LED Lumen Depreciation and L70: How Driver Stability Affects Long-Term Light Output

Every LED product sheet carries a lifetime number, usually expressed as L70 followed by hours. L70 means the time at which light output has decayed to 70% of its initial value. The number is not a marketing guess; it is the output of a standardized calculation chain starting with IES LM-80 testing and ending with an IES TM-21 projection. Understanding that chain, and where the driver fits into it, helps separate honest lifetime claims from optimistic ones.

LM-80, formally Approved Method: Measuring Luminous Flux and Color Maintenance of LED Packages, Arrays and Modules, prescribes how to age LEDs under controlled conditions. Samples operate at three case temperatures, typically 55, 85, and a manufacturer-chosen third point, for a minimum of 6,000 hours with 10,000 preferred. At 1,000-hour intervals, luminous flux and chromaticity are measured. The result is a table showing the percentage of initial flux remaining at each hour point, for each temperature.

TM-21 takes those data points and fits an exponential decay model of the form Phi(t) = Phi_0 times e raised to negative alpha times t, where Phi_0 is initial flux and alpha is the degradation rate. The model is fitted to the last 5,000 hours of LM-80 data using least-squares regression on the natural logarithm of normalized flux. Solving for the time at which Phi(t) equals 0.70 times Phi_0 gives the L70 lifetime. L80 and L50 are calculated the same way.

TM-21 imposes one critical discipline: the six-times rule. The projected lifetime may not exceed six times the actual test duration. A manufacturer testing 6,000 hours can claim L70 of no more than 36,000 hours, regardless of what the curve suggests. To claim 50,000 hours, LM-80 must run at least 8,334 hours. This rule exists because exponential extrapolation becomes unreliable beyond a certain horizon, and it gives specifiers a way to spot inflated claims: if a sheet says L70 100,000 hours but the LM-80 report shows only 6,000 hours, the number violates the standard.

Where does the driver enter this picture? LM-80 measures the LED source in isolation, operated under ideal constant-current conditions with precise temperature control. In a real luminaire, the LED is powered by a driver and mounted in a thermal environment far from ideal. The driver's output quality directly affects the stress the LED experiences, and stress is what drives degradation. Consider current ripple: a driver with 30% peak-to-peak ripple at 100Hz drives LED current from 70% to 130% of nominal 100 times per second. Junction temperature follows with a small but real swing. Thermal cycling, even at high frequency and small amplitude, accelerates degradation of the phosphor layer and die-attach material. Over tens of thousands of hours, a high-ripple driver can cause the LED to depreciate faster than its LM-80 data predicts. Ripple below 10% keeps the LED operating close to the steady-state conditions under which LM-80 was measured.