LED Grow Light Lifespan: What L70, L90 and TM-21 Really Mean
A grow light is usually sold with a number that looks like a lifetime — 50,000 hours, 100,000 hours, or a five-year warranty printed next to it. Those figures come from a testing and projection chain that is well defined, and once the chain is visible it becomes clear which numbers are measured, which are extrapolated, and which are simply commercial. This guide walks through that chain and through the failure modes it does not cover.
- What L70, L80 and L90 actually measure
- Where the numbers come from: LM-80 and TM-21
- Why a rated life is not a warranty
- What shortens lifespan in practice
- How to read a lifespan claim in a specification
- What a lifespan figure does not tell you
- Frequently asked questions
- About the publisher
- References
What L70, L80 and L90 actually measure
Light-emitting diodes do not usually stop working the way a filament lamp does. They lose output gradually. Because there is no single moment of failure, the industry describes LED life as lumen maintenance: the share of the original light output that remains after a given number of operating hours.
The notation is written as Lρ, where ρ is the percentage of initial output that is still present. The most common points are L90, L80, L70 and L50. L70 means the time until the LED holds 70 percent of its initial light output; L90 means the time until it holds 90 percent. These are milestones on a decay curve, not a warranty date and not a moment of failure.
| Notation | What it means | Practical reading in a grow room | Unit |
|---|---|---|---|
| L90 | Time until output falls to 90 percent of the initial value. | A ten percent loss is small but it is the first point at which a light recipe drift becomes measurable on a quantum sensor. | hours |
| L80 | Time until output falls to 80 percent of the initial value. | For a facility that re-plans PPFD between crops, this is often the point at which the fixture is re-hung or dimmed differently rather than replaced. | hours |
| L70 | Time until output falls to 70 percent of the initial value. The figure most often quoted as “lifetime”. | A thirty percent loss changes the delivered DLI at the same mounting height, so the setting that worked when the fixtures were new may no longer be enough. | hours |
| L50 | Time until output falls to 50 percent of the initial value. | Rarely quoted for horticultural fixtures, but it appears in datasets where degradation is faster than the manufacturer expected. | hours |
Where the numbers come from: LM-80 and TM-21
A rated lumen maintenance life is not measured directly. Nobody runs a grow light for 100,000 hours before shipping it. The figure is built from a test plus a projection, and both parts have published rules.
| Step | What it defines | Consequence for a buyer |
|---|---|---|
| IES LM-80 | The test method for measuring how a LED package, array or module loses light over time while held at fixed case temperatures, with light output sampled at intervals. | Ask for the report, not just the number. ENERGY STAR guidance states that a report must include a minimum of one reported case temperature, so the data may exist at a single temperature rather than several. |
| IES TM-21 | The method for projecting long-term lumen maintenance from the LM-80 dataset, including the limit on how far the data may be extrapolated. | Under TM-21 the reported figure is limited to six times the actual test duration, so the achievable claim is capped by how long the test really ran. |
| Published rated life | The headline figure that appears on a datasheet or in a quotation. | It should state the percentage maintained, the hours, and the operating conditions. A figure without conditions is not traceable. |
The extrapolation limit is the part of the chain that is checked least often. A worked example from the Cree LED application note on long-term lumen maintenance makes it concrete: with 6,000 hours of actual test data, the maximum figure that may be reported under TM-21 is “>36,000” hours, because six times 6,000 is 36,000. The same note shows a dataset with 10,000 hours of testing supporting an L70 reported as “>60,000” hours, while an L85 point from the same data set is reported as 28,500 hours.
The arithmetic consequence is simple. If a supplier advertises an L70 of 100,000 hours, the underlying LM-80 dataset has to be roughly 16,700 hours long before that projection is permitted. That is the size of test campaign worth asking about, and it is a far more useful question than arguing about the headline number itself.
Why a rated life is not a warranty
Warranty length and rated lumen maintenance life are usually printed next to each other, which is why they get confused. They answer different questions. The rated life describes how the light output is expected to decline in the LED light source. The warranty describes the period during which the seller will repair or replace the product when it fails to meet the agreed specification.
A short rated life with a long warranty is entirely possible, and so is the reverse. When comparing two quotations, treat them as two separate columns: the rated life to judge how the light recipe will drift, and the warranty terms to judge commercial risk. Neither one can be used to infer the other.
What shortens lifespan in practice
The decay curve published in a datasheet belongs to a particular operating point. Cree LED’s application note displays lumen maintenance curves for several drive currents and junction temperatures, which is the clearest way to see that the curve is not a property of the LED alone. The factors that move a horticultural fixture away from the published curve are practical and predictable.
- Junction temperature. The LED die runs hotter than the case, and the case runs hotter than the room. A fixture with a thin heat path, a passive housing in still air, or a fixture installed close to a ceiling will sit at a higher operating point than the same fixture in an open room.
- Drive current. Running a given LED at higher current raises output per die but also raises the die temperature for the same thermal design, which is why a fixture driven to its maximum is not on the same curve as the same fixture dimmed.
- The driver, not the LED. Drivers contain electrolytic capacitors, which are the components most affected by heat and are a common reason a fixture stops working long before its LEDs reach their L70 point. This is one reason the LM-80 report cannot be used as a statement about the whole luminaire.
- Environment. High humidity, condensation on optics, dust on a diffuser and salt or fertiliser mist all reduce the light that reaches the canopy without changing the LED die at all.
- Scheduled dimming. A fixture operated at reduced output for part of the day accumulates fewer stress hours than one held at full output, so two identical fixtures on different recipes will not age at the same rate.
How to read a lifespan claim in a specification
A lifespan line in a quotation or datasheet can be checked in a few minutes with the following questions. Each one has a specific answer that either exists or does not.
- Which percentage? A bare “50,000 hours” without L70, L80 or L90 is ambiguous. Ask which maintenance point is meant.
- Which case temperature? The figure is only valid at the case temperature used in the test. ENERGY STAR guidance notes that a report need only include a minimum of one reported case temperature, so a single-temperature dataset is normal and the temperature should be stated.
- How much test data? The reported figure cannot exceed six times the LM-80 duration under TM-21, so the test length can be derived from the claim. A supplier that cannot produce the duration cannot support the claim.
- Is the LED datasheet or the luminaire being quoted? LM-80 data belongs to the LED source. The assembled fixture adds a driver, optics and seals that are outside that dataset.
- What is the driver rated for? Ask separately about the driver’s own rated life and its temperature rating, since drivers rather than LEDs are frequently the limiting component.
- What does the warranty actually cover? Read whether it covers the driver as well as the LED board, and whether it is pro-rata or replacement.
What a lifespan figure does not tell you
A rated lumen maintenance life is a single number standing in for a curve. It does not describe colour shift, which changes the spectrum delivered to the canopy even when total output is still inside specification. It does not describe the survival rate of the population — a rated figure is a projection for a sample, not a guarantee that every unit in a room will still be within tolerance on the same date. It does not describe the driver. And it does not describe what happens to the optics under a year of humidity and dust, which is often the difference between a fixture that measures as new and one that does not.
For planning purposes the practical approach is to treat the rated life as a maintenance interval rather than a service life: choose a mounting height and a dimming schedule, then re-measure the canopy with a quantum sensor at intervals and adjust as output declines. That keeps the crop plan tied to measured light rather than to a projection.
Frequently asked questions
Is L70 the same as the life of the fixture?
No. L70 is the number of hours after which the light source is expected to hold 70 percent of its initial output. It describes the gradual decline of the LEDs, not the point at which the fixture stops working. The driver, the thermal path, the seals and the growing environment can all end the service life of a fixture before the LEDs reach their L70 point.
How many hours of testing support a 100,000 hour L70 claim?
Under IES TM-21 the reported lumen maintenance life is limited to six times the actual test duration, so 6,000 hours of LM-80 data supports a maximum reported figure of more than 36,000 hours. A claim of 100,000 hours implies roughly 16,700 hours of real LM-80 data behind it, which is why asking for the test duration is a more productive question than disputing the number itself.
Does a longer warranty mean a longer rated life?
No. A warranty is a commercial promise about defects and replacement, while a rated lumen maintenance life is a technical projection of light output decline. The two are independent, and a fixture can reasonably carry a multi-year warranty alongside a rated life of tens of thousands of hours.
Why does temperature appear in every lumen maintenance discussion?
Because lumen maintenance curves are published for specific operating conditions, and drive current and junction temperature are the two conditions that move the curve most. A fixture that runs its LEDs hotter, or drives them harder, will not follow the curve published for a cooler, gentler operating point.
What does an LM-80 report not cover?
LM-80 covers the LED package, array or module. It does not cover the driver, the optics, the housing seals or the assembled luminaire, and it says nothing about how the fixture behaves in a humid or dusty grow room. ENERGY STAR guidance also notes that a report need only include a minimum of one reported case temperature, so the temperature at which the data was taken has to be checked rather than assumed.
About the publisher
Zhongshan City Ruixian Electronics Factory (XineLam) designs and manufactures LED lighting and LED grow light modules. The company has 17 years of experience in LED lighting and holds 300+ patents in China and internationally. This article is published as technical background for growers and system integrators; it describes published test standards and does not replace the test report issued for a specific product.
Where this matters in practice: for multi-layer racks and small propagation rooms, where fixtures are replaced on a schedule and light recipes are tuned between crops, a rated life with stated conditions is easier to plan around than a headline number without them. For high-bay greenhouse top-lighting, the reachable maintenance interval and the driver rating tend to matter more than the LED figure alone.
References
- Cree LED. XLamp Long-Term Lumen Maintenance (application note) — definitions of L50/L70/L85, LM-80 compliant testing, and the TM-21 limit of six times the test duration. https://assets.cree-led.com/a/da/x/XLamp-Lumen-Maintenance.pdf
- U.S. ENERGY STAR. ENERGY STAR Guidance Regarding Lumen Maintenance Testing and Reporting, 9 September 2011 — minimum reporting requirements for LM-80 data and reference to IES TM-21-11 for interpolation. https://www.energystar.gov/sites/default/files/specs//ENERGY_STAR_Final_Lumen_Maintenance_Guidance_0.pdf
- DesignLights Consortium. Horticultural Lighting Technical Requirements V4.0 — LM-80 applicability section and the current photosynthetic photon efficacy threshold for listed horticultural luminaires. https://designlights.org/our-work/horticultural-lighting/technical-requirements/hort-v4-0/