CN106405442A - LED service life prediction method under actual operation environment - Google Patents

Abstract

The invention discloses an LED life prediction method based on an actual operating environment, and belongs to the technical field of LED product testing. The present invention establishes a single LED life model and a batch of LED life models with the same acceleration factor through the acceleration test report data, and obtains the real-time junction temperature feedback system diagram of the p-n junction, which is a key factor affecting the life of the LED, to obtain the real-time junction temperature feedback system of the LED temperature, the actual junction temperature of the LED and the batch LED life model are used to predict the life of the LED. The method of the time required for a certain proportion of LED failures under the confidence interval is beneficial to guide LED heat dissipation and system design, and improve the life and reliability of LED lighting products. This method can be extended to the life prediction of LED drivers and the entire LED lighting system.

Description

LED life prediction method based on actual operating environment

technical field

The invention discloses an LED life prediction method based on an actual operating environment, and belongs to the technical field of LED product testing.

Background technique

Compared with traditional light sources, LEDs have the advantages of high efficiency and long life, and have been widely used in recent years. At present, the lifespan of LED products on the market is usually rated as 25,000-50,000 hours, but users find that LED products often fail before the nominal lifespan, and cannot achieve high reliability operation. In addition to the LED product failure caused by the LED driver, the main reasons for the unequal life of the LED light source and the nominal life are: (1) The definition of LED life is relatively vague, and there is a lack of understanding of factors such as working conditions, failure standards, reliability, and confidence Consider; (2) The nominal life is usually tested under the specified current and temperature conditions, but the working conditions of the actual LED products are not constant, and even exceed the rated working conditions under severe working conditions.

At present, the definition of LED life mostly adopts the L _p life model, that is, when the output luminous flux is lower than p% of the initial value, the LED is considered to be invalid. Usually, the L ₇₀ standard is used for outdoor lighting, and the L ₉₀ standard is used for indoor general lighting. However, the tens of thousands of hours of L _p life cannot be obtained through experimental testing. According to the industry standard IES LM-80, LED manufacturers should provide test data of output luminous flux not less than 6000 hours, and the driving current is set to Typical operating current, at least three ambient temperatures: 55°C, 85°C and optional temperature. The life beyond the test time is the least square curve fitting of the luminous flux data in the LM-80 report, that is, the IES TM-21 standard, to predict the time required to reach the initial luminous flux p%. However, IES TM-21 uses averaged data, ignoring the life difference between samples, and cannot read reliability information. IES LM-80 only gives data under several test conditions, and cannot obtain life models and prediction methods under different working conditions. Therefore, correctly predicting the life of LEDs in the actual operating environment is conducive to guiding LED heat dissipation and system design, improving the life and reliability of LED lighting products, and truly realizing green lighting.

Contents of the invention

The purpose of the present invention is to address the shortcomings of the above-mentioned background technology and provide a LED life prediction method based on the actual operating environment, especially related to a method of actually predicting a certain proportion of LED failures under various confidence levels or confidence intervals under changing working environments The method of the required time realizes the LED life prediction under the changing working environment, and solves the technical problem that the average data is used to predict the LED life and ignores the life difference between samples, which leads to the inability to read the reliability information.

The present invention adopts following technical scheme for realizing above-mentioned purpose of the invention:

A method for predicting the lifetime of an LED based on an actual operating environment, comprising the following steps:

Step 1. According to the actual operating environment of the LED and the selected LED model, determine the ambient temperature T _A and the driving current I _F curve during operation, and select the LED life standard L _p according to the LED application, that is, the time it takes for the LED to reach the initial luminous flux p% .

Step 2. Determine a batch of LED lifetime models B _X :

2-1) The lifetime model L _p of a single LED obtained from the LED failure mechanism is:

Among them, A _p is a constant, n is a scaling factor, E _a is molecular activation energy, k _B is Boltzmann's constant, L _p (I _F , T _J ) is determined by the LED drive current I _F and junction temperature T _J The time required to reach the initial luminous flux p% under the stress level;

2-2) A batch of LED life model B _X , which means that within the working time of B _X , there is a failure rate of X% in this batch of LED products, that is, the output luminous flux of X% products drops to p% of the initial luminous flux, That is, the lifetime when the batch LED failure rate is X%,

Among them, the constant A _X , the molecular activation energy E _a , and the scaling factor n are to be found;

2-3) B _X has the same acceleration factor AF as L _p :

Among them, (I _F0 , T _J0 ) is the stress level determined by the initial driving current I _F0 and the initial junction temperature T _J0 , (I _F , T _J ) is the accelerated stress level in actual operation (driving current I _F and junction temperature T J0 _J determined stress levels). Using the IES LM-80 report data of LED products, select at least three sets of output luminous flux data under the stress level determined by different driving current and junction temperature conditions (I _F , T _J ), according to the L _p lifetime standard selected in step 1 Obtain at least three sets of _Lp data under the determined stress level of different drive current _IF and junction temperature _TJ ;

2-4) Make the _Lp data in step 2-3) obey the Weibull distribution, the cumulative failure probability curve F(t) of the Weibull distribution:

Among them, F(B _X )=X%, R(t) is a reliability function, t is time, η is a characteristic parameter based on the B _63.2 standard, that is, η=B _63.2 , β is a shape parameter, and the B _63.2 standard is expressed in The lifetime when the failure rate of batch LEDs is 63.2% at a stress level determined by a certain drive current and junction temperature;

2-5) The L _p data in step 2-3) is linearly fitted by the least squares method, and at least three different Weber curves are fitted according to the L _p life standard selected in step 1, and the readings are different. The drive current and junction temperature determine the β, η, and B _X values at the stress level;

2-6) Get three groups of stress levels and corresponding η values in step 2-5), and substitute it into the acceleration factor AF formula in step 2-3), wherein,

Constitute two sets of equations to obtain the values of molecular activation energy E _a and scaling factor n;

2-7) Put the molecular activation energy E _a , scaling factor n and B _X value under any set of ( _IF , T _J ) determined stress levels into the B _X lifetime model in step 2-2) to obtain A _X .

Step 3. Obtain the junction temperature curve of the LED in the actual operating environment.

3-1) Obtain the conduction voltage drop V _F of the LED at different package surface temperatures T _c , the LED radiant optical power P _opt , the thermal resistance Θ _jc of the LED from the internal pn junction to the package surface, and the driving force through experimental tests or device manuals. The relationship curve between the current I _F ;

3-2) Calculate the internal pn junction temperature T _J of the LED by the formula:

T _J (T _A , P _heat ,Θ _hs-a )=T _A +P _heat ·Θ _ja =T _A +P _heat ·(Θ _jc +Θ _c-hs +mΘ _hs-a ) (6),

Among them, the electric power P _e1 consumed by the LED = I _F V _F , the heat power P _heat emitted by the LED = P _e1 -P _opt , and its thermoelectric conversion coefficient k _h = P _heat /P _e1 =1-P _opt /P _e1 , m is the number of LEDs connected in series on the same radiator, Θ _c-hs is the sum of the thermal resistances of all devices between the LED surface and the radiator, and Θ _hs-a is the thermal resistance of the radiator;

3-3) Substitute various curve relationships in step 3-1) into step 3-2), construct the real-time junction temperature T _J feedback system diagram shown in Figure 10, and obtain the junction temperature curve of the LED in the actual operating environment.

Step 4. Calculating the life value B _X of the LED based on the actual operating environment.

4-1) Put the i-th stress level (T _J , I _F ) _i in the actual operating environment into the B _X life formula in step 2 to obtain the life of batch LEDs under the i-th stress level in a cycle operating environment B _Xi data;

4-2) Calculate the consumption life CL of batch LEDs in one operation cycle:

Among them, t _i is the cumulative working time of the LED under the (T _J , I _F ) _i stress level, and k is the total number of stress levels in one cycle of operation environment;

4-3) Calculation of LED life value B _X (hour) working in a cycle operating environment:

The size of the shape parameter β obtained under different environmental stresses in steps 2-5) should be the same.

The present invention adopts the above technical scheme, and has the following beneficial effects: a single LED life model and a batch LED life model with the same acceleration factor are established through the acceleration test report data; the key factor affecting the LED life is obtained by constructing a p-n junction internal junction temperature feedback system diagram Real-time junction temperature when LED is working; accurate prediction of LED life by actual junction temperature and batch LED life model, not only solves the problem that IES TM-21 uses averaged data to predict LED life and ignores the life difference between samples. It also solves the technical problem that the limited test conditions and data in the accelerated test cannot reflect the LED life under changing working conditions in the actual operating environment, which is conducive to guiding LED heat dissipation and system design, and improving Life and reliability of LED lighting products. This method can be extended to the lifetime prediction of LED drivers and the whole LED lighting system.

Description of drawings

Fig. 1 is the LED output luminous flux curve when I _F =0.35A, T _A =120°C, T _J =129°C.

Fig. 2 is the LED output luminous flux curve when I _F =0.7A, T _A =55°C, T _J =74°C.

Fig. 3 is the LED output luminous flux curve when I _F =1A, T _A =85°C, T _J =112°C.

FIG. 4 is the LED cumulative failure probability curve F(t) based on the lifetime criterion L ₉₀ .

FIG. 5 is an LED cumulative failure probability curve F(t) based on the lifetime criterion L ₇₀ .

Fig. 6 is a distribution diagram of B _X based on different driving current I _F and junction temperature T _J.

Fig. 7 is the V-I characteristic curve of LED.

Fig. 8 is a characteristic curve of the photoelectric conversion ratio P _opt /P _e1 of the LED.

Fig. 9 is a cumulative structure function diagram of the heat flow path.

Figure 10 is a diagram of the LED real-time junction temperature feedback system.

Figure 11 is the indoor lighting experiment curve based on the L ₉₀ standard.

Figure 12 is the experimental curve of outdoor street lighting based on the L ₇₀ standard.

detailed description

The technical solution of the invention will be described in detail below in conjunction with the accompanying drawings.

The experiment selects Luxeon K2 white light LED from Lumileds Company. Select three sets of output luminous flux data under different driving current and junction temperature conditions (I _F , T _J ) from its IES LM-80 report data, select the L _p lifetime standard, and obtain three sets of different driving current I _F and junction temperature The _Lp data under _TJ conditions are shown in Figure 1, Figure 2, and Figure 3. Make the L _p data obey the Weibull distribution, and use the least square method to linearly fit, and get three different Weibull curves. The size of the shape parameter β obtained under different environmental stresses (I _F , T _J ) and L _p life standards should be the same. From Fig. 4 and Fig. 5, the activation energy E _a , scaling factor n, characteristic parameter η, and B _X values under different failure rates of the B _X life model under different L _p standards can be obtained.

Fig. 6 is based on the life model B _X under the conditions of Fig. 4 and Fig. 5. The life standards L ₇₀ and L ₉₀ are selected. Taking the failure rate of 1% and 10% as examples, the life model is:

Select Luxeon K2 white light LED from Lumileds company, solder it to the PCB circuit board to realize electrical connection, the PCB board is attached to the heat sink, and use thermal interface material TIM (Thermal Interface Material) to fill the air gap between the two to achieve further heat conduction.

In order to accurately obtain the photoelectric and thermal characteristics of the LED in the experiment, T3Ster (Thermal Transient Tester) equipment was used for characteristic testing. The above-mentioned LED structure was removed from the radiator and placed on the metal plate of T3Ster, and the metal plate was used to replace the radiator. Adjust the temperature of the metal plate, that is, the temperature of the heat sink, and conduct experimental tests. Set up 36 groups of different stress levels for experiments, the driving current I _F is randomly distributed as 0.25A, 0.5A, 0.75A, 1A, and the radiator temperature T _hs is 10°C, 20°C to 90°C.

Using the T3Ster test system, the volt-ampere characteristic curve shown in Figure 7, the output photoelectric conversion ratio P _opt /P _e1 characteristic curve shown in Figure 8, and the cumulative structure function diagram of the heat flow path in Figure 9 can be obtained. Wherein, when the thermal capacitance C _th in Fig. 9 reaches positive infinity, the corresponding thermal resistance Θ _j-hs value is the sum of thermal resistances of all devices between the LED pn junction and the metal plate.

Substitute the three-dimensional diagrams of Figure 7, Figure 8, and Figure 9 into Figure 10 to construct a real-time LED junction temperature feedback system diagram. Figure 10 uses the current curve of an outdoor street lamp, and its ambient temperature curves are sampled from Aalborg, Denmark and Singapore. . The radiator temperature T _hs which is easy to measure is adopted here as the feedback signal. Use the junction temperature formula T _J =T _hs +P _heat ·Θ _j-hs to obtain the pn junction junction temperature T _J . Knowing the drive current I _F and the junction temperature T _J , from the life formula B _X , get all B _Xi , use the formula Get the life of LED in the actual operating environment and realize life prediction.

Select a single LED shown in Figure 7 for indoor lighting experiments, using the L ₉₀ standard, the driving current is 0.35A, the working time is from 19:00 to 24:00 every day, and the indoor ambient temperature is assumed to be a constant temperature of 22°C (half a year) and 26°C (half a year), using the above-mentioned life prediction steps, the relationship curve between the life and the heat sink thermal resistance shown in Figure 11 is obtained. It can be seen from FIG. 11 that the use of a heat sink with a good heat dissipation effect, that is, a small equivalent thermal resistance, can effectively improve the life of the LED light source. In order to achieve the same life of 50,000 hours, compared with the B ₁₀ life standard, the B ₁ life standard has higher requirements on the performance of the radiator, and its maximum equivalent thermal resistance is 51.5°C/W, while the B ₁₀ life standard can be used The maximum equivalent thermal resistance of the radiator is 76.2°C/W. Figure 11 can provide heat dissipation reference for LED lighting design. Each LED should select a suitable heat sink according to its actual operating environment and reliability requirements.

The lighting brightness of a single LED is difficult to meet the needs of most occasions. Here, 18 LEDs of the same type are connected in series to conduct street lighting experiments. The experimental locations are Aalborg, Denmark and Singapore. The ambient temperatures of the two places are shown in Figure 10. , the difference is large. The driving current of the LED is 0.7A, and the working time is from 19:00 to 5:00 the next day. Using the L ₇₀ standard, the predicted life B ₁₀ and thermal resistance curve are shown in Figure 12. It can be seen from Figure 12 that the ambient temperature of the LED has a great influence on its life, and the high temperature will shorten its life. To achieve the same 50,000-hour lifespan, the same street light requires different radiators in Singapore and Aalborg. Therefore, LED lighting should propose corresponding thermal management schemes according to different use environments to improve the life and reliability of LED lighting systems.

Claims (5)

1. Based on the LED life prediction method under the actual operating environment, it is characterized in that, Step A, according to the actual operating environment of the LED and its model, determine its environmental temperature change curve and driving current change curve, and select the LED life standard according to the LED application occasion; Step B. Establishing a single LED lifetime model L _p (I _F , T _J ) according to the LED failure mechanism: and batch LED lifetime model B _X (I _F ,T _J ): The single LED lifetime model and the batch LED lifetime model have the same acceleration factor AF(n, E _a ): L _p (I _F , T _J ) is the time required for the LED to reach the initial luminous flux p% under the stress level determined by the driving current I _F and the junction temperature T _J , A _p is a constant of the life model of a single LED, n is a scaling factor, E _a is the molecular activation energy, k _B is the Boltzmann constant, B _X (I _F , T _J ) is the output luminous flux of X% LEDs under the stress level determined by the driving current I _F and the junction temperature T _J of the batch LED The time required to drop to p% of the initial luminous flux, i.e., B _X (I _F , T _J ) is the life time when the batch LED failure rate is X% at the stress level determined by the driving current I _F and the junction temperature T _J , A _X is the constant of the batch LED life model, (I _F0 , T _J0 ) is the stress level determined by the initial drive current I _F0 and the initial junction temperature T _J0 , and the LED life under different stress levels is fitted according to the LED life standard selected in step A The Weibull distribution curve of the molecule is used to determine the molecular activation energy E _a , the scaling factor n, and the value of the constant A _X of the batch LED lifetime model; Step C, obtaining the junction temperature curve of the LED in the actual operating environment; Step D, predicting the lifetime of the batch LEDs according to the junction temperature curve of the LEDs in the actual operating environment and the lifetime model of the batch LEDs. 2. According to claim 1, based on the LED life prediction method under the actual operating environment, it is characterized in that, in step B, according to the LED life standard selected in step A, the Weber distribution curve of LED life under different stress levels is fitted to determine the molecular The activation energy E _a , scaling factor n, and the value of the constant A _X of the batch LED lifetime model, the specific method is as follows: From the IES LM-80 report data of LED products, select at least three sets of luminous flux data under different drive currents and junction temperatures to determine the stress level, and determine at least three sets of different drive currents and junction temperatures to determine the stress level according to the LED life standard selected in step A. The life data under; The cumulative failure probability curve F(t) is obtained from the life data described in Weibull distribution: F(B _X )=X%, R(t) is a reliability function, t is time, η is a characteristic parameter based on the B _63.2 standard, that is, η=B _63.2 , β is a shape parameter, and the B _63.2 standard indicates Lifetime at 63.2% failure rate of batch LEDs at stress levels determined by current and junction temperature; According to the LED life standard selected in step A, the least square method is used to linearly fit the life data to obtain at least three sets of Weber curves under the stress levels determined by different drive currents and junction temperatures, and read the stress levels under different drive currents and junction temperatures. The shape parameter β, the characteristic parameter η, and the life-span B _X when the batch LED failure rate is X%; According to the principle that the ratio of the acceleration factor and the ratio of the characteristic parameters under different driving currents and junction temperatures are equal to the ratio of the characteristic parameters, the molecular activation energy E _a and the scaling factor n are determined, and then combined with any set of driving currents and junction temperatures to determine the batch size under the stress level Lifetime at _X % LED Failure Rate Solve for the constant AX of the batch LED lifetime model. 3. According to claim 1 or 2, the LED life prediction method based on the actual operating environment is characterized in that, the specific method of step C is: Respectively obtain the relationship curve between the turn-on voltage drop V _F of the LED at different package surface temperatures T _c , the LED radiated light power P _opt , the thermal resistance Θ _jc of the LED from the internal pn junction to the package surface, and the driving current I _F ; The LED is determined by the expression: T _J (T _A , P _heat ,Θ _hs-a ) = T _A +P _heat ·Θ _ja =T _A +P _heat ·(Θ _jc +Θ _c-hs +mΘ _hs-a ) The junction temperature curve in the actual operating environment, where T _J is the junction temperature, T _A is the ambient temperature, P _heat is the thermal power emitted by the LED, P _heat = P _e1 -P _opt , P _e1 is the electric power consumed by the LED, P _e1 = I _F V _F , Θ _c-hs is the sum of the thermal resistance of all devices between the LED surface and the heat sink, Θ _hs-a is the heat sink resistance, m is the number of LEDs connected in series on the same heat sink. 4. according to claim 3, based on the LED life prediction method under the actual operating environment, it is characterized in that, the specific method of step D is: The stress level of the LEDs in the actual operating environment is obtained from the junction temperature curve of the LEDs obtained in step C under the actual operating environment, and then combined with the batch LED life model established in step B to determine the batch of LEDs under the i-th stress level in a cycle operating environment The lifetime B _Xi ; Calculate the consumption life CL of a batch of LEDs in one operating cycle: t _i is the accumulated working time of the LED under the i-th stress level in one operation cycle, and k is the total number of stress levels in one cycle of operation environment; The ratio of the working time of LEDs in one operating cycle to the consumption life of batch LEDs in one operating cycle is used as the predicted value of the life of batch LEDs in a cycle operating environment. 5 . The LED lifetime prediction method based on actual operating environment according to claim 2 , wherein the value of the shape parameter β under the stress level determined by the different driving currents and junction temperatures is the same.