How to predict LED lifespan

How to predict LED lifespan

LEDs also have a certain lifespan. Early LEDs were just gifts such as flashlights and table lamps, which were not used for long and had no prominent lifespan issues. But now LEDs have begun to be widely used in outdoor and indoor lighting, especially high-power LED street lamps, which have high power, high heat, and long working hours, and their lifespan issues are very prominent. The myth that LEDs must have a lifespan of 100,000 hours seems to have been completely shattered. So what exactly is the problem?

If the power supply and driver failures are not considered, the life of an LED is reflected in its light decay, that is, the brightness becomes dimmer and dimmer over time until it finally goes out. The life is usually defined as the time it takes to decay by 30%.

So can the life of LEDs be predicted? This question cannot be answered simply and requires a discussion from the beginning.

1. LED light decay:

Most white LEDs are made by irradiating yellow phosphor with blue LED. There are two main reasons for LED light decay. One is the light decay of blue LED itself, which decays much faster than red, yellow and green LED. Another is the light decay of phosphor, which decays very seriously at high temperature. The light decay of LEDs of different brands is different. Usually LED manufacturers can provide a set of standard light decay curves. For example, the light decay curve of Cree in the United States is shown in Figure 1.

Figure 1. Light decay curve of Cree's LED

As can be seen from the figure, the light decay of LED is related to its junction temperature. The so-called junction temperature is the temperature of the semiconductor PN junction. The higher the junction temperature, the earlier the light decay occurs, that is, the shorter the life. As can be seen from the figure, if the junction temperature is 105 degrees, the life of the LED will only be more than 10,000 hours when the brightness drops to 70%, 20,000 hours at 95 degrees, and 50,000 hours at 75 degrees, and it can be extended to 90,000 hours at 65 degrees. Therefore, the key to extending the life is to reduce the junction temperature. However, these data are only suitable for Cree's LEDs. They are not suitable for LEDs from other companies. For example, the light decay curve of Lumiled's LuxeonK2 is shown in Figure 2.

Figure 2. Lumiled's LuxeonK2 light decay curve

When the junction temperature increases from 115°C to 135°C, the lifetime decreases from 50,000 hours to 20,000 hours.

The light attenuation curves of other companies can be obtained from the original manufacturers.

2. How to prolong the life of LED

From the figure, we can conclude that the key to prolonging its life is to reduce its junction temperature. And the key to reducing the junction temperature is to have a good heat sink, which can dissipate the heat generated by the LED in time.

Here we are not going to discuss how to design a heat sink, but which heat sink has a better heat dissipation effect. In fact, this is a question of measuring the junction temperature. If we can measure the junction temperature that any heat sink can reach, we can not only compare the heat dissipation effects of various heat sinks, but also know the LED life that can be achieved after using this heat sink.

3. How to measure junction temperature

Junction temperature seems to be a temperature measurement problem, but the junction temperature to be measured is inside the LED, and you can't just put a thermometer or thermocouple into the PN junction to measure its temperature. Of course, its shell temperature can still be measured with a thermocouple, and then based on the given thermal resistance Rjc (junction to shell), its junction temperature can be inferred. However, after the heat sink is installed, the problem becomes complicated again. Because the LED is usually soldered to the aluminum substrate, and the aluminum substrate is installed on the heat sink, if you can only measure the temperature of the heat sink shell, then you must know the values ​​of many thermal resistances to infer the junction temperature. Including Rjc (junction to shell), Rcm (shell to aluminum substrate, in fact, it should also include the thermal resistance of the film printed board), Rms (aluminum substrate to heat sink), Rsa (heat sink to air), as long as one of the data is inaccurate, it will affect the accuracy of the test. Figure 3 shows a schematic diagram of the thermal resistances from the LED to the heat sink. Many thermal resistances are combined, which makes its accuracy more limited. In other words, the accuracy of inferring the junction temperature from the measured heat sink surface temperature is even worse.

Figure 3. Schematic diagram of thermal resistance from LED to heat sink

Fortunately, there is an indirect way to measure temperature, that is to measure voltage. So what voltage is the junction temperature related to? What is the relationship?

We must first start with the volt-ampere characteristics of LEDs.

4. Temperature coefficient of LED volt-ampere characteristics

We know that LED is a semiconductor diode. Like all diodes, it has a volt-ampere characteristic. Like all semiconductor diodes, this volt-ampere characteristic has a temperature characteristic. Its characteristic is that when the temperature rises, the volt-ampere characteristic shifts to the left. Figure 4 shows the temperature characteristic of the volt-ampere characteristic of LED.

Figure 4. Temperature characteristics of LED volt-ampere characteristics

Assuming that the LED is powered by a constant current of Io, when the junction temperature is T1, the voltage is V1, and when the junction temperature rises to T2, the entire volt-ampere characteristic shifts to the left, the current Io remains unchanged, and the voltage becomes V2. The two voltage differences are divided by the temperature, and its temperature coefficient can be obtained, expressed in mV/oC. For ordinary silicon diodes, this temperature coefficient is about -2mV/oC. However, most LEDs are not made of silicon materials, so its temperature coefficient must also be measured separately. Fortunately, most of the data sheets of various LED manufacturers give its temperature coefficient. For example, for Cree's XLamp7090XR-E high-power LED, its temperature coefficient is -4mV/oC. It is twice as large as ordinary silicon diodes. The temperature coefficient of the volt-ampere characteristic of LuxeonRebel from Philips-Lumileds in the United States is -2—4mV/oC. As for the array LED (BXRA) of Bridgelux in the United States, more detailed data is given.

Figure 2. Lumiled's LuxeonK2 light decay curve

However, the data they provided is so broad that it has no value for use.

In any case, as long as the temperature coefficient of the LED is known, it is easy to infer the junction temperature of the LED from measuring the forward voltage of the LED.

5. How to specifically measure the junction temperature of LEDs.

Now let's take Cree's XLamp7090XR-E as an example to explain how to specifically measure the junction temperature of an LED. It is required that the LED has been installed in a heat sink and a constant current driver is used as the power supply. At the same time, the two wires connected to the LED should be led out. Before power is turned on, connect the voltmeter to the output terminal (the positive and negative poles of the LED), then turn on the power supply. Before the LED heats up, immediately read the voltmeter reading, which is equivalent to the value of V1. Then wait for at least 1 hour until it has reached thermal equilibrium, and then measure it again. The voltage across the LED is equivalent to V2. Subtract these two values ​​to get the difference. Divide it by 4mV to get the junction temperature. In fact, most LEDs are connected in series and then in parallel. This does not matter. The voltage difference at this time is contributed by many LEDs in series, so the voltage difference should be divided by the number of LEDs in series and then divided by 4mV to get its junction temperature. For example, if there are 10 LEDs in series and 2 in parallel, the voltage measured for the first time is 33V, and the voltage measured for the second time after thermal equilibrium is 30V, and the voltage difference is 3V. This number should be divided by the number of LEDs in series (10) to get 0.3V, and then divided by 4mV to get 75 degrees. Assuming that the ambient temperature before powering on is 20 degrees, the junction temperature should be 95 degrees at this time.

The junction temperature obtained by this method is definitely much more accurate than measuring the temperature of the heat sink with a thermocouple and then calculating its junction temperature.

6. How to predict the life of this lamp?

It seems that it should be very simple to infer the life from the junction temperature. Just check the curve in Figure 1, and you can know that the life corresponding to the junction temperature of 95 degrees can be obtained as 20,000 hours. However, this method is still credible for indoor LED lamps. If it is applied to outdoor LED lamps, especially high-power LED street lamps, there are still many uncertainties. The biggest problem is that the heat dissipation efficiency of the radiator of the LED street lamp decreases over time. This is due to the accumulation of dust and bird droppings, which reduces its heat dissipation efficiency. Also, because there is strong ultraviolet light outdoors, it will also reduce the life of the LED. Ultraviolet light mainly plays a big role in the aging of the encapsulated epoxy resin. If silicone is used, it can be improved. Ultraviolet light also has some bad effects on the aging of phosphors, but it is not very serious.

However, this method is effective in comparing the heat dissipation effects of two heat sinks. Obviously, the heat sink with a smaller left shift of the volt-ampere characteristic has a better heat dissipation effect. In addition, it is still accurate to predict the life of indoor LED lamps.