 5,000 hours at 105 °C. Put it in a driver where the can sits at 65 °C. The exponent is (105 − 65) / 10 = 4, so life is 5,000 × 16 = 80,000 hours. Run the same capacitor at 85 °C and the exponent is 2: 5,000 × 4 = 20,000 hours. At 95 °C it is 10,000 hours. At 105 °C it is 5,000 hours, exactly as printed.

That is the whole mechanism behind the 50,000-hour claim. Somebody took a capacitor rated 5,000 hours at 105 °C, assumed a 68 °C can temperature in a well-ventilated luminaire, and published the result. It is not a lie. It is a calculation for a mounting condition that may not be yours.

Capacitor can temperature Life multiplier Life from a 5,000 h / 105 °C part
55 °C ×32 160,000 h
65 °C ×16 80,000 h
75 °C ×8 40,000 h
85 °C ×4 20,000 h
95 °C ×2 10,000 h
105 °C ×1 5,000 h

Look at the first two rows against the last two. A 40 °C difference in can temperature is a 32-fold difference in predicted life. No amount of component quality compensates for that. This is why the single most effective thing you can do for driver life is not buying a better brand — it is moving the driver 30 cm down inside the cabinet.

Where the 13 °C comes from

Our 20,000-hour figure assumes a 85 °C can temperature. A 50,000-hour claim assumes about 72 °C. Both are plausible in different installations, and the difference between them is almost entirely the enclosure.

When you measure this properly, you find three temperature steps stacked on top of each other:

  1. Room ambient. Usually 25–30 °C in a plant room, 35–40 °C in a ceiling void above a shop, 45 °C plus in a sealed roof space in summer.
  1. Enclosure rise. An IP20 open-frame driver in a ventilated steel cabinet adds 8–12 °C. The same driver in a sealed IP67 plastic box adds 15–25 °C, because the only path out is through the plastic.
  1. Internal rise from case to capacitor core. Another 5–12 °C depending on how close the capacitor sits to the switching devices and the output diodes.

So a driver in a ventilated indoor cabinet at 28 °C room ambient can easily have its capacitor at 50 °C — which is 160,000 hours of capacitor life and means the driver will die of something else. The same driver, same model, mounted in a sealed box in a ceiling void at 45 °C, has its capacitor at 82 °C — 8,000 hours, or eighteen months at 12 hours a day.

Same product. Two different products from the customer’s point of view. Neither datasheet is wrong; both are describing a condition.

How to measure it in twenty minutes

You need a thermocouple meter, some Kapton tape, and four hours. A Fluke 87V with the standard bead probe is fine; a datalogger is better because you want the steady state, not the first reading.

  1. Mount the driver in its actual enclosure, in its actual position, with the actual load. Not on a bench.
  1. Tape the thermocouple to the side of the largest electrolytic capacitor can, not to the heatsink, not to the case. The capacitor is what you are trying to predict.
  1. Run at 80–100% load for at least four hours. Small drivers reach thermal equilibrium in 40 minutes; 400 W units in a sealed box can take three hours.
  1. Record the room ambient at the same time, at three heights, and take the hottest.
  1. Calculate. Use the measured can temperature in the formula above, not the ambient.

The number that comes out will surprise you, and it will be different from the datasheet on both sides — sometimes better, usually worse.

One warning from experience: do not measure in winter and extrapolate to summer. A ceiling void in Milan goes from 22 °C in March to 48 °C in July. If you size a warranty on a March measurement you will be replacing drivers in September, and the failure will look like a manufacturing defect when it is a seasonal load case.

MTBF is not lifetime, and the confusion is systematic

Some datasheets quote 200,000 hours MTBF. Some buyers read that as “23 years” and treat it as a life figure. It is not.

MTBF — mean time between failures — comes from reliability prediction handbooks such as MIL-HDBK-217 or Telcordia SR-332. Both models assume a constant failure rate: the probability of failure per hour does not change with age. That assumption is reasonable for random failures in electronic components during their useful life, and it is completely wrong for wear-out mechanisms like electrolyte evaporation, which accelerate with age.

The consequence is that MTBF tells you about random failures — a solder joint, a MOSFET that was marginal from the start — and tells you nothing about the day the capacitor dries out. A driver can have a 200,000-hour MTBF and a 12,000-hour wear-out life. The MTBF describes the first twelve months; the capacitor describes the fifth year.

There is also a dirty secret in MIL-HDBK-217: the predicted MTBF depends heavily on the “environment” factor you select. Ground benign versus ground fixed versus a cabinet in a factory changes the number by a factor of three. Two suppliers can publish numbers three times apart using the same handbook and the same components, and both be technically correct.

If a supplier gives you MTBF and no life figure, ask for the capacitor part number, its rated hours and temperature, and the assumed case temperature. That is the only set of numbers that actually predicts when units come back.

What actually kills drivers in the field

We warranty our indoor range for three years and we log every return. Over the last four years the pattern has been remarkably consistent, and it is not what the component-level analysis would predict.

Roughly half of the returns are thermal in origin — capacitor dry-out or a thermal shutdown that the customer interpreted as a dead unit. About a quarter are input-side damage: surges, sustained overvoltage from a generator, or a wiring error during commissioning. The remaining quarter is a mix of mechanical damage, water ingress on outdoor models with a badly torqued gland, and a small number of genuine component failures.

The thermal half is almost always traceable to one of three things: the driver mounted at the top of an enclosure where the hot air collects, the driver mounted with its ventilation slots facing a wall 20 mm away, or the driver mounted in a sealed box with no derating applied. All three are installation decisions made by someone who was not thinking about the capacitor.

This is why we now put a mounting note in the box for the 100W IP20 indoor LED driver: mount in the lower half of the enclosure, keep 30 mm clear above the ventilation slots, and if the enclosure is sealed, derate by 20% or ask us for the temperature rise data. It reads like a legal disclaimer. It is actually the difference between 8,000 and 80,000 hours.

The three things that move the number most

Ranked by effect per euro spent:

1. Mounting position. Moving a driver from the top of a cabinet to the bottom typically drops the local ambient by 10–15 °C, which is a 2–3× life multiplier. Cost: zero. Effect: larger than any component upgrade you can buy.

2. Enclosure ventilation. Adding a louvre or a fan to a sealed cabinet that is running at 15 °C above ambient. Cost: €15. Effect: 2× on life, plus it fixes the output derating you are probably already suffering without knowing.

3. Capacitor specification. Upgrading from a 105 °C / 5,000 h general-purpose part to a 105 °C / 12,000 h long-life series. Cost: 4–7% on the driver. Effect: 2.4× on the capacitor life — but only if the case temperature is under control. Put a long-life capacitor in a 95 °C environment and you have bought 24,000 hours instead of 10,000, which is better and still not 50,000.

Note what is not on that list: output power. Load has a second-order effect compared with temperature. A driver at 60% load in a 60 °C enclosure will die long before a driver at 100% load in a 40 °C enclosure, because the capacitor temperature follows the enclosure far more than it follows the load. Buyers routinely derate for life and then install in a hot box, which is the engineering equivalent of wearing a seatbelt and driving faster.

What to write into a specification

If you want a life figure you can actually hold a supplier to, write it like this:

Driver shall achieve ≥40,000 hours at 80% load with a capacitor case temperature not exceeding 75 °C, calculated using the Arrhenius 10 °C rule from the capacitor manufacturer’s rated life. Supplier shall state the capacitor series, its rated hours and temperature, and the assumed case temperature. MTBF figures alone are not acceptable.

That clause does three things. It defines the condition, it defines the method, and it makes the supplier commit to a component you can check. Any supplier who cannot answer it has not done the calculation, which means their 50,000-hour claim was copied from a template.

For projects where the driver is inaccessible — a ceiling void, a pole-top cabinet, a signage box on a roof — the calculation should drive the product choice, not the other way round. That is when it makes sense to move up a size and run at 50% load, or to specify the model with the bigger case and the external heatsink face. Both cost less than a lift ticket and an electrician for a warranty replacement.

If you want, send us the enclosure dimensions, the ambient profile and the daily running hours, and we will run the numbers for the 150W indoor power supply or whichever model fits, and tell you what life it will actually deliver in that box — including the case where the answer is “this model is wrong for your enclosure, use the 100 W at 70%”.

FAQ

How long do LED drivers last in practice? Between 8,000 and 80,000 hours, depending almost entirely on capacitor case temperature. That is not a hedge; it is the measured spread from the same product in different enclosures. At 12 hours a day, 8,000 hours is under two years and 80,000 hours is eighteen.

Does the 50,000-hour figure on a datasheet mean the driver? Usually it means the electrolytic capacitor, at an assumed case temperature of about 70 °C. Ask for the assumed temperature. If the supplier cannot name it, the number is decorative.

Is MTBF the same as lifetime? No. MTBF assumes a constant failure rate and describes random failures during useful life. Wear-out from electrolyte evaporation accelerates with age and is not modelled. A 200,000-hour MTBF is compatible with a 10,000-hour life.

Does running at 50% load double the life? Not on its own. Load affects the internal temperature rise, but the enclosure usually dominates. 50% load in the same box might buy you 20–30%, not 100%. Move the driver to a cooler position first.

Why do drivers fail in summer but not winter? Because capacitor life is exponential in temperature and the failure is a threshold effect: the units that were at 90 °C in July were at 70 °C in March and would have lasted five more years at that temperature. Seasonal installations mask the problem for one cycle, then fail in a cluster.

Want the real number for your enclosure? Send the cabinet dimensions, the ambient profile and the running hours, and we will calculate the life for the 100W IP20 indoor LED driver or any model in the indoor range before you place the order. We will also tell you when the honest answer is a smaller unit run cooler.

SANPU

Stay in the loop

Subscribe to our free newsletter.