LM-80 与 TM-21：完整对比
| Parameter | LM-80 | TM-21 | Winner |
|---|---|---|---|
| Standard Type | Test method (IES LM-80-15) | Projection method (IES TM-21-19) | — |
| What It Measures | Actual lumen depreciation at 0, 1000, 3000, 5000, 6000+ hours | Extrapolated lumen maintenance at 25k, 50k, 100k hours | LM-80 (real data) |
| Minimum Test Duration | 6,000 hours (8.2 months) per IES LM-80-15 | No test — uses LM-80 data as input | TM-21 (no wait) |
| Temperature Requirements | Three case temperatures: 55°C, 85°C, and a third point per standard | Uses LM-80 data at the temperature closest to in-situ conditions | LM-80 (more granular) |
| Output Metric | Lumen maintenance percentage at each test interval | Projected L70, L80, or L90 life in hours | TM-21 (actionable) |
| Sample Size | Minimum 20 units per test condition per IES LM-80-15 | Uses LM-80 sample data; no separate sample requirement | LM-80 (statistical rigor) |
| Cost to Perform | $15,000–$30,000 per LED model for full 6,000-hour test | Minimal — calculation based on existing LM-80 data | TM-21 (cheaper) |
| Regulatory Acceptance | Required by ENERGY STAR, DLC, Title 24, and most building codes | Required by ENERGY STAR and DLC for lifetime claims | Tie (both required) |
| Failure Mode Detection | Catches early failures, solder joint issues, phosphor degradation | Assumes exponential decay; misses sudden failure modes | LM-80 (real-world) |
| Applicable Products | LED packages, modules, and arrays only | Same as LM-80 — applies to LED light sources, not complete luminaires | Tie |
LM-80 gives you the raw truth. You're looking at actual photometric measurements taken at 0, 1000, 3000, 5000, and 6000 hours minimum, with the standard requiring data at three different case temperatures. I've seen tests where a supposedly premium LED dropped 8% in the first 3000 hours — that's something TM-21 would never catch on its own because the projection assumes a smooth exponential curve. The LM-80 data tells you if the LED has a manufacturing defect or a phosphor stability issue right out of the gate.
TM-21 takes that data and applies an exponential decay model (Arrhenius-based) to project forward. Here's the catch: the projection is only valid for six times the test duration. So with 6,000 hours of LM-80 data, you can project out to 36,000 hours. Want a 50,000-hour L70 claim? You need at least 8,333 hours of LM-80 data. Most reputable manufacturers run 10,000 hours or more to get comfortable projections. What does this mean in practice? A TM-21 projection of L70 at 60,000 hours based on 6,000 hours of data is mathematically suspect — the uncertainty balloons beyond 36,000 hours. I've seen spec sheets claiming 100,000-hour life on 6,000-hour data, and that's just not credible engineering.
Bottom line: LM-80 is the evidence, TM-21 is the interpretation. A good spec includes both, with the LM-80 test duration clearly stated.
Running a full LM-80 test on a single LED model across three temperatures with 20 samples per condition will set you back $15,000 to $30,000. That's for the photometric testing alone — doesn't include the cost of the LED samples or the engineering time to set up the test boards. For a manufacturer with 50 LED models in their portfolio, that's potentially $1 million in testing. That's why you'll see many smaller brands using generic LM-80 data from the LED package manufacturer rather than testing their own modules.
TM-21 costs essentially nothing once you have the LM-80 data. It's a calculation — you can do it in Excel with the formula from the standard. But here's the hidden cost: if your TM-21 projection doesn't meet the L70 threshold your customer needs, you're either retesting with longer duration (more LM-80 cost) or redesigning the thermal management. I've been on calls where a client had to scrap a whole fixture design because the TM-21 projection showed L70 at only 35,000 hours — they'd already spent $25,000 on LM-80 testing. The payback on doing TM-21 projections early in the design phase is enormous.
LM-80 is essential for any application where you need to validate that the LED itself is reliable — think industrial high-bay, outdoor parking lot lighting, or any installation where relamping is expensive. If you're specifying for a 24/7 operation like a warehouse or a hospital corridor, you want to see the raw LM-80 data, not just the TM-21 projection. I've seen TM-21 projections that looked great on paper but the LM-80 data showed a 5% drop in the first 1000 hours that the exponential model smoothed over.
TM-21 is what you use when you need to compare lifetime claims across different manufacturers. Every ENERGY STAR and DLC listed product has a TM-21 projection — it's the only way to compare apples to apples. For retail or office applications where the expected life is 50,000 hours, a TM-21 projection based on 10,000 hours of LM-80 data is perfectly adequate. Just don't use it for mission-critical applications where a sudden failure would be catastrophic — TM-21 won't predict that.
LM-80 Pros: Real measured data, catches early failures, required by all major energy programs, gives you confidence in the LED's intrinsic reliability. Cons: Expensive, takes 6-12 months to complete, only tests the LED package not the complete luminaire, doesn't account for driver failures or thermal design of the fixture.
TM-21 Pros: Free once you have LM-80 data, enables lifetime comparisons, required for regulatory claims, mathematically sound when applied correctly. Cons: Only as good as the input data, assumes exponential decay, limited projection window (6x test duration), can't predict catastrophic failures or wear-out mechanisms.
| Use Case | Recommended | Reason |
|---|---|---|
| New product development — LED selection | LM-80 first | You need to validate the LED's intrinsic reliability before designing the fixture around it |
| Comparing lifetime claims from different manufacturers | TM-21 | Only TM-21 gives you a standardized projection to compare across brands |
| ENERGY STAR or DLC listing application | Both required | ENERGY STAR requires LM-80 data and TM-21 projection for all listed products |
| Industrial high-bay with 24/7 operation | LM-80 with 10,000+ hour data | Extended LM-80 duration reduces projection uncertainty for long-life applications |
| Retail or office with 50,000-hour expected life | TM-21 on 6,000-hour LM-80 data | Standard 6,000-hour LM-80 gives you a valid 36,000-hour TM-21 projection — adequate for most commercial applications |
| Warranty validation for a 10-year warranty | Both, with 10,000-hour LM-80 minimum | 10-year warranty requires L70 at 87,600 hours — you need at least 14,600 hours of LM-80 data for a valid TM-21 projection |
Use Compare2Best's comparison engine to compare products from verified manufacturers.
Start Comparing