LM-80 measures a real part for 6,000 or 10,000 hours. Nobody runs a fixture for 50,000 hours to prove a claim, so the industry needed a defined way to extend measured data into a projected number. That method is IES TM-21, and its limits are as important as its formula.
What LM-80 Does and Does Not Cover
LM-80 is a test of the LED package or module, not of a finished luminaire. It requires at least 6,000 hours of operation, 25 parts per case temperature, at three case temperatures, commonly 55C, 85C and 105C. Every 1,000 hours the lab records lumen maintenance and color shift.
The report gives measured data only. It contains no lifetime figure. Anyone quoting an “LM-80 lifetime” is quoting something the standard does not produce.
The TM-21 Model
IES TM-21 fits an exponential decay to the measured maintenance curve:
Phi(t) = B x exp(-alpha x t)
Only data past 5,000 hours may be used for the fit, and the fit quality is checked before a projection is issued. The output is the time at which lumen output crosses a chosen threshold: L70 is 70% of the initial value, L80 is 80%, L90 is 90%. Most procurement work is written around L70.
The critical constraint is the reporting limit. TM-21 allows a projected value no longer than six times the LM-80 test duration.
| LM-80 duration | TM-21 reporting limit | Practical meaning |
|---|---|---|
| 6,000 h | 36,000 h | Older or lower-spec parts |
| 10,000 h | 60,000 h | Current baseline for new designs |
| Threshold not reached | “>limit” notation | Report the limit, not a fabricated number |
A part tested for 6,000 hours can therefore never carry a validated L70 of 50,000 hours. If a datasheet shows that combination, the projection is outside TM-21 or the test duration is longer than stated.
What a Valid Report Contains
Read the report before the number. Four items decide whether the L70 figure means anything: the drive current used in the LM-80 test, the case temperature, whether the projection passed the TM-21 validity checks, and whether the luminaire drives the part at the same current.
If the fixture runs the part at a higher current than the LM-80 test, the projection does not transfer. Current and junction temperature both accelerate decay, and TM-21 has no mechanism to correct for a current increase.
Where Projects Go Wrong
- Using package-level LM-80 data to claim a luminaire lifetime. The driver, the optical materials and the solder joints are not in the test.
- Comparing L70 values taken at different case temperatures. A 105C projection is not the same statement as a 55C projection.
- Ignoring color shift. TM-21 addresses lumen maintenance; chromaticity drift is reported separately and matters more in retail and medical work.
- Treating the six-times limit as optional marketing headroom. It is the boundary of the method.
A Workable Procurement Rule
Ask for the LM-80 report and the TM-21 projection as separate documents. Match the test current to the drive current in the fixture. Match the case temperature to the measured or modelled temperature in the application. If the matching temperature is 85C but the report only covers 55C, the projection cannot be used as written.
FAQ
Can TM-21 predict 100,000 hours? No. A 10,000-hour LM-80 test supports a projection up to 60,000 hours. Anything beyond that sits outside the method and should be treated as an estimate, not a verified figure.
Does L70 mean the fixture stops working? No. It means the source has fallen to 70% of its initial light output. The fixture still operates, and in many industrial spaces the practical replacement point is set by a maintained illuminance target rather than by L70 itself.
Is a longer LM-80 test always better? It gives a longer reporting window, which is useful. The more decisive factors are still the test current and the case temperature, because they determine whether the data applies to the fixture being specified.
Related reading
