LM-80 vs TM-21: How LED “Lifetime” Claims Are Calculated (Without the Hype)
LED “50,000–100,000 hour” claims aren’t magic—mostly they’re spreadsheet outputs chained to LM-80 test data and TM-21 projection rules. Here’s how the sausage is made, where it breaks, and how to spot the hype fast.
Not always on purpose, but the industry has trained itself to treat “50,000 hours” like a fact you can print on a box, even though it’s usually a conditional forecast built from LED package lab data, a projection method with hard limits, and a long list of real-world failure modes that the math never touches. So when a spec sheet says “L70 100,000 h,” what did they actually test, for how long, at what temperature, and what exactly is being projected?
And why do so many buyers still accept the number without asking the only question that matters: “Show me the LM-80 and the TM-21 output”?
Table of Contents
LM-80 is measurement. TM-21 is math.
LM-80 is the boring part, which is exactly why it’s valuable. It’s a standardized way to measure lumen maintenance (lumen depreciation) of an LED light source (package/array/module) over time at controlled conditions—think fixed drive current, fixed case temperature, periodic measurements, and a minimum test duration that’s long enough to have real signal. The output is data: how many lumens you still have at 1,000 hours, 2,000 hours… through at least 6,000 hours in many reports.
TM-21 is where marketing departments start smiling. TM-21 takes LM-80 lumen-maintenance data and projects it forward using curve-fitting rules, then outputs an estimated time to a lumen-maintenance threshold like L70 (70% of initial lumens) or L80, L90. It’s not “fake.” It’s just not a guarantee.
Here’s the hard truth: TM-21 can only be as honest as the LM-80 data and the thermal assumptions behind it.
The 6x rule: the industry’s speed limit (and the most ignored number on spec sheets)
TM-21 includes an extrapolation cap that’s widely summarized as the “6x rule”: you don’t get to project forever off a short test. The IES has been blunt that marketing claims should not exceed the maximum projection limits allowed by TM-21, and that overstated claims are a real problem.
That’s why “100,000 hours from a 6,000-hour LM-80” should trigger your reflexive skepticism. 6,000 × 6 = 36,000. Not 100,000.
Yes, companies get cute with wording (“calculated,” “design life,” “engineering estimate,” “typical”), but if the claim is positioned like a comparable lifetime metric, the buyer is being nudged into a false equivalence.
Reported vs projected: the single most common buyer trap
LM-80 reports often include a TM-21 projection summary, and you’ll see language like “L70(6K)” or “L70(10K).” That parenthetical matters because it tells you how much test data was used.
Example: a Nichia LM-80 report for a warm-white LED package (NFSL757D, nominal 2700 K) shows a TM-21 projection line of “L70(6K) > 36,700 hours” tied to 6,000 hours of testing. That “greater than” is not a victory lap; it’s basically the method saying, “I can’t responsibly publish beyond the allowed boundary, but the curve hasn’t hit L70 yet.”
So when you see “L70 > 50,000 h” without the “(6K)/(10K)” context and without the actual TM-21 output, you’re not reading engineering—you’re reading sales.
“LED lifetime” is often not luminaire lifetime. That gap costs money.
I’m going to be blunt: most lifetime claims are LED-package-centric, while your failure in the field is usually driver, solder, optics, heat, or controls.
EPA’s own ENERGY STAR downlights comment response admits a point the industry hates to talk about in public: maximum driver case temperature has “consistently been a source of luminaire failures” and remains a significant failure category, which is why they kept thermal verification requirements. That’s not theory. That’s program-level scar tissue.
And controls? Even uglier. A U.S. Department of Energy report on 0–10V control for LED streetlights measured massive variation in response: ~53 percentage points average range in relative power draw across tested control voltages, and 9 of 19 drivers evaluated were noncompliant with ANSI C137.1-2022. Translation: you can “meet spec” on paper and still deliver unpredictable behavior in deployment. That unpredictability is a reliability story, not just an energy story—because heat follows power, and drivers die faster when they run hotter than intended.
So if your procurement language says “L70 50,000 hours,” but you don’t lock down driver thermal performance, you’re buying a number that won’t be the first thing to fail.
How serious programs police the hype (and why that should shape your spec)
Two places where the nonsense gets squeezed out: ENERGY STAR and the DLC.
EPA confirmed the official ANSI/IES TM-21 Calculator release date (June 1, 2023) and noted concerns that it can produce more conservative projections than the older ENERGY STAR calculator; EPA’s response was to propose simplifying lifetime requirements (e.g., a single 25,000-hour requirement in Draft 2 logic) rather than letting projection disputes become a loophole. That’s not “lowering the bar.” It’s acknowledging that projection math isn’t a warranty.
Now look at DLC horticultural lighting (where long operating hours and thermal stress are normal). DLC V3.0 (revised May 13, 2024) requires Q90 ≥ 36,000 hours using LM-80 + TM-21 (or LM-84 + TM-28), demands in-situ temperature measurement testing (ISTMT), and explicitly requires submission of the TM-21 calculator report (PDF + JSON) as pathways transition away from older ENERGY STAR calculators. They also require driver documentation showing ≥ 50,000 hours lifetime at temperature, tied to measured in-situ conditions.
That’s the playbook: data + in-situ thermal reality + documentation that can be audited.
The minimum documents I think a professional buyer should demand
If a supplier gets cagey, treat it as the answer.
LM-80 test report (full report, not a summary) You want test duration, drive current, case temperatures, sample count, lumen maintenance tables, chromaticity shift.
TM-21 output (calculator report, preferably with the official tool output) If the claim says L70, the projection should say L70, with the “(6K)/(10K)” context.
In-situ temperature evidence (LED and driver) If the luminaire runs the LED package hotter than the LM-80 case temperature, the projection is optimistic by definition.
Driver lifetime-at-temperature curve + TMP location Programs keep hammering this because drivers are frequent first-failure components.
Warranty language that matches the claim “L70 100,000 h” with a 3-year warranty is comedy.
If you’re sourcing custom builds, push this upstream during design—thermal stack, driver choice, and documentation packaging—through an OEM/ODM LED lighting services process instead of trying to “paper over” gaps after tooling. And if you need a clean place to centralize files and compliance artifacts, start with a vendor’s resources and downloads library so your team isn’t chasing PDFs in email threads.
Comparison table: what each “lifetime” artifact can (and can’t) prove
Item
What it actually is
What it can support
What it cannot support
LM-80 test report
Measured lumen maintenance of an LED light source at defined currents/temps over thousands of hours
Evidence that a specific LED package/array/module depreciates slowly under those lab conditions
Curve-fit projection of LM-80 lumen data with strict limits on publishable projection
A defensible reported L70/L80/L90 projection within method limits
“Guaranteed life,” anything beyond the projection cap, or claims that ignore hotter in-fixture temperatures
Spec-sheet “50,000–100,000 h”
Usually a simplified, sometimes cherry-picked statement
A starting point for questions
A procurement-grade durability promise
Program requirements (ENERGY STAR/DLC)
Audit-driven rules tying projections to tools, temps, documentation
A template for buyer specs that survive reality
A substitute for your site conditions and maintenance economics
Field reliability
What breaks in the real world
The only thing finance cares about
Something you can infer from LED-package lumen math alone
FAQs
How is LED “lifetime” calculated using LM-80 and TM-21?
LM-80 and TM-21 “calculate LED lifetime” by measuring lumen maintenance of an LED light source for thousands of hours under controlled temperatures and currents (LM-80), then fitting that measured decay data with the TM-21 projection method to estimate when light output will drop to a threshold like L70 or L80. After that definition, the catch: the projection is only as credible as the LM-80 dataset and the assumption that your in-fixture temperatures match the test conditions.
What is the TM-21 extrapolation limit (the “6x rule”)?
The TM-21 extrapolation limit (often summarized as the “6x rule”) is a restriction that prevents publishing lumen-maintenance lifetime projections that extend beyond a multiple of the actual LM-80 test duration, specifically to stop unrealistic long-life claims being generated from short-term data. Practically, if someone is implying 100,000 hours off 6,000 hours of LM-80, you should demand the TM-21 report and scrutinize how they’re wording the claim.
What does “L70” mean in LED lumen maintenance?
L70 is a lumen-maintenance threshold meaning the time at which a light source is projected or measured to deliver 70% of its initial luminous flux, so it’s about gradual dimming (lumen depreciation), not sudden failure, flicker, color shift, or driver death. That distinction matters because many field replacements happen due to driver faults or unacceptable color shift long before the LED package reaches L70.
What’s the difference between “reported” and “projected” LED lifetime?
“Reported” lifetime in TM-21 context is the publishable, method-compliant lumen-maintenance life output constrained by TM-21’s limits and reporting rules, while “projected” (or “calculated”) lifetime can refer to the raw curve-fit estimate that may extend beyond what standards bodies consider appropriate to publish as a comparable claim. If a spec sheet omits whether the value is TM-21 reported, treat it as incomplete.
Why do ENERGY STAR and DLC care so much about temperature and drivers?
ENERGY STAR and DLC emphasize temperature and drivers because luminaire failures and performance drift are strongly driven by thermal stress on electronics, and programs have repeatedly observed driver temperature as a major failure category; they therefore require thermal verification and documentation that ties lifetime claims to in-situ operating conditions. A “good LED” run hot in a bad fixture is still a bad product in the field.
Conclusion
If you want fewer surprises in year 3, stop buying lifetime slogans and start buying evidence. Send your short list of fixtures and applications (warehouse linear, retail track, outdoor area) and require: LM-80 report, TM-21 calculator output, and in-situ LED/driver temperature documentation in the submittal package. If a vendor can’t produce that cleanly, they’re not ready for professional procurement—full stop.
