The key to power LED thermal design: how to manage thermal resistance

As we all know, the luminous properties of LEDs are closely related to their operating conditions. The forward current applied to the LED is the main influencing factor. The higher the current, the more luminous flux the LED produces. Unfortunately, LEDs are driven by a constant current source, and when the temperature of the LED rises, its light output will drop sharply. Figure 1 shows the effect of common LED basic parameters on the output spectrum. In addition, this figure also shows that the efficiency and luminous color of the LED will also shift at the peak wavelength.

The Importance of LED Thermal Characteristics

Since the light output of LEDs changes with temperature, good thermal management is an important issue in power LED lighting applications. By reducing the temperature of the LED, we can keep it efficient. In practical applications, the lower the temperature of the LED, the more lumens it outputs.

This means that in actual applications, the actual thermal resistance of the LED from the junction to the ambient is an important factor in the design of LED lighting. Unfortunately, different LED suppliers provide a wide range of thermal resistance and other temperature-related characteristic parameters. Therefore, different thermal standards organizations have also begun to develop standards for LED thermal management. Today, the JEDEC JC15 association is drafting a new standard for measuring the thermal resistance of LEDs. In addition, the International Lighting Committee has established two new technical committees (TC-2-63 and TC-2-64) to deal with LED thermal issues. A consensus has gradually been reached among these associations that suppliers must consider the actual optical power Popt (in other words, the radiant luminous flux) when calculating the thermal resistance of LEDs using Equation 1:

In the formula, the product of the LED forward current and forward voltage (IFxVF) is the electrical power required for the LED to operate, and ∆Tj is the change in junction temperature of the LED.

Ignoring the optical power when determining the thermal resistance of an LED will result in a lower thermal resistance than the actual LED application. If LED lighting designers use this data to calculate the light output of an LED lamp, the result is that their designs often fail to meet the actual light output requirements. The actual thermal resistance will be higher, and the LED junction temperature will be higher accordingly. As a result, the luminous flux emitted by the actual LED lighting device will be lower than expected. Obtaining the actual thermal characteristics of the LED is the key to successful LED design.

Thermal Characterization: Simulation and Physical Testing

Thermal simulation can help designers understand the heat dissipation of their LED products. Because the heat emitted by LED light sources generally enters the surrounding environment through natural convection, CFD analysis tools are necessary to determine the heat dissipation performance of different design solutions.

Figure 2 shows the thermal simulation results of a modified MR16 LED lamp in a JEDEC standard natural convection test environment.

In order to establish an accurate thermal simulation model, the thermal resistance value of the LED in actual application must be determined. The thermal resistance value of the LED in actual application can usually be completed by measuring instruments such as Tr3ster. Tr3ster is a product developed by the Mentor Graphics MicReD team. Figure 3 is the test equipment used for the LED thermal test in Figure 2.

Figure 4 is a curve of the relationship between the junction temperature and Zth of an LED measured by the Tr3ster thermal transient tester. This test result can be used to obtain detailed structural information on the thermal path of the LED, which mainly refers to the heat transfer path between the PN junction of the LED and the environment. This detailed structural information is described in the form of a curve of the relationship between thermal resistance and thermal capacitance. This type of curve is also called a structure function. The structure function can help designers determine the thermal resistance of each part of the entire LED thermal design, including the LED node, TIM, heat sink or lighting equipment.

Figure 5 shows that 50% of the total junction-to-ambient thermal resistance in the entire LED photo device is caused by the LED itself. Structure functions can not only help with structural analysis (for example, die attach failure detection), but also help generate simplified thermal models of the package component dynamics. Such simplified models can be directly used by CFD software. (Some semiconductor suppliers have also begun to provide transient models of their product thermal performance)

Combined thermal and photometric measurements

Figures 4 and 5 provide some very useful comparisons for interpretation, but thermal characterization data is essential for actual design work. Therefore, when calculating the actual thermal resistance value, the optical power of the LED must be clearly understood.

In order to obtain this information, a thermal test device (compliant with the applicable thermal test standards [3]) must be able to test the LED optical power. LED optical power testing must comply with the relevant standards of the CIE Association [4]. Figure 6 is a description of such a test system. The Tr3ster thermal test instrument provides an electrical power to the LED under test in the TERALED system, which is an automatic photometric measurement device consisting of an integrating sphere and a detector. In addition, the entire system also includes electronic control and test data processing software. The LED booster (the small box on the left of Figure 6) allows the system to test multi-chip high forward voltage (VF>10V) LEDs.

By measuring LEDs with Tr3ster, we can obtain data such as radiant flux, luminous flux, light output characteristics, and colorability while obtaining LED thermal resistance. We can measure these LED characteristic values ​​under different reference temperatures and forward current conditions. Adding thermal transient testing to the photometric measurement process does not significantly increase the test time. Today, power LEDs can usually reach a stable temperature within 30~60S after being attached to a cold plate. Therefore, including thermal transient testing in the photometric measurement process does not increase the test time much.

Effect of reference temperature

The tricky part is that the total thermal resistance of an LED is highly dependent on the ambient temperature. This means that when predicting the thermal performance of an LED, the test environment (reference temperature) must be specified. If the photometric measurement and thermal resistance measurement are performed at the same time, the reference temperature is the temperature of the cold plate.

The data in the LED manual is based on an ambient temperature of 25°C, but the actual operating ambient temperature of the LED is often 50°C, and can even reach 80°C. Its junction temperature may range from 80°C to 110°C. A higher LED operating temperature will cause a significant drop in LED luminous flux.

Figure 7 shows the relationship between the luminous flux and reference temperature of the Cree MCE series white LED. These tests are mainly based on two different heat dissipation design schemes. The test mainly consists of two different PCBs, a metal chip and an FR4 device. In addition, different thermal interface materials are used between the PCB and the heat sink. As the heat sink temperature continues to increase, the luminous flux continues to decrease.

Because the same LED was used in both tests, the expected test results were two parallel curves, but this was not the case. The total junction-to-ambient thermal resistance also changes with the reference temperature. The structure function in Figure 8 also shows that the heat flow path changes with temperature. The initial 1.5K/W thermal resistance is caused by the internal package of the LED. The subsequent thermal resistance corresponds to the PCB and the TIM material between the PCB and the LED package. The last part is the thermal interface material between the PCB and the heat sink. In the test of the TG2500 sample, both layers of TIM material showed that their thermal resistance was highly dependent on temperature, resulting in a 20% change in the total thermal resistance. The structure function in Figure 8 is mapped to the various components of the LED.

Luminous flux as a function of true junction temperature

Once the heat loss and reference temperature value of the LED at each reference temperature are obtained, the actual LED junction temperature can be calculated using Formula 2:

Among them, Pheat = IFxVF (Tref) - Popt (Tref) is also used in Formula 1, and RthJA is the measured thermal resistance. If the test data in Figure 7 is reprocessed and the luminous flux is used as a function of the LED junction temperature, we find that the results of the two tests are almost the same (as shown in Figure 9). The two almost overlapping luminous flux and junction temperature curves indicate that the LED chip and its package have the same light output characteristics in our two tests.

Among them, Pheat = IFxVF (Tref) - Popt (Tref) is also used in Formula 1, and RthJA is the measured thermal resistance. If the test data in Figure 7 is reprocessed and the luminous flux is used as a function of the LED junction temperature, we find that the results of the two tests are almost the same (as shown in Figure 9). The two almost overlapping luminous flux and junction temperature curves indicate that the LED chip and its package have the same light output characteristics in our two tests.

in conclusion

Temperature is an important factor affecting the performance of LED lighting equipment, not only affecting its expected service life but also determining its performance. Lower operating temperatures can achieve more luminous flux. Since most LED suppliers do not conduct LED thermal resistance tests and photometric tests at the same time, the actual thermal resistance value of LEDs cannot be provided in today's LED specifications. Therefore, the thermal resistance value provided by LED suppliers is lower than the thermal resistance value of LEDs in actual applications. If you want to obtain the thermal performance of LEDs through CFD simulation, then knowing the actual thermal resistance value of LEDs is necessary. If this information is not available, then combining photometric measurements and thermal transient tests, and performing some test data processing, you can obtain relevant information about thermal resistance.

Few LED specifications specify the light output characteristics at various temperatures. By determining the thermal resistance and heat loss of the LED under test, the light output characteristics can be described as a function of the actual junction temperature. This can eliminate the effect of different ambient temperatures on the actual thermal resistance during the test. When the light output characteristics can correspond to the actual junction temperature, it becomes possible to accurately compare different LED lighting devices.