Lumen degradation analysis of LED lamps based on the subsystem isolation method

Abstract

The lumen degradation of LED lamps undergoing an accelerated aging test is investigated. The entire LED lamp is divided into three subsystems, namely, driver, lampshade, and LED light source. The parameters of output power [Watts (W)], transmittance (%), and lumen flux (lm) are adopted in the analysis of the degradation of the driver, lampshade, and LED light source, respectively. Two groups of LED lamps are aged under the ambient temperatures of 25°C and 85°C, respectively, with the aging time of 2000 h. The lumen degradation of the lamps is from 3.8% to 4.9% for the group under a temperature of 25°C and from 10.6% to 12.7% for the group under a temperature of 85°C. The LED light source is the most aggressive part of the three subsystems, which accounts for 70.5% of the lumen degradation of the LED lamp on average. The lampshade is the second degradation source, which causes 21.5% of the total amount on average. The driver is the third degradation source, which causes 6.5% under 25°C and 2.8% under 85°C of the total amount on average.

Full text

Lumen degradation analysis of LED lamps based
on the subsystem isolation method
HONG-LIANG KE,1,*JIAN HAO,1JIAN-HUI TU,1PEI-XIAN MIAO,1CHAO-QUAN WANG,1
JING-ZHONG CUI,1QIANG SUN,2AND REN-TAO SUN3
1Lanzhou Space Technology Institute of Physics, Science and Technology on Vacuum Technology and Physics Laboratory,
No.100 Feiyan Roud, Chengguan District, Lanzhou, Gansu Province 730000, China
2Department of Optoelectronics Research and Development Center, CIOMP-Chinese Academy of Science,
No. 3888 East South-Lake Road, Changchun, Jilin Province 130033, China
3College of Communication Engineering, Jilin University, No. 5988 Renmin Street, Changchun, Jilin Province 130022, China
*Corresponding author: pirlo2008snooker@126.com
Received 7 September 2017; revised 21 December 2017; accepted 30 December 2017; posted 2 January 2018 (Doc. ID 306558);
published 31 January 2018
The lumen degradation of LED lamps undergoing an accelerated aging test is investigated. The entire LED lamp
is divided into three subsystems, namely, driver, lampshade, and LED light source. The parameters of output
power [Watts (W)], transmittance (%), and lumen flux (lm) are adopted in the analysis of the degradation of the
driver, lampshade, and LED light source, respectively. Two groups of LED lamps are aged under the ambient
temperatures of 25°C and 85°C, respectively, with the aging time of 2000 h. The lumen degradation of the lamps
is from 3.8% to 4.9% for the group under a temperature of 25°C and from 10.6% to 12.7% for the group under a
temperature of 85°C. The LED light source is the most aggressive part of the three subsystems, which accounts for
70.5% of the lumen degradation of the LED lamp on average. The lampshade is the second degradation source,
which causes 21.5% of the total amount on average. The driver is the third degradation source, which causes
6.5% under 25°C and 2.8% under 85°C of the total amount on average. © 2018 Optical Society of America
OCIS codes: (000.2190) Experimental physics; (350.4800) Optical standards and testing.
https://doi.org/10.1364/AO.57.000849
1. INTRODUCTION
Recently, white LEDs converted by blue chip and yellow
phosphor have been developed owing to their high efficiency,
environmental benefits, and long lifetime [1–3]. To verify the
long-lifetime performance, Energy Star proposed a well-known
6000 h test for the LED light source [4], as well as the LED
lamps [5]. The LED lifetime of L70%, defined as the time for
luminous maintenance to drop to 70%, could be predicted by
TM-21-11 [6] with the lumen degradation acquired in the
6000 h test. However, 6000 h is not a long enough period
to catch up with the present LED production, and therefore
the thermal stress is always applied in the LED accelerated
aging test to shorten the test time in most studies [7–11].
The failure mechanism analysis and the lifetime prediction for
LED lamps under accelerated aging tests remain challenging
tasks. First, there are various unexpected failure modes at the sys-
tem level under different stress levels [12,13]. Second, the lifetime
spans of the subsystems always differ greatly, which makes the
lifetime of the LED lamp limited to the worst subsystem.
In recent years, some researchers [14,15] divided the LED
lamp into several subsystems, which are the driver, the LED
light source, and the lampshade and fixture. The analysis for
the different subsystems is conducted. For the LED light
source, studies [16–20] focused on the reliability tests and life-
time prediction of LED packages/modules. Yoon et al.[21]
made a comparison between LED packages and LED lamps
in the reliability analysis. It was shown that the shape param-
eters of the Weibull distribution in the two cases were, respec-
tively, 8.87–11.12 and 14.87–19.82, which means the failure
mechanism was different. IES LM-82 [22] pointed out that the
thermal condition of the LED light source should be consid-
ered when it was applied in a certain lighting system and differ-
ent thermal conditions corresponding to different reliabilities.
As a result, the lifetime of the LED lamp cannot be simply
taken as that of the LED light source.
For the LED driver, De Santi et al.[23] indicated that the
output power of the driver decreased over time, and one of the
sample’s power decreased by 5% after 2000 h aging under an
ambient temperature of 40°C. Sun et al.[24] pointed out that
the output power of the LED driver with isolated components
decreased by the same rate in a 300 h aging test under ambient
temperatures of 55°C and 105°C. Therefore, the decrease of the
Research Article Vol. 57, No. 4 / 1 February 2018 / Applied Optics 849
1559-128X/18/040849-06 Journal © 2018 Optical Society of America

output power of the LED driver results in a decrease of the
lumen flux of the LED lamp.
For the LED lampshade, mostly made of polymethyl
methacrylate (PMMA), Lu et al.[25] investigated the failure
mechanism of PMMA under different thermal stresses. The
results showed that the transmittance of PMMA decreased
significantly in the wavelength band from 380 nm to 730 nm,
causing a 10.2% decrease of the lumen flux for 360 h of aging
under an ambient temperature of 55°C. The degradation of the
lampshade is the darkening of PMMA, which reduces the trans-
mittance of PMMA and is therefore responsible for the decrease
of the luminous flux as well as the color shift of the LED lamp.
The main purpose of this research is to investigate the deg-
radation of each subsystem of the LED lamp, namely the driver,
lampshade, and LED light source, under the thermal acceler-
ated aging test. The variation in the output power (W) of the
driver, the variation in transmittance (%) of the lampshade, and
the variation in the lumen flux (lm) of the LED light source are
taken as the parameters for evaluating the degradation of each
subsystem. The proportion of lumen degradation of the LED
lamp caused by each subsystem is given for comparison. The
second purpose is to investigate the difference in the degrada-
tion of the subsystems under different thermal stress levels, and
the normal temperature of 25°C and the elevated temperature
of 85°C are applied in this research.
2. THEORETICAL ANALYSIS
The total lumen degradation of the LED lamp over time Dall is
divided into the three parts D1,D2, and D3, which are caused
by the driver, the lampshade, and the LED light source,
respectively, as given by
Φ0−ΦtDall D1D2D3;(1)
where Φ0is the initial lumen flux of the LED lamp before aging
and Φtis the lumen flux after an aging time t.
For the LED driver, the correspondent lumen degradation
of D1can be evaluated by the variation in its output power
using
D1W0−Wt×μ;(2)
where W0and Wtare, respectively, the output power of the
LED driver before and after thours of aging and μis the varia-
tion rate of the lumen flux of the LED light source with respect
to the output power of the driver.
For the lampshade, the correspondent lumen degradation of
D2is evaluated by the variation of the transmittance of Tλ,
which is given by
TλSPD1λ∕SPD2λ;(3)
where SPD1λand SPD2λare the spectral power distribu-
tion (SPD) of the LED lamp with and without the lampshade,
respectively. Then D2can be calculated by
D2Z780
380
Km×T0λ−Ttλ ×Vλ×SPDtdλ;(4)
where T0λand Ttλare the transmittance of the lampshade
before and after thours of aging, respectively. Kmis 683 lm/W,
Vλis the vision function under photopic vision, and SPDtis
the spectral power distribution of the LED lamp without the
lampshade after thours of aging. The visible wavelength band
from 380 nm to 780 nm is adopted.
For the LED light source, the correspondent lumen degra-
dation of D3is evaluated by the degradation of the LED itself,
which is expressed as
D3Φ0
0−Φ0
t;(5)
where Φ0
0and Φ0
tare the lumen flux of the LED light source
under the rated current before and after thours of aging, re-
spectively. With the acquired Dall,D1,D2, and D3, the
proportion of the lumen degradation of each subsystem is
obtained and compared.
3. EXPERIMENTS
A. Test Samples
The subsystems of the driver, the lampshade, the LED light
source, the heat sink, and the lampholder E27 of the tested
LED lamp in this research are shown in Fig. 1. The driver
is placed outside the LED lamp for the measurement of the
output power. The main parameters of each subsystem are
listed in Table 1.
Six LED lamps from the same batch are divided into two
groups, with samples 1, 2, and 3 as the first group and samples
4, 5, and 6 as the second group. The samples in the first group
are aged under the normal temperature of 25°C while the sam-
ples in the second group are aged under the elevated temper-
ature of 85°C. The total aging time for both groups is 2000 h,
and the parameters measurement for each subsystem is taken
before and after the aging process.
B. Test for LED Lamp and Subsystems
The experimental apparatus is shown in Fig. 2. The LED lamp
is fixed inside a temperature chamber, and outside the chamber
Fig. 1. Subsystems of the LED lamp.
Table 1. Main Parameters of Each Subsystem
Composition
Thirty 0.15 W, GaN-Based White
LEDs Converted by Y3Al5O12:Ce
LED
source Rated input/output
135 mA (DC)/340 lm,
3500 K
Driver Rated input/output 220V(AC)/135 mA (DC)
Lampshade Material PMMA
Heat sink Material Ceramics
850 Vol. 57, No. 4 / 1 February 2018 / Applied Optics Research Article

a 1.5 m integrating sphere connected to a spectrometer is used
for the collection of optical parameters of the LED lamp. The
side opening of the integrating sphere is closely connected to
the window of the chamber. To achieve thermal isolation
between the integrating sphere and the chamber, a special hood
made of vacuum glasses is placed at the window.
By using this measurement system, the optical parameters of
the LED lamp under different thermal conditions can be
measured. First, different ambient temperatures are controlled
by the chamber, and the correspondent thermal condition of
the LED lamp is monitored by a thermal infrared imager fixed
on the top of the chamber. Second, the lamp is pushed into the
integrating sphere to ensure that the measurement is with
highest lumen flux of the sample. Finally, the output of the
LED lamp under a certain thermal condition is given by the
spectrometer.
The measurement steps for the LED lamp, driver,
lampshade, and LED light source are given as follows.
Step-1 for the LED lamp: The chamber temperature is con-
trolled to 25°C, and therefore the samples in the second group
after 2000 h of aging must be fully cooled down before testing.
The tested sample is fixed inside the chamber and preheated for
20 min before the measurement. The LED driver is placed
outside the chamber and powered by 220 V (AC). The lumen
flux of Φand the spectral power distribution of SPD1of LED
lamp are then measured by the spectrometer.
Step-2 for the driver: During the optical parameter measure-
ment in step-1, the stable output power (W) of the driver is
simultaneously measured with a digital multimeter outside
the chamber.
Step-3 for the lampshade: In the following steps, the lampshade
is removed from the tested lamp, and thereby the thermal
condition and the junction temperature of the LED lamp
inevitably vary. Cai et al.[13] indicated that the variation in
the temperature of the LED heat sink could be 8–10°C, before
and after removing the lampshade. Figure 3shows the compari-
son of the infrared image of the LED lamp with and without
the lampshade, under the normal working conditions. Note
that at the ambient temperature of 25°C, the heat sink temper-
ature (TH) of the LED lamp is 62°C in the case with the lamp-
shade, and it is 54°C in the case without the lampshade.
Therefore, when the lampshade is removed, the chamber tem-
perature should be adjusted to ensure the THis unchanged.
Then the sample is pushed into the integrating sphere, and
the spectral power distribution of SPD2is given by the spec-
trometer. In our previous research in Ref. [20], the LED junc-
tion temperature measured by the forward voltage method
shows an upward trend over the aging time, which gives a varia-
tion of about 6°C–8°C after 3000 h of aging under an ambient
temperature of 80°C. Therefore, the thermal condition of the
LED lamp should be tested and adjusted during the aging
process. It is emphasized that the integrating sphere system
should be re-calibrated with the auxiliary lamp before the
measurement, due to the change of the tested target.
Step-4 for LED light source: The LED driver and AC power are
substituted by a DC power, and the rated current of 135 mA
(DC) is applied to the sample. The lampshade is removed, and
the chamber temperature is adjusted as described in step-3.
After 20 min preheating, the lumen flux of Φ0is measured.
In this way the influence of the LED driver and the lampshade
on the lumen degradation is excluded, and the variation of Φ0
over 2000 h can be regarded as the degradation of the LED
light source itself.
4. RESULTS AND ANALYSIS
A. Lumen Degradation of LED Lamp Dall
Figure 4shows the difference of the lumen flux measured be-
fore and after 2000 h of aging for the six samples in step-1.
Note that the degradation of Dall is 13.9 lm (4.5%), 11.9 lm
(3.8%), 14.7 lm (4.9%), 38.4 lm (12.7%), 37.0 lm (11.8%),
Fig. 2. Sketch map of the measurement system.
Fig. 3. Comparison of the infrared image of the LED lamp with
lampshade and that without lampshade.
Fig. 4. Comparison of the lumen flux measured in Step-1 at 0 h and
2000 h.
Research Article Vol. 57, No. 4 / 1 February 2018 / Applied Optics 851

and 33.1 lm (10.6%), respectively, for the six samples after
2000 h of aging.
B. Lumen Degradation of Driver D1
To evaluate the effect of the degradation of the driver on the
lumen degradation of the LED lamp, the variation rate of the
lumen flux of the LED light source with respect to the output
power of the driver μshould be determined first. Figure 5
shows the experimental results of the lumen flux under differ-
ent driver output power for sample 1, before and after 2000 h
of aging. Note that the lumen flux can be linearly fitted for both
cases with R-square (coefficient of multiple determination)
higher than 0.99. The slope of the fitting for the data before
aging is 56.71 (lm/W), and it is 57.11 (lm/W) for the data after
2000 h of aging. As a result, the average value of 56.91 is taken
as the variation rate of μfor sample 1. The μvalues for the other
five samples are, respectively, 55.32, 57.66, 60.35, 59.62, and
55.30 (lm/W).
Table 2lists the output power of the driver before and after
the 2000 h of aging measured in step-2, and the decreased out-
put power of ΔWfor the six samples. Note that the decreased
output power of the driver over 2000 h aging at a normal tem-
perature of 25°C is 0.015 W on average, and it is 0.018 Wat
elevated temperatures of 85°C. This indicates that an elevated
temperature of 85°C has little effect on the degradation of the
driver. With the decreased output power and the μvalue, the
lumen degradation of D1caused by the driver is calculated
according to Eq. (2). The values of D1for the samples are listed
in the last column of Table 2. Note that the averaged value of
Fig. 5. Lumen flux as a function of output power for sample 1.
Table 2. Degradation of Driver D1
Aging
Temperature
(°C)
Power (W)
at 0 h
Power (W)
at 2000 h
ΔPower
(W)
μ
lm/W D1lm
25
No. 1 4.725 4.711 0.014 56.91 0.81
No. 2 4.635 4.620 0.015 55.32 0.83
No. 3 4.821 4.804 0.017 57.66 0.98
85
No. 4 4.748 4.548 0.020 60.35 1.20
No. 5 4.622 4.607 0.015 59.62 0.88
No. 6 4.535 4.517 0.018 55.30 1.00
Fig. 6. (a) SPD1and SPD2of sample 1 at 0 h and 2000 h. (b) The
transmittance of sample 1 at 0 h and 2000 h. (c) The transmittance
reduction of sample 1 after 2000 h of aging. (d) The transmittance
reduction of the other five samples.
852 Vol. 57, No. 4 / 1 February 2018 / Applied Optics Research Article

D1is 0.87 lm for the samples aging at 25°C, and is 1.03 lm for
the samples aging at 85°C.
C. Lumen Degradation of Lampshade D2
To investigate the transmittance of the lampshade, the SPD1in
step-1 and the SPD2in step-3 are compared. Figure 6(a) shows
the SPD1and SPD2of sample 1 at 0 h and after 2000 h of
aging, and Fig. 6(b) shows the calculated transmittance T0
(before aging) and Tt(after 2000 h of aging) according to
Eq. (3). Apparently, Ttis lower than T0at each wavelength
over the visible band. The transmittance reduction (T0−Tt)
is shown in Fig. 6(c). Note that as the wavelength increases,
T0−Ttdecreases within the band from 400 nm to 500 nm
and becomes steady with the value of 1.1–1.4% within the
band from 500 nm to 800 nm. Figure 6(d) shows the calculated
T0−Ttfor the other five samples. Note that the values of
T0−Ttare larger for the samples under an aging temperature
of 85°C compared with those under 25°C, indicating that the
degradation of the lampshade strongly correlates to the elevated
temperature.
With the obtained T0,Tt, and SPD2at 2000 h, the cor-
respondent lumen degradation of D2caused by the lampshade
is calculated according to Eq. (4). The values of D2are listed in
Table 3, which are from 2.4 to 3.1 lm for samples 1, 2, and 3,
and are from 7.6 to 8.1 lm for samples 4, 5, and 6.
D. Lumen Degradation of LED Light Source D3
The lumen flux of the LED light sources of the sample is ac-
quired in step-4. The values of Φ0(before aging) and Φ0
t(after
2000 h of aging) are listed in the third and fourth row, respec-
tively, of Table 4. According to Eq. (5) the correspondent
lumen degradations of D3are calculated and listed in the last
row of Table 4. Note that the lumen degradation caused by the
LED light source over 2000 h of aging at a normal temperature
of 25°C is 9.3 lm on average, and it is 26.0 lm at an elevated
temperature of 85°C. The latter is about 2.8 times larger than
the former, implying that thermal stress plays an important role
in the degradation of the LED light source.
E. Analysis of Dall ,D1,D2, and D3
For sample 1, the lumen degradation of the LED lamp of Dall is
13.9 lm after 2000 h of aging at a temperature of 25°C (see
Fig. 4). The lumen degradation caused by the driver of D1
is 0.81 lm (see Table 2), which accounts for 5.8% of Dall. The
lumen degradation caused by the lampshade of D2is 3.0 lm
(see Table 3), which accounts for 21.7% of Dall. The lumen
degradation caused by the LED light source of D3is 9.6 lm
(see Table 4), which accounts for 69.1% of Dall. It is noticed
that the sum of D1,D2, and D3is 96.6% of Dall. The remain-
ing 3.4% of the total lumen degradation is caused by the mea-
surement errors and some interactions among each subsystem,
which can result in a faster degradation of the whole LED lamp.
For example, the lighting by the LED light source has a direct
effect on the degradation of the lampshade, and the degradation
of the light source therefore effects the degradation of the lamp-
shade. However, the degradation value caused by this effect
cannot be detected in step-3. Similar interactions among the
three subsystems contribute to the remaining 3.4% of the total
lumen degradation.
Table 5lists the results for the six samples. These results
show that the lumen degradation of the aged LED lamp is
mainly due to the degradation of the LED light source,
which causes about 70.5% of the total amount on average.
The lampshade is the second degradation source, which causes
about 21.3% of the total amount on average. The driver is the
third degradation source, which causes about 6.5% (at a normal
temperature of 25°C) and about 2.8% (at an elevated
temperature of 85°C) of the total amount on average.
5. CONCLUSIONS
To investigate the degradation of each subsystem of the LED
lamp in the aging test, the LED lamp is divided into the three
subsystems of driver, lampshade, and LED light source. Two
groups of aging tests, one test at a normal temperature of 25°C
and the other at an elevated temperature of 85°C, are
conducted for 2000 h.
It is shown that the decreased output power of the LED
driver is 0.015 W on average for the aging test at a normal
temperature of 25°C, and it is 0.018 W for the aging test at
an elevated temperature of 85°C. The small difference indicates
that the elevated temperature of 85°C has little effect on the
driver degradation. The degradation of the driver causes a
Table 3. Degradation of Lampshade D2
Aging Temperature (°C) 25 85
No. 123456
D2(lm) 3.0 2.4 3.1 8.1 7.6 7.6
Table 4. Degradation of LED Light Source D3
Aging Temperature
(°C) 25°C85°C
No. 123456
Φ0at 0 h (lm) 335.2 342.5 338.0 339.3 341.8 336.4
Φ0
tat 2000 h (lm) 325.3 334.0 327.6 311.0 314.3 312.2
D3(lm) 9.6 8.3 10.1 27.6 26.8 23.6
Table 5. D1∕Dall,D2∕Dall , and D3∕Dall for Six Samples
Aging Temperature (°C) Driver D1∕Dall Lampshade D2∕Dall LED Light Source D3∕Dall Remaining 1−D1D2D3∕Dall
25
No. 1 5.8% 21.7% 69.1% 3.4%
No. 2 6.9% 20.3% 69.7% 3.0%
No. 3 6.7% 21.4% 68.7% 3.2%
85
No. 4 3.1% 21.1% 71.8% 4.0%
No. 5 2.3% 20.5% 72.4% 4.8%
No. 6 3.0% 23.0% 71.3% 2.7%
Research Article Vol. 57, No. 4 / 1 February 2018 / Applied Optics 853

correspondent lumen degradation of 6.5% (at a normal tem-
perature of 25°C) and 2.8% (at an elevated temperature of
85°C) of the total amount on average.
The degradation of the lampshade, coming from the
changing of transmittance, causes a correspondent lumen
degradation of 21.3% of the total amount on average.
The degradation of the LED light source causes a corre-
spondent lumen degradation of 70.5% of the total amount
on average, meaning that the degradation of the LED lamp
is mainly attributed to the LED light source.
The elevated temperature of 85°C plays an important role in
the degradation of the LED light source and lampshade, with
the lumen degradation approximately three times that at a
normal temperature of 25°C in this research.
As a starting point, the scope of this study is limited to the
LED lamp used in this research. For other kinds of LED lamps
with different driver circuits, lamp structures, materials, chips,
phosphors, and so on, the worst subsystem should be experi-
mentally determined, but we do believe that the proposed
subsystem isolation method can provide a useful way for
estimating the reality of LED lamps, especially with current
multifarious LED lighting products.
Funding. Chinese Academy of Sciences (CAS), Cui-Can
Project (KZCC-EW-102); CAS, 863 Project (2013AA03A116,
2015AA03A101).
REFERENCES
1. C.-C. Tsai, W.-C. Cheng, J.-K. Chang, S.-Y. Huang, J.-S. Liou, G.-H.
Chen, Y.-C. Huang, J.-S. Wang, and W.-H. Cheng, “Thermal-stability
comparison of glass- and silicone-based high-power phosphor-con-
verted white-light-emitting diodes under thermal aging,”IEEE
Trans. Device Mater. Rel. 14,4–8 (2014).
2. H.-L. Ke, L. Jing, Q. Gao, Y. Wang, J. Hao, Q. Sun, and Z.-J. Xu,
“Comparison of online and offline test in LED accelerated reliability
test under temperature stress,”Appl. Opt. 54, 9906–9910 (2015).
3. J. Hao, Q. Sun, L. Jing, Y. Wang, J. Zhao, H. Zhang, H. Ke, Q. Gao, X.
Wang, and Y. Zhang, “Comparison of lifetime predictions with LED
lamps and light source modules in accelerated aging tests,”in
17th International Conference on Thermal, Mechanical and
Multi-Physics Simulation and Experiments in Microelectronics and
Microsystems (2016), paper 11.
4. IESNA Testing Procedures Committee, LM-80-08: Measuring Lumen
Maintenance of LED Light Sources (Illuminating Engineering Society,
2008).
5. “ENERGY STAR@Program Requirements for Integral LED
Lamps–Partner Commitments,”2011.
6. R. Tuttle, K. Haraguchi, M. Hodapp, J. Jiao, C. Miller, Y. Ohno, T.
Pulsipher, E. Radkov, Er. Richman, and D. Szombatfalvy,
TM-21-11: Projecting Long Term Lumen Maintenance of LED Light
Source (Illuminating Engineering Society, 2012).
7. Y. Gu and N. Narendran, “White LED performance,”Proc. SPIE 5530,
119–124 (2004).
8. Y.-F. Su, S.-Y. Yang, T.-Y. Hung, C.-C. Lee, and K.-N. Chiang, “Light
degradation test and design of thermal performance for high-power
light-emitting diodes,”Microelectron. Reliab. 52, 794–803 (2012).
9. J. Hao, D. Li, C. He, Q. Sun, and H. Ke, “Step-down accelerated aging
test for LED lamps based on Nelson models,”Optik 149,69–80
(2017).
10. C. Qian, X. J. Fan, J. J. Fan, C. A. Yuan, and G. Q. Zhang, “An
accelerated test method of luminous flux depreciation for LED
luminaires and lamps,”Reliability Eng. Syst. Saf. 147,84–92 (2016).
11. S. Koh, C. Yuan, B. Sun, B. Li, X. Fan, and G. Q. Zhang, “Product level
accelerated lifetime test for indoor LED luminaires,”in 14th
International Conference on Thermal, Mechanical and Multi-Physics
Simulation and Experiments in Microelectronics and Microsystems,
EuroSimE2013 (2013).
12. W. D. van Driel and X. J. Fan, Solid State Lighting Reliability:
Components to System (Springer, 2012).
13. M. Cai, D. G. Yang, S. Koh, C. A. Yuan, W. B. Chen, B. Y. Wu, and
G. Q. Zhang, “Accelerated testing method of LED luminaries,”in 14th
International Conference on Thermal, Mechanical and Multi-Physics
Simulation and Experiments in Microelectronics and Microsystems,
EuroSimE 2013 (2013).
14. M. Cai, D. Yang, K. Tian, P. Zhang, X. Chen, L. Liu, and G. Zhang,
“Step-stress accelerated testing of high-power LED lamps based on
subsystem isolation method,”Microelectron. Reliab. 55, 1784–1789
(2015).
15. X. P. Li, L. Chen, and M. Chen, “An approach of LED lamp system
lifetime prediction,”in IEEE International Conference on Quality and
Reliability (ICQR) (2011).
16. M. Cai, W. B. Chen, L. L. Liang, M. Gong, W. C. Tian, H. Y. Tang, S.
Koh, C. A. Yuan, Z. Zhang, G. Q. Zhang, and D. G. Yang, “Highly
accelerated life testing of LED luminaries,”in International Conference
on Electronic Packaging Technology & High Density Packaging
(2012).
17. N. Narendran and Y. Gu, “Life of LED-based white light sources,”
J. Disp. Technol. 1, 167–171 (2005).
18. N. Narendran, Y. Gu, J. P. Freyssinier, H. Yu, and L. Deng, “Solid-
state lighting: Failure analysis of white LEDs,”J. Cryst. Growth
268, 449–456 (2004).
19. N. Narendran, L. Deng, R. M. Pysar, Y. Gu, and H. Yu, “Performance
characteristics of high-power light-emitting diodes,”Proc. SPIE 5187,
267–275 (2004).
20. H.-L. Ke, L. Jing, J. Hao, Q. Gao, Y. Wang, Q. Sun, and Z.-J. Xu,
“Analysis of junction temperature and modification of luminous flux
degradation for white LEDs in a thermal accelerated reliability test,”
Appl. Opt. 55, 5909–5916 (2016).
21. Y. G. Yoon, J. P. Hyung, U. H. Jeong, H. W. Lim, and J. S. Jang,
“Life time comparison of LED package and the self-ballasted LED
lamps by simple linear regression analysis,”Microelectron. Reliab. 55,
1779–1783 (2015).
22. Illuminating Engineering Society, LM-82-12 Approve Method:
Characterization of LED Light Engines and LED Lamps for Electrical
and Photometric Properties as a Function of Temperature
(Illuminating Engineering Society, 2012).
23. C. De Santi, M. Dal Lago, M. Buffolo, D. Monti, M. Meneghini, G.
Meneghesso, and E. Zanoni, “Failure causes and mechanisms of
retrofit LED lamps,”Microelectron. Reliab. 55, 1765–1769 (2015).
24. B. Sun, S. W. Koh, C. Yuan, X. Fan, and G. Zhang, “Accelerated life-
time test for isolated components in linear drivers of high-voltage LED
system,”in 14th International Conference on Thermal, Mechanical
and Multi-Physics Simulation and Experiments in Microelectronics
and Microsystems, EuroSimE 2013 (2013).
25. G. Lu, W. D. van Driel, X. Fan, M. YazdanMehr, J. Fan, C. Qian, K. M.
B. Jansen, and G. Q. Zhang, “Colour shift and mechanism
investigation on the PMMA diffuser used in LED-based luminaires,”
Opt. Mater. 54, 282–287 (2016).
854 Vol. 57, No. 4 / 1 February 2018 / Applied Optics Research Article