TM-21….What it IS and IS NOT!
?TM-21 does notdetermine traditional life or “time to failure” of an LED Lighting system
–The useful life of an LED system has many components that need to be considered (lamp, driver, lens, etc.)
–LEDs degrade (like all light sources) but for potentially very long periods of time. Instead of outright failure, LEDs will eventually dim to a point that is too low to serve their purpose
?TM-21 doesproject the lumen maintenance of an LED source (package/array/module)
….. Which can then be used to project the expected lumen output of the source as part of a system (fixture)
15| Solid-State Lighting Program
www.ssl.energy.gov
TM-21 Method:
Normalize and Average All Data
?Use all data.
–LM-80 requires 6,000 hours with data at 1,000 hour increments. –TM-21 applies all test data at any increment
Sample # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Average 0 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.0000 500 0.970 0.987 0.984 0.990 0.981 0.988 0.990 0.988 0.989 0.982 0.977 0.988 0.985 0.976 0.985 0.977 0.966 0.998 0.985 0.975 0.9831 1000 2000 3000 4000 5000 6000 0.957 0.973 0.966 0.977 0.963 0.975 0.978 0.973 0.975 0.965 0.956 0.975 0.969 0.960 0.971 0.962 0.950 0.983 0.970 0.961 0.9680 0.962 0.976 0.967 0.980 0.969 0.979 0.978 0.974 0.978 0.964 0.960 0.980 0.971 0.966 0.978 0.969 0.954 0.989 0.976 0.967 0.9719 0.957 0.971 0.960 0.976 0.965 0.974 0.974 0.968 0.974 0.957 0.956 0.977 0.965 0.962 0.975 0.964 0.944 0.984 0.969 0.961 0.9667 0.950 0.967 0.954 0.970 0.959 0.968 0.962 0.962 0.968 0.948 0.950 0.970 0.956 0.957 0.969 0.958 0.938 0.977 0.963 0.952 0.9599 0.944 0.960 0.947 0.967 0.953 0.964 0.958 0.957 0.964 0.942 0.946 0.967 0.949 0.953 0.965 0.956 0.935 0.972 0.958 0.948 0.9553 0.947 0.960 0.949 0.965 0.953 0.966 0.954 0.955 0.966 0.936 0.946 0.961 0.945 0.953 0.966 0.952 0.937 0.971 0.957 0.944 0.9542 ?Normalize all data to 1 (100%) at 0 hours
?Average each point for all
samples of the device for each test condition
–Averaging is done for simplicity of application
–Variance associated with multiple samples is not
considered usable for projection 16| Solid-State Lighting Program
www.ssl.energy.gov
TM-21 Sample Size
?Suggested minimum sample size is set at 20 for each given temperature and drive current (matched with a 6-time multiplier)?Relative symmetry of known LED data supports normality of data at 20 units?Limited reduction in uncertainty at 30 units and severe increase at 10.17| Solid-State Lighting Program
www.ssl.energy.gov
TM-21 Use Latest Data for Curve Fit
?Initial data variability (i.e., “hump”) is
difficult for models to evaluate (0-1000 hr)?Later data exhibits more characteristic decay curve of interest
–Non-chip decay (encapsulant, etc.) occurs early and with varying effects on decay curve–Later decay is chip-driven and relatively consistent with exponential curve
–Verification with long duration data sets
(>10,000 hr) shows better model to reality fit with last 5,000 hours of 10,000 hour data
?For 6,000 hours of data (LM-80 minimum) and up to 10,000 hours: Use last 5,000 hours
?For > 10,000 hours: Use the last ? of the collected data
18| Solid-State Lighting Program
www.ssl.energy.gov
TM-21 Curve Fit and Projection
?Apply exponential least squares curve fit
Φ(t)= Bexp(-αt)
Where:t= operating time in hours;
Φ(t) = averaged normalized luminous flux output at time t;B= projected initial constant derived by the least squares curve-fit;α= decay rate constant derived by the least squares curve-fit.
?Project lumen maintenance “life”
Lp= ln(100*B/p)/α
Where:
Lp= lumen maintenance life in hours where pis the maintained percentage of initial lumen output
–Can accommodate user identified Lp(I.e. L70L50)–If Lpreached during testing –must use that value
19| Solid-State Lighting Program
www.ssl.energy.gov
DOE解读IES LM79,LM80和TM21的差异
