Abstract: With the continuous development of semiconductor lighting technology, LEDs have been applied to more and more lighting fields, leading to the emergence of high-power LEDs. However, as LED power gradually increases, heat dissipation becomes extremely prominent. Thermal resistance, as a key parameter for evaluating heat dissipation performance, has significant theoretical and practical implications for analysis. Excessive thermal resistance leads to excessively high temperatures, directly affecting the wavelength, luminous efficiency, forward voltage drop, and lifespan of high-power LEDs. Therefore, this paper proposes a real-time thermal resistance analysis method for high-power LEDs based on multi-point temperature measurement. A data acquisition scheme was designed, and the uncertainty of the high-power LED thermal resistance analysis was calculated, verifying the authenticity and reliability of the entire analysis method.
1 Introduction
With the continuous increase in LED power, heat dissipation has become a major factor restricting the widespread adoption of high-power LEDs. As LED power increases, its packaging structure becomes more complex, while its size decreases, resulting in higher power density. Higher power density leads to more severe heat dissipation problems. Poor heat dissipation causes the PN junction temperature of high-power LEDs to rise, leading to light decay, shorter lifespan, and in severe cases, chip burnout. Research on the heat dissipation problem of high-power LEDs will play a crucial role in their future promotion. The analysis of thermal resistance has significant theoretical and practical implications.
Theoretical and Model Establishment of Thermal Resistance of 2 High-Power LEDs
2.1 Thermal Resistance Theory
For high-power LEDs, heat is mainly generated through two methods: thermal conduction and thermal convection.
The principle of heat conduction follows Fourier's law, which can be expressed as follows: during heat conduction per unit time, the amount of heat passing through a certain cross section is directly proportional to the cross-sectional area perpendicular to this section and the rate of temperature change, as shown in formula (1):
(1)
For thermal convection, the phenomenon is less pronounced in liquids than in gases. This can be expressed using Newton's law of cooling:
(2)
Or
(3)
As can be seen from equations (2) and (3), the heat transferred by convective heat transfer is proportional to the temperature difference between the fluid and the solid surface and the heat transfer area between them. The proportionality coefficient is the surface heat transfer coefficient. The process of convective heat transfer is relatively complex. The geometry of the heat transfer surface, the changes in the physical properties and physical state of the fluid, and the boundary conditions of the heat transfer surface will all have a certain impact on convective heat transfer.
2.2 Thermal Resistance Model
The concept of thermal resistance is similar to that of electrical resistance. The temperature difference between two points on a heat transfer path is equivalent to voltage, the heat generated during heat transfer is equivalent to current, and thermal resistance is equivalent to electrical resistance. Thermal resistance is defined as the ratio of the temperature difference between two specified points or regions in thermal equilibrium to the power dissipated between those two points, with units of °C/W, and expressed by the formula:
(4)
Where Rth is the thermal resistance, t1 and t2 are the temperatures at the two points, and P is the power consumed.
Analysis of the heat transfer mechanism of high-power LEDs shows that for a single-layer flat wall heat conduction process, if the temperatures of the two surfaces are t1 and t2 respectively, then according to Fourier's law, i.e., equation (1):
(5)
Right now,
(6)
For high-power LEDs, the thermal resistance encountered during the heat transfer between solids via thermal conduction is called thermal resistance.
In the process of convective heat transfer, according to Newton's law of cooling, namely equations (2) and (3), we can derive:
(7)
Right now,
(8)
The heat dissipation structure of a high-power LED typically consists of the following parts: a high-power LED chip module, a metal circuit board, and an external heat sink. The chip is soldered to the metal circuit board via pads, and the printed circuit board is connected to the external heat sink to dissipate heat to the outside. A typical high-power LED heat dissipation structure diagram is shown in Figure 1.
Figure 1. Heat dissipation structure diagram of high-power LED
According to the heat dissipation structure of high-power LEDs, the heat dissipation process of high-power LEDs is as follows: the high-power LED chip emits light and generates heat. The heat is transferred from the inside of the chip to the metal circuit board through thermal conduction, and then transferred to the external heat sink through the metal circuit board. Finally, the heat is transferred between the external heat sink and the outside air through thermal convection.
Figure 2 Thermal resistance model of high-power LED
Based on the overall structure and heat dissipation process of high-power LEDs, a thermal resistance model for high-power LEDs can be summarized, as shown in Figure 2. The total thermal resistance Rth of a high-power LED is the sum of the thermal resistances along the heat dissipation path from the LED chip to the external environment. This mainly includes the internal thermal resistance RS-B of the LED chip (the thermal resistance between the LED chip and the metal circuit board during heat conduction); the thermal resistance RB-S between the metal circuit board and the external heat sink; the thermal resistance RS-A between the external heat sink and the outside air during heat convection; and the thermal resistance during heat convection between the chip and the air. However, since the heat dissipation effect between the chip and the external environment is not significant, it can be ignored. Therefore, the total thermal resistance of a high-power LED is the sum of its components.
3LED Thermal Resistance Analysis
The thermal resistance analysis of high-power LEDs is mainly described from the following four aspects: the design of the data acquisition scheme for high-power LED thermal resistance analysis, the real-time thermal resistance analysis of high-power LEDs, and the further analysis of thermal resistance using the Hilbert-Huang transform.
3.1 Data Acquisition Design
The hardware and software design of the data acquisition system based on a microcontroller are shown in Figure 3.
Figure 3 System Overall Structure Diagram
This system uses a PIC microcontroller as its core and a digital temperature sensor to collect multi-point temperature data from high-power LEDs. The digital temperature signal is directly sent to the PIC microcontroller for processing, enabling multi-point temperature data acquisition from the high-power LEDs. Simultaneously, the voltage and current signals across the high-power LEDs are acquired and converted from digital signals by the PIC microcontroller using an A/D converter. The processed voltage value is then sent to a computer. Finally, the processed multi-point temperature data, voltage, and current data are transmitted to a host computer via an RS-232 interface, providing data support for real-time thermal resistance analysis of the high-power LEDs.
3.2 Hardware Design of Data Acquisition System
The core control chip in this paper uses the PIC16F877 microcontroller manufactured by Microchip. The multi-point temperature data detection in this system utilizes the new generation single-bus digital temperature sensor DS18B20 from Dallas Semiconductor. A single-bus configuration is adopted, meaning multiple DS18B20s are connected to the same pin and operated according to the ROM code. Its hardware schematic is shown in Figure 4, where GND is the ground line; DQ is the data line, which can both send and receive data; this pin is an open-drain output and is normally high; VDD is an optional external power supply terminal, which can be connected to +5V and should be grounded when using parasitic power supply. This system uses VDD power supply and connects DQ to the microcontroller to achieve temperature detection.
Figure 4 Pin configuration diagram of DS18B20
This paper connects the input terminals of the operational amplifier AD620AN (its +IN and -IN pins) to the two ends of R2. The voltage at the output terminal OUT is the voltage across R2. The obtained voltage signal is sent to the A/D conversion port AN0 of the microcontroller for processing, and then the voltage across the LED is calculated based on the resistance ratio of R1 and R2. The specific circuit is shown in Figure 5.
Figure 5 Schematic diagram of voltage acquisition principle
A 1K sampling resistor is connected to measure the voltage signal of the sampling resistor. The signal is then sent to the A/D conversion port AN1 of the microcontroller for processing to obtain the current value passing through the LED.
Finally, the temperature data from the DS18B20 and the voltage and current data across the high-power LED are transmitted to the PIC16F877 microcontroller, which then transmits them to the computer via serial port for processing. This system uses RS-232 to connect the microcontroller and the host computer for serial communication.
3.3 Software Design of the Data Acquisition System
The PIC16F877 microcontroller serves as the data acquisition terminal for the temperature data acquisition system. It primarily controls three DS18B20 digital temperature sensors, enabling the reading of temperature data, acquisition of real-time data from the DS1302 sensor, and communication with the host computer. Based on the system's functional requirements, the multi-point temperature data acquisition system program consists of four modules: the main program, the temperature data acquisition module, the voltage data acquisition module, and the serial communication module. Its flowchart is shown in Figure 6.
Figure 6 Temperature Acquisition Flowchart
Real-time thermal resistance analysis of 4 high-power LEDs
This paper realizes real-time data transmission between a microcontroller and a computer via an RS-232 serial port. Furthermore, based on the analysis method of thermal resistance of high-power LEDs, it implements real-time analysis of thermal resistance using VC++ programming, including all transient and steady-state processes.
After data acquisition, a visualization system for high-power LED thermal resistance analysis needs to be designed. This system transmits temperature and voltage data to a computer via an RS-232 serial port for display and processing. It mainly includes three modules: a serial communication module, a data display module, and a data processing module, which will be described in detail below. This system uses the MSComm control to design the serial communication module.
After transmitting three-point temperature, voltage, and current data via serial port, the next step is to process the data, including real-time display of thermal resistance analysis and display of temperature and power parameters.
To enable real-time analysis of thermal resistance, this paper presents the thermal resistance using a real-time dynamic curve. The thermal resistance analysis results are first stored in a global data structure. Events are triggered by the transmitted data, and functions are implemented within the function to read the data and plot the curve based on it, displaying the temperature and power values in real time, as shown in Figure 7.
Figure 7 Software Interface
4.1 Example of Thermal Resistance Analysis
Taking a high-power LED as an example, thermal resistance analysis is applied to practical engineering, analyzing the thermal resistance under different heat sink conditions. Since the selection of a suitable heat sink material has a significant impact on the thermal resistance of a high-power LED, the heat dissipation performance of the high-power LED can be improved through thermal resistance analysis.
This paper uses different aluminum substrates as heat sinks and analyzes the changes in thermal resistance under different high-power LED aluminum substrates through multi-point temperature measurement, and finally obtains an optimal heat sink design scheme. Table 1 shows the thermal resistance obtained under different heat sinks.
Among them, the substrate thermal resistance is the thermal resistance from the PN junction of the high-power LED to the substrate; the air thermal resistance is the thermal resistance from the PN junction of the LED device to the air. These resistances vary depending on the temperature at different points. Table 1 shows that the high-power LED substrate with AlN as the heat sink has the lowest thermal resistance and the best heat dissipation performance; while the high-power LED substrate with Al as the heat sink has the highest thermal resistance and the worst heat dissipation performance. Therefore, AlN can be chosen as the heat sink material to reduce the thermal resistance of the LED device. As shown in Table 4-1, different aluminum substrates not only cause differences in substrate thermal resistance but also affect the thermal resistance of the environment. This indicates that choosing a suitable heat sink has a significant impact on the thermal resistance of high-power LEDs, which in turn determines the overall heat dissipation performance of the high-power LED.
Table 1. Thermal resistance analysis results under different heat sinks
4.2 Thermal resistance analysis based on Hilbert-Huang transformation
Since the thermal resistance analysis results obtained from the experiment may be mixed with noise signals, this paper uses Hilbert-Huang transform based on empirical mode decomposition (EMD) to further process the thermal resistance analysis results based on multi-point temperature measurements. The Hilbert-Huang transform for signal processing mainly involves two steps: first, applying EMD to obtain a finite number of IMFs, and then applying Hilbert transform and instantaneous frequency method to obtain the Hilbert spectrum of the signal.
For thermal resistance analysis, the EMD process is as follows:
First, based on the thermal resistance analysis result X(t), determine all its local extreme points. Let m1 be the average value of the upper and lower envelopes, and h1 be the difference from the average value m1.
(9)
When h1(t) satisfies the definition of an intrinsic mode function, then h1(t) is used as the first IMF component of X(t); when h1(t) does not satisfy the definition of an IMF, the above steps are repeated with h1(t) as data. Then:
(10)
Wherein: the average value of the upper and lower envelopes of the data h1(t) is m11(t).
By separating c1 from the original data X(t), we can obtain:
(11)
Repeat the above steps, using r1(t) as the new raw data, until a second component c2 that satisfies the IMF appears. Continue this cycle until the original signal no longer contains an IMF component.
(12)
To ensure that each IMF has actual physical meaning, the second criterion for determining an IMF can be transformed into the following standard, as shown in equation (13):
(13)
Where 0, 1, ..., T represent all times included in the average envelope; m1k(t) is the average envelope obtained by the IMF component extraction module in the current iteration; m1(k-1)(t) is the average envelope obtained by the IMF component extraction module in the previous iteration; and a reasonable SD is within the range of 0.2 to 0.3. The entire loop ends when no IMF-satisfied component can be extracted and rn(t) becomes a monotonic function. The original data can be represented by the sum of the IMF components and the final residual, denoted as:
(14)
The entire process of decomposing the thermal resistance analysis results of high-power LEDs using the EMD method is shown in Figure 8.
Figure 8 EMD Flowchart
4.3 Hilbert-Huang Transform Based on EMD
After obtaining a finite number of IMFs using EMD, a Hilbert transform must be performed on each IMF. The transformation formula for real functions is as follows:
(15)
According to the convolution theorem:
(16)
Therefore, the essence of the Hilbert transform is that x(t) passes through a Hilbert filter, and the Hilbert filter is actually a linear network with an impulse response of 1/πt, whose transfer function is shown in the formula:
(17)
φ(t) and A(t) are called the instantaneous phase and instantaneous envelope, respectively. The derivative of φ(t) ω(t) is called the instantaneous frequency, and its expression is as follows:
(18)
(19)
(20)
Therefore, in essence, obtaining the instantaneous parameters of a signal is equivalent to obtaining its conjugate signal. The Hilbert transform extracts the instantaneous features of steady-state signals, especially the instantaneous frequency features, which has significant practical engineering implications. Thus, we can say that non-steady-state data, after EMD decomposition and subsequent transformation, can retain its original physical meaning.
The Hilbert transform of the original signal X(t) is expressed as:
(twenty one)
Define the real part as a Hilbert spectrum, denoted as , then
( twenty two)
The Hilbert spectrum reflects the total amplitude or energy of each IMF component across frequencies.
Figure 9 shows the raw data for thermal resistance analysis of a high-power LED. The Hilbert-Huang transform divides the thermal resistance analysis results into 12 IMF components. After EMD, the first 11 IMFs in the thermal resistance analysis results have low energy and wide frequency spectrum; based on their noise characteristics, they can be removed as noise. Since EMD decomposition is essentially a screening and separation process, decomposing the original signal into the sum of all IMFs and the remaining total trend, the noise components can be directly subtracted from the original signal; this is the denoising process. Figure 10 shows the thermal resistance analysis results after Hilbert-Huang transform.Figure 10. Thermal resistance analysis results based on Hilbert-Huang transformation.
5. Conclusion
Thermal resistance, as a key parameter for evaluating heat dissipation performance, has significant theoretical and practical implications for analysis. Excessive thermal resistance leads to excessively high temperatures, directly affecting the wavelength, luminous efficiency, forward voltage drop, and lifespan of high-power LEDs. This paper proposes a novel analytical method for the thermal resistance of high-power LEDs. Based on research into the fundamental theory of thermal resistance, a real-time thermal resistance analysis method based on multi-point temperature measurement is proposed and successfully applied to practical high-power LEDs. This method enables the analysis of thermal resistance and is used to optimize the heat dissipation performance of aluminum substrates. Furthermore, the Hilbert-Huang transform is applied to the thermal resistance analysis for further analysis. Verification through application in practical high-power LEDs plays a positive role in promoting the further development of high-power LEDs.
