Multi-chip parallel power module thermal network model correction method based on thermal diffusion angle
Through the thermal network model correction method of multi-chip parallel power module based on thermal diffusion angle, the over-temperature chip is screened and the effective convection heat exchange area is replaced equivalently, which solves the problem of heat diffusion calculation error and achieves a more accurate junction temperature distribution Evaluation and thermal design optimization.
Patent Information
- Application Number
- CN202411012796.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-26
- Publication Date
- 2024-11-19
AI Technical Summary
When calculating the junction temperature of the existing multi-chip parallel power module, the convection heat exchange area is incomplete when heat diffused to the heat dissipation substrate, resulting in large calculation errors, and it is impossible to accurately evaluate the junction temperature distribution of the chip inside the power module.
The thermal network model correction method of multi-chip parallel power module based on thermal diffusion angle is adopted. By screening the over-temperature chip, the effective convective heat exchange area is performed, and the convective heat exchange area of the non-over-temperature chip is corrected to reduce iterative calculations. Steps: Calculate the junction temperature deviation and obtain the correction result of the minimum junction temperature difference value through the exhaustive method.
It improves the accuracy of junction temperature distribution, reduces calculation complexity and time-consuming, and can more accurately evaluate the junction temperature distribution of the chip inside the power module, guiding thermal design optimization.
Smart Images

Figure CN118981999A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of microelectronics, and in particular relates to a thermal network model correction method for a multi-chip parallel power module based on a heat diffusion angle. Background Art
[0002] As a wide bandgap material, silicon carbide has the advantages of high breakdown field strength, high saturated electron drift rate, and high thermal conductivity. In addition to having ideal gate insulation characteristics, high-speed switching performance, low on-resistance, and high stability, the most significant advantage of SiC MOSFET is that it has a simple drive circuit and is well compatible with the drive circuits of existing silicon-based power devices. However, due to the small chip area and high current density of SiC MOSFET, and with the continuous improvement of its power level, power density, and switching frequency, compared with silicon-based devices, SiC MOSFET needs to withstand higher operating temperatures. More than half of the failures are caused by excessive device temperature, and the failure rate of the device will nearly double for every 10°C increase in temperature. Therefore, it is very necessary to accurately evaluate the temperature distribution of each chip in the power module and optimize the thermal design of the power module during the module design stage to improve the thermal reliability of the SiC MOSFET module.
[0003] The existing SiC MOSFET modules can generally be divided into a multi-layer structure. The presence of the insulation gel makes the heat in the module mainly transferred from top to bottom, until the bottom heat dissipation substrate. Due to heat diffusion, the effective heat conduction area continues to increase. At present, most of the methods for calculating the junction temperature are calculated using the heat diffusion model. Specifically, the heat flux density curve of the vertical downward path from the center of the chip is used to obtain the heat diffusion angle and calculate each parameter. The junction temperature error of this calculation method is small. Since there is generally more than one chip inside the module, that is, there are multiple heat sources, and they are all soldered on the same substrate, there is thermal coupling. Therefore, the method for calculating the junction temperature in this case is to calculate the self-heating thermal resistance and coupling thermal resistance based on the improved single-chip Cauer network equivalent circuit model and establish the thermal impedance matrix of the multi-chip module, and use the linear superposition principle to predict the junction temperature of each chip. Although finite element simulation can also obtain the junction temperature, the thermal network model can more deeply analyze the heat conduction and heat exchange process inside the power module, and can guide the thermal design and optimization of the power module by calculating parameters such as thermal resistance. In addition, the thermal network model can also consider more factors, such as the thermal conductivity of the material, the convection heat transfer coefficient, etc., so as to more accurately evaluate the junction temperature distribution of the chip, and provide a more reliable basis for its design and optimization, and then provide guidance to regulate the heat distribution characteristics of the power module and improve the reliability of the power module.
[0004] However, in the current mainstream calculation methods, whether it is a model that only considers heat diffusion or thermal coupling, when the heat of the chip located at the edge of the power device (DBC) diffuses to the heat dissipation substrate, the radius of the convective heat transfer equivalent circle may exceed the boundary of the heat dissipation substrate, so that the convective heat transfer area during calculation is no longer the area of a complete circle, which in turn leads to a large calculation error of the thermal network model, and it is impossible to accurately evaluate the junction temperature distribution of the parallel chips inside the power module, and it is also impossible to guide the thermal design of the power module.
[0005] In summary, it is necessary to propose a correction method for the thermal network model of a multi-chip parallel power module based on the heat diffusion angle, so as to calibrate the calculated junction temperature and obtain a more accurate junction temperature distribution result. Summary of the invention
[0006] The present invention provides a method for correcting a thermal network model of a multi-chip parallel power module based on a heat diffusion angle to solve the above problem.
[0007] The present invention adopts the following technical solutions:
[0008] A method for correcting a thermal network model of a multi-chip parallel power module based on a heat diffusion angle, the steps comprising:
[0009] Establishing a thermal network model of a multi-chip parallel power module based on a heat diffusion angle, and solving the calculated junction temperature of each chip in the multi-chip parallel power module based on the thermal network model;
[0010] Calculating the junction temperature deviation between the calculated junction temperature of each chip and the corresponding simulated junction temperature by a difference comparison method;
[0011] Screening an over-temperature chip in the multi-chip parallel power module, wherein the junction temperature deviation of the over-temperature chip exceeds a threshold value;
[0012] An effective convection heat transfer area of the over-temperature chip is equivalently replaced; the object of the equivalent replacement is the effective convection heat transfer area of the non-over-temperature chip in the multi-chip parallel power module;
[0013] The secondary calculated junction temperature of the over-temperature chip after equivalent replacement is calculated by an exhaustive method, and the secondary calculated junction temperature corresponding to the minimum junction temperature difference is used as the correction result.
[0014] Optionally, establishing a thermal network model of a multi-chip parallel power module based on a heat diffusion angle includes:
[0015] Establishing a three-dimensional model of the multi-chip parallel power module, and dividing the multi-chip parallel power module based on chip specifications and / or chip types to form at least one sub-unit;
[0016] The subunits are simplified into a multi-layer structure based on the three-dimensional model.
[0017] Optionally, the process of calculating the junction temperature includes:
[0018] Obtaining heat flux density data of each chip in the subunit by finite element method, and drawing a heat flux density curve based on the heat flux density data;
[0019] Calculate the heat diffusion angle based on the heat flux density curve, and calculate the thermal resistance of the chip through the heat diffusion angle;
[0020] The thermal resistance of each chip in the subunit is solved based on the heat diffusion angle of each layer under the chip.
[0021] Optionally, the function formula of the heat flux curve is:
[0022]
[0023] Where r is the effective heat conduction circle radius, P is the total chip power, and q is the heat flux density.
[0024] Optionally, the thermal resistance R th The formula is:
[0025]
[0026] Where r1 is the radius of the effective heat conduction circle at the junction of the current layer and the upper layer, d is the thickness of each layer, α is the heat diffusion angle of each layer, λ is the thermal conductivity of each layer material, and x is the horizontal coordinate of each layer in the heat flux density curve.
[0027] Optionally, the junction temperature calculation solution process further includes:
[0028] Solving the effective convection heat transfer area based on the heat diffusion angle, and calculating the convection heat transfer thermal resistance through the effective convection heat transfer area;
[0029] The effective convection heat exchange area is the effective heat conduction area corresponding to the heat dissipation substrate at the bottom layer of the subunit.
[0030] Optionally, the convection heat transfer resistance R nv The formula is:
[0031]
[0032] Where h is the convective heat transfer coefficient, A s is the effective convective heat transfer area.
[0033] Optionally, the calculated junction temperature is solved by total thermal resistance, and the total thermal resistance R0 is calculated as follows:
[0034] R0=R th +R nv ;
[0035] The calculated junction temperature T c The solution formula is:
[0036] T c =R0·P+T a ;
[0037] Where, T a is the ambient temperature.
[0038] Optionally, the junction temperature deviation e is calculated as:
[0039]
[0040] Where, T A is the simulated junction temperature obtained by finite element simulation calculation.
[0041] Optionally, also include:
[0042] Screening the objects of the equivalent replacement, extracting effective replacement chips whose convective heat transfer equivalent graphs are circular in the non-overtemperature chips, and sorting the effective replacement chips according to their distances from the overtemperature chips, and calculating the junction temperature deviations in sequence;
[0043] The method for extracting the effective replacement chip is: selecting a chip whose distance from any point on the edge of the heat dissipation substrate to the center of the chip is greater than or equal to the radius of the equivalent circle of convection heat transfer.
[0044] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:
[0045] The present invention screens over-temperature chips and performs equivalent replacement of their effective convection heat transfer areas. There is no need for iterative calculation, which reduces the steps of iterative calculation. The corresponding correction value is the effective convection heat transfer area previously calculated for the adjacent chip. The calculation is relatively simple and time-saving, and the problem of large junction temperature error caused by the effective convection heat transfer area being limited by the size of the heat dissipation substrate is solved. The junction temperature distribution can be obtained more accurately, which is convenient for evaluating the junction temperature distribution of parallel chips inside the power module and guiding the thermal design of the power module. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0047] Figure 1 is a flow chart of the method of the present invention;
[0048] Figure 2 is a top view of a multi-chip parallel power module in a specific embodiment of the present invention;
[0049] Figure 3 It is a partially enlarged stereoscopic diagram and a simplified structural diagram of a multi-chip parallel power module in a specific embodiment of the present invention;
[0050] Figure 4 is a flow chart of the present invention for calculating junction temperature;
[0051] Figure 5 It is a comparison diagram of convective heat transfer in a specific embodiment of the present invention.
[0052] In the figure: 1. first subunit, 2. second subunit, 3. third subunit, 4. fourth subunit, 5. first chip, 6. second chip, 7. third chip, 8. chip body, 9. chip solder layer, 10. DBC upper copper layer, 11. DBC ceramic layer, 12. DBC lower copper layer, 13. DBC solder layer, 14. heat dissipation substrate. DETAILED DESCRIPTION
[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0054] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.
[0055] It should also be noted that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products unless otherwise specified.
[0056] The present invention provides a method for correcting a thermal network model of a multi-chip parallel power module based on a heat diffusion angle. Figure 1 As shown, the steps include:
[0057] A thermal network model of a multi-chip parallel power module based on the heat diffusion angle is established, and the calculated junction temperature of each chip in the multi-chip parallel power module is solved based on the thermal network model.
[0058] Specifically, Figure 2 and Figure 3As shown, a three-dimensional model of a multi-chip parallel power module is established, and the multi-chip parallel power module is divided based on chip specifications and chip types. In this embodiment, it is divided into four sub-units according to functional blocks, namely, a first sub-unit 1, a second sub-unit 2, a third sub-unit 3, and a fourth sub-unit 4. In order to simplify the description, only the first sub-unit 1 is used as an example. There are a total of three chips in the first sub-unit 1, namely, a first chip 5, a second chip 6, and a third chip 7.
[0059] Further, based on this three-dimensional model, the first subunit 1 is simplified into a multi-layer structure, specifically, as Figure 2 As shown, from top to bottom are chip body 8, chip solder layer 9, DBC upper copper layer 10, DBC ceramic layer 11, DBC lower copper layer 12, DBC solder layer 13 and heat dissipation substrate 14. Combining the basic principle, it can be known that heat is transmitted downward from chip body 8 to heat dissipation substrate 14.
[0060] Before calculating the junction temperature, it is also necessary to obtain the heat flux density data and simulated junction temperature (T A ), and draw the heat flux curve based on the heat flux data. Figure 4 As shown, the junction temperature of each chip is calculated, where the function formula of the heat flux density curve is:
[0061]
[0062] Where r is the effective heat conduction circle radius, P is the total chip power, and q is the heat flux density.
[0063] The above formula can be used to obtain the radius variation curve of the corresponding effective thermal conductivity figure (conventionally calculated as a circle) for each chip. The abscissa of the curve represents the distance from the point on the path to the center of the top of the chip, and the ordinate represents the radius of the effective thermal conductivity figure. The segmented slope of the curve is the tangent value of the heat diffusion angle of each layer under the chip. The ordinate value corresponding to the maximum value of the abscissa is the radius value of the effective convection heat transfer figure on the bottom surface of the heat dissipation substrate corresponding to the corresponding chip.
[0064] Based on the heat flux density curve, the heat diffusion angle of each chip and each layer is obtained, and the thermal resistance of the chip is solved through the heat diffusion angle. th The formula is:
[0065]
[0066] Where r1 is the radius of the effective heat conduction circle (effective heat conduction pattern) at the junction of the current layer and the upper layer, d is the thickness of each layer, α is the heat diffusion angle of each layer, λ is the thermal conductivity of each layer material, and x is the horizontal coordinate of each layer in the heat flux density curve.
[0067] Furthermore, the effective convection heat transfer area is solved based on the heat diffusion angle, and the convection heat transfer thermal resistance is calculated through the effective convection heat transfer area. The effective convection heat transfer area of the chip is the effective thermal conductivity area corresponding to the heat dissipation substrate at the bottom. Therefore, the convection heat transfer thermal resistance R nv The formula is:
[0068]
[0069] Where h is the convective heat transfer coefficient, A s is the effective convective heat transfer area.
[0070] The calculated junction temperature corresponding to the chip is solved by the total thermal resistance. The total thermal resistance R0 is the sum of the thermal resistance of thermal conduction and the thermal resistance of convection heat transfer. The formula is:
[0071] R0=R th +R nv ;
[0072] Based on this, calculate the junction temperature T c The solution formula is:
[0073] T c =R0·P+T a ;
[0074] Where, T a is the ambient temperature.
[0075] The junction temperature deviation between the calculated junction temperature of each chip and the corresponding simulated junction temperature is calculated by the difference comparison method. The formula for the junction temperature deviation e is:
[0076]
[0077] Where, T A is the simulated junction temperature obtained by finite element simulation calculation. Accordingly, in this embodiment, the junction temperature deviations e5, e6, and e7 of the first chip 5, the second chip 6, and the third chip 7 can be obtained in sequence.
[0078] Thus, the following Table 1 is obtained in a specific embodiment:
[0079]
[0080] Table 1
[0081] Depend on Figure 5It can be seen that the effective convection heat transfer pattern on the bottom surface of the heat dissipation substrate corresponding to the first chip 5 in Figure (a) is not a circle, the effective convection heat transfer pattern on the bottom surface of the heat dissipation substrate corresponding to the second chip 6 in Figure (b) is not a circle, and the effective convection heat transfer pattern on the bottom surface of the heat dissipation substrate corresponding to the third chip 7 in Figure (c) is approximately a circle; and the conventional thermal network model calculation method based on the heat diffusion angle is not applicable in the two cases of Figure (a) and Figure (b). The reason is that when the effective convection heat transfer pattern is no longer a circle due to the lateral size restriction of the heat dissipation substrate, the effective convection heat transfer area of the bottom surface of the heat dissipation substrate of the chip calculated by using the heat flux density change curve of the vertical downward path of the chip center is too small, which makes the deviation of the calculated junction temperature large, exceeding the set threshold value of 3.1%.
[0082] Therefore, in the first subunit 1 of the multi-chip parallel power module of this embodiment, over-temperature chips are screened, namely the first chip 5 and the second chip 6 .
[0083] In order to correct the calculated junction temperature, it needs to be adjusted. It can be deduced that the reason for the deviation lies in the size of the effective convection heat transfer area, so the key to the correction is to increase the area to a suitable value. Therefore, it is necessary to perform an equivalent replacement on the effective convection heat transfer area of the over-temperature chip. Among them, the object of equivalent replacement is the effective convection heat transfer area of the non-over-temperature chip in the multi-chip parallel power module. In this embodiment, since there are two over-temperature chips and the non-over-temperature chip is only the third chip 7, it is sufficient to replace the effective convection heat transfer area of the third chip 7, but in other scenarios there may be multiple chips available for replacement, so a corresponding screening strategy is needed, as follows:
[0084] Screen the objects of equivalent replacement, and extract effective replacement chips whose convective heat transfer equivalent graphs are circles among multiple non-overtemperature chips, such as chip n-chip m. It is preferred to use the in-model calculation method to extract suitable effective replacement chips, that is: select chips whose distance from any point on the edge of the heat dissipation substrate to the center of the chip is greater than or equal to the radius of the convective heat transfer equivalent circle, thereby ensuring that the obtained replacement chip satisfies the effective convective heat transfer graph on its heat dissipation substrate as a complete circle. Afterwards, sort the effective replacement chips according to their distance from the overtemperature chip, chip n, chip n+1...chip m-1, chip m, chip n is closest to the overtemperature chip, and starting from chip n, calculate the junction temperature deviation in sequence. The exhaustive method is used to calculate the secondary calculated junction temperature of the over-temperature chip after equivalent replacement, and the secondary calculated junction temperature corresponding to the minimum junction temperature difference is selected as the correction result; similarly, the termination means can also be based on meeting the threshold. If the obtained junction temperature deviation e is large and does not meet the threshold, the effective convection heat transfer area value of the chip adjacent to the required chip is continued to be replaced; if the obtained junction temperature deviation e is small and meets the threshold requirement, no replacement is performed and the correction is completed.
[0085] In this embodiment, since the three chips are in the same subunit and their work or functions are closely related and affect each other, the differences between various parameters will not be too large. Consider using the effective convection heat transfer area value of the third chip 7 applicable to the previous model to replace the effective convection heat transfer area of the bottom surface of the heat dissipation substrate of the first chip 5 and the second chip 6 for correction. That is, the quick correction method is: the effective convection heat transfer areas of the first and second chips 5 and 6 are replaced by the corresponding values of the third chip 7, while the calculated value of the third chip 7 itself remains unchanged, that is, in the formula of convection heat transfer thermal resistance, the A used to solve the first and second chips s5 , A s6 Replaced with the third chip A s7 After calculation, the following table 2 is obtained.
[0086]
[0087] Table 2
[0088] It can be seen from the table that after correction, the deviation values are all within the threshold range, which can well reflect the junction temperature distribution, and give full play to the advantages of the traditional model, deeply analyze the heat conduction and heat exchange process inside the power module, and can guide the thermal design and optimization of the power module by calculating parameters such as thermal resistance.
[0089] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for correcting a thermal network model of a multi-chip parallel power module based on a heat diffusion angle, characterized in that the steps include: Establishing a thermal network model of a multi-chip parallel power module based on a heat diffusion angle, and solving the calculated junction temperature of each chip in the multi-chip parallel power module based on the thermal network model; Calculating the junction temperature deviation between the calculated junction temperature of each chip and the corresponding simulated junction temperature by a difference comparison method; Screening an over-temperature chip in the multi-chip parallel power module, wherein the junction temperature deviation of the over-temperature chip exceeds a threshold value; An effective convection heat transfer area of the over-temperature chip is equivalently replaced; the object of the equivalent replacement is the effective convection heat transfer area of the non-over-temperature chip in the multi-chip parallel power module; The secondary calculated junction temperature of the over-temperature chip after equivalent replacement is calculated by an exhaustive method, and the secondary calculated junction temperature corresponding to the minimum junction temperature difference is used as the correction result.
2. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 1, characterized in that: The thermal network model of multi-chip parallel power module based on heat diffusion angle includes: Establishing a three-dimensional model of the multi-chip parallel power module, and dividing the multi-chip parallel power module based on chip specifications and / or chip types to form at least one sub-unit; The subunits are simplified into a multi-layer structure based on the three-dimensional model.
3. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 2, characterized in that: The process of calculating the junction temperature includes: Obtaining heat flux density data of each chip in the subunit by finite element method, and drawing a heat flux density curve based on the heat flux density data; Calculate the heat diffusion angle based on the heat flux density curve, and calculate the thermal resistance of the chip through the heat diffusion angle; The thermal resistance of each chip in the subunit is solved based on the heat diffusion angle of each layer under the chip.
4. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 3, characterized in that: The function formula of the heat flux curve is: Where r is the effective heat conduction circle radius, P is the total chip power, and q is the heat flux density.
5. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 4, characterized in that: The thermal resistance R th The formula is: Where r1 is the radius of the effective heat conduction circle at the junction of the current layer and the upper layer, d is the thickness of each layer, α is the heat diffusion angle of each layer, λ is the thermal conductivity of each layer material, and x is the horizontal coordinate of each layer in the heat flux density curve.
6. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 5, characterized in that: The junction temperature calculation solution process also includes: Solving the effective convection heat transfer area based on the heat diffusion angle, and calculating the convection heat transfer thermal resistance through the effective convection heat transfer area; The effective convection heat exchange area is the effective heat conduction area corresponding to the heat dissipation substrate at the bottom layer of the subunit.
7. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 6, characterized in that: The convection heat transfer resistance R nv The formula is: Where h is the convective heat transfer coefficient, A s is the effective convective heat transfer area.
8. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 7, characterized in that: The calculated junction temperature is solved by the total thermal resistance, and the formula of the total thermal resistance R0 is: R0=R th +R nv ; The calculated junction temperature T c The solution formula is: T c =R0·P+T a ; Where, T a is the ambient temperature.
9. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 8, characterized in that: The formula for the junction temperature deviation e is: Where, T A is the simulated junction temperature obtained by finite element simulation calculation.
10. The method for correcting the thermal network model of a multi-chip parallel power module based on the heat diffusion angle according to claim 1, characterized in that: Also includes: Screening the objects of the equivalent replacement, extracting effective replacement chips whose convective heat transfer equivalent graphs are circular in the non-overtemperature chips, and sorting the effective replacement chips according to their distances from the overtemperature chips, and calculating the junction temperature deviations in sequence; The method for extracting the effective replacement chip is: selecting a chip whose distance from any point on the edge of the heat dissipation substrate to the center of the chip is greater than or equal to the radius of the equivalent circle of convection heat transfer.
