A method for determining thermal decoupling boundary conditions of a compact thermal network model
Patent Information
- Application Number
- CN202210777206.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-30
- Publication Date
- 2026-10-16
- Estimated Expiration
- 2042-06-30
AI Technical Summary
[0050] The method for determining the thermal decoupling boundary conditions of a compact thermal network model provided by this invention allows for the decomposition of the compact thermal network model into independent thermal network models for temperature calculation when the compact thermal network model satisfies the thermal decoupling boundary conditions. This improves the computational efficiency of the compact thermal network model while maintaining a given accuracy. Compared with traditional temperature calculations using compact thermal network models, the beneficial results of this invention are reflected in:
Smart Images
Figure CN115186469B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor manufacturing technology, and more specifically to a method for determining the thermal decoupling boundary conditions of a compact thermal network model. Background Technology
[0002] In power electronic system reliability research, thermal network models are widely used for real-time reliability prediction based on junction temperature and task profiles. One-dimensional Foster and Cauer thermal networks are both one-dimensional models and do not consider thermal coupling effects. Two-dimensional thermal network models consider thermal coupling effects; however, as the number of chips in an IGBT module increases, the complexity of the two-dimensional thermal network model significantly increases, leading to a decrease in computational efficiency. Therefore, it is necessary to simplify the calculation of coupled thermal network models as much as possible within a certain estimation accuracy range.
[0003] To address the computational efficiency issue of coupled thermal network models for IGBT power modules, based on existing publicly available references, some research methods are proposed as follows:
[0004] Reference 1, "Wang Z, Qiao W. An Online Frequency-Domain Junction Temperature Estimation Method for IGBT Modules[J].IEEE Transactions on Power Electronics, 2015, 30(9): 4633-4637," is based on the frequency response of a linear time-invariant system and realizes the online estimation of the junction temperature of IGBT power modules in the frequency domain, which can significantly improve computational efficiency.
[0005] Reference 2, “Ma K, Bahman AS, Beczkowski S, et al. Complete loss and thermal model of power semiconductors including device rating information[J].IEEE Transactions on Power Electronics, 2014.”, proposes a two-stage power loss curve to achieve better thermal estimation while reducing computational burden.
[0006] However, these methods are all proposed for Foster thermal networks, while Cauer thermal networks are more difficult to quantitatively analyze due to their special network structure.
[0007] Reference 3, "Zhang Y, Wang H, Wang Z, et al. A simplification method for power device thermal modeling with quantitative error analysis[J].IEEE Journal of Emerging and Selected Topics in Power Electronics, 2019," proposes a simplified method for power device thermal modeling with quantitative error analysis, which can achieve the minimum thermal estimation workload under the maximum allowable error.
[0008] The above literature optimizes the temperature calculation algorithm to improve computational efficiency, but there is currently little research on simplifying the thermal network model.
[0009] Reference 4, "Zhang Y, Wang H, Wang Z, et al. Simplified multi-time scale thermal model considering thermal coupling in IGBT modules[C] / / 2019 IEEE Applied Power Electronics Conference and Exposition(APEC).IEEE, 2019.", proposes a multi-time scale Foster thermal network model considering thermal coupling in IGBT modules, which simplifies the profile analysis of long-term missions.
[0010] In summary, current research on improving the computational efficiency of heat network models has the following shortcomings:
[0011] 1) There are very few studies in the existing literature on reducing the computational burden by simplifying the thermal network model;
[0012] 2) Reference 4 presents a simplified study of the Foster network, but the Foster network cannot describe the actual physical meaning. Further research is needed on a compact Cauer thermal network model with physical meaning. Summary of the Invention
[0013] The problem this invention aims to solve is the conflict between the accuracy and efficiency of junction temperature calculation in multi-chip IGBT power module thermal network models, in order to compensate for the shortcomings of compact Cauer thermal network models for power modules in improving computational efficiency.
[0014] This invention is achieved through the following technical solution:
[0015] A method for determining the thermal decoupling boundary conditions of a compact thermal network model, wherein the compact thermal network model is used to describe the heat transfer behavior of one IGBT chip and one Diode chip in an IGBT power module, and considers the thermal coupling effect between the two chips.
[0016] The compact thermal network model includes 6 temperature nodes, 2 input loss current sources, 8 thermal conductivities, and 6 thermal capacities; the 6 temperature nodes are denoted as temperature node N. i i = 1, 2, 3, 4, 5, 6, temperature node N i The temperature is denoted as the junction temperature T. i i = 1, 2, 3, 4, 5, 6; the eight thermal conductivities are denoted as thermal conductivities G. 1_3 Thermal conductivity G 2_4 Thermal conductivity G 3_5 Thermal conductivity G 4_6 Thermal conductivity G 3_4 Thermal conductivity G 5_6 Thermal conductivity G 5_0 Thermal conductivity G 6_0 The six heat capacities are denoted as heat capacity C. i i = 1, 2, 3, 4, 5, 6; the two input loss current sources are respectively denoted as input loss current source P. T and input loss current source P D ;
[0017] The compact thermal network model is a two-dimensional Cauer thermal network structure, divided into three layers: temperature nodes N1 and N2 form the first layer, temperature nodes N3 and N4 form the second layer, and temperature nodes N5 and N6 form the third layer. Temperature nodes N1, N3, and N5 are aligned in the vertical direction, while temperature nodes N2, N4, and N6 are aligned in the other vertical direction.
[0018] With input loss current source P T Starting from P, we form an IGBT substring. Specifically, the input loss current source P... T One end is grounded, and the other end is connected to the temperature node N1. Thermal conductivity G 1_3 Located between temperature node N1 and temperature node N3, thermal conductivity G 3_5 Located between temperature junctions N3 and N5, thermal conductivity G 5_0 One end of the capacitor is connected to temperature node N5, and the other end is grounded. One end of the heat capacitors C1, C3, and C5 is connected to temperature nodes N1, N3, and N5 respectively, and the other end is grounded. The input loss current source P... DA Diode substring is formed starting from the input loss current source P. D One end is grounded, and the other end is connected to the temperature junction N2. Thermal conductivity G 2_4 Located between temperature node N2 and temperature node N4, thermal conductivity G 4_6 Located between temperature node N4 and temperature node N6, thermal conductivity G 6_0 One end of the heat capacity C2 is connected to temperature node N6, and the other end is grounded. One end of the heat capacity C2, heat capacity C4, and heat capacity C6 are connected to temperature node N2, temperature node N4, and temperature node N6 respectively, and the other end is grounded.
[0019] Thermal conductivity G 3_4 The thermal conductivity G is set between the temperature node N3 of the IGBT substring and the temperature node N4 of the Diode substring. 5_6 It is set between the temperature node N5 of the IGBT substring and the temperature node N6 of the Diode substring;
[0020] The specific steps for determining the thermal decoupling boundary conditions are as follows:
[0021] Step 1: Establish the state-space equations of the compact thermal network model.
[0022] Let the compact thermal network model be denoted as thermal network model M1, and the state-space equation of thermal network model M1 is expressed as follows:
[0023]
[0024] Where t is time, and A is the coefficient matrix of the heat network model M1, its expression is:
[0025]
[0026] B is the input matrix of the thermal network model M1, and its expression is:
[0027]
[0028] C is the output matrix of the heat network matrix M1, and its expression is:
[0029] C = [0 1 0 0 0 0];
[0030] T is the node temperature matrix of the thermal network matrix M1, and its expression is:
[0031] T = [T1 T2 T3 T4 T5 T6] T ;
[0032] P is the input loss matrix of the thermal network matrix M1, and its expression is:
[0033] P = [P]T P D ] T ;
[0034] Step 2: Establish the transfer functions of heat network models M1 and M2.
[0035] Step 2.1, establish the transfer function and transfer function
[0036] Based on the state-space equation of the heat network model M1 obtained in step 1, the transfer function expression of the heat network model M1 in the complex frequency domain is obtained as follows:
[0037]
[0038] Where s is the Laplacian operator and I is the identity matrix. The input loss current source P of the thermal network model M1 T The transfer function to the node temperature T2, The input loss current source P of the thermal network model M1 D The transfer function to the node temperature T2;
[0039] Step 2.2, establish the transfer function
[0040] Remove the IGBT substrings and thermal conductivity G from the thermal network model M1. 3_4 Thermal conductivity G 5_6 A one-dimensional Cauer thermal network model is obtained, denoted as thermal network model M2; in the complex frequency domain, the input loss current source P of thermal network model M2 is... D The transfer function expression to the node temperature T2 is:
[0041]
[0042] Step 3, solve for the coupling coefficient ΔG(w)
[0043] Using the transfer function obtained in equation (1) and transfer function The transfer function obtained in equation (2) Establish coupling coefficient ΔC Z The expression for (w) is:
[0044]
[0045] Where j is the complex unit, and w is the coupling coefficient ΔG Z angular frequency of (w);
[0046] Step 4: Solve the thermal decoupling boundary conditions
[0047] Using the transfer function obtained in equation (1) The coupling coefficient ΔG obtained in equation (3) Z (w), the expression for solving the thermal decoupling boundary conditions is established as follows:
[0048]
[0049] In the formula, L -1 {} represents the inverse Laplace transform, ΔG Z The coupling coefficient ΔG Z The magnitude of (w) and ΔG Z =|ΔG Z (w)|,P T (s) represents the input loss current source P. T The Laplace transform form, P D (s) represents the input loss current source P. D The Laplace transform form, E D The given error.
[0050] The method for determining the thermal decoupling boundary conditions of a compact thermal network model provided by this invention allows for the decomposition of the compact thermal network model into independent thermal network models for temperature calculation when the compact thermal network model satisfies the thermal decoupling boundary conditions. This improves the computational efficiency of the compact thermal network model while maintaining a given accuracy. Compared with traditional temperature calculations using compact thermal network models, the beneficial results of this invention are reflected in:
[0051] This invention is applicable to compact thermal network models of multi-chip IGBT power modules that consider thermal coupling effects. The proposed method for determining thermal decoupling boundary conditions can decompose the compact thermal network model into independent thermal network models for calculation when the compact thermal network model meets the thermal decoupling boundary conditions. This not only ensures the accuracy of junction temperature calculation for IGBT power modules, but also simplifies the compact thermal network model and reduces the computational burden of complex compact thermal network models. Attached Figure Description
[0052] Figure 1 This is a flowchart of the method for determining the present invention.
[0053] Figure 2 This is a structural diagram of the compact thermal network model M1 involved in the present invention.
[0054] Figure 3 This is a structural diagram of the thermal network model M2 involved in the present invention.
[0055] Figure 4This is a comparison curve of the Diode chip junction temperature calculated by the compact thermal network model and the Diode chip junction temperature calculated by the independent thermal network model in a given instance, under the condition of thermal decoupling boundary conditions.
[0056] Figure 5 This is a comparison curve of the Diode chip junction temperature calculated by the compact thermal network model and the Diode chip junction temperature calculated by the independent thermal network model in a given instance, without satisfying the thermal decoupling boundary conditions. Detailed Implementation
[0057] The present invention will be further described in detail below with reference to the accompanying drawings.
[0058] This invention provides a method for determining the thermal decoupling boundary conditions of a compact thermal network model. The compact thermal network model is used to describe the heat transfer behavior of one IGBT chip and one Diode chip in an IGBT power module, and considers the thermal coupling effect between the two chips.
[0059] Figure 2 This is a structural diagram of the compact thermal network model M1 involved in the present invention. Figure 2 As can be seen, the compact thermal network model involved in the definition method of the present invention includes 6 temperature nodes, 2 input loss current sources, 8 thermal conductivities, and 6 thermal capacities; the 6 temperature nodes are denoted as temperature node N. i i = 1, 2, 3, 4, 5, 6, temperature node N i The temperature is denoted as the junction temperature T. i i = 1, 2, 3, 4, 5, 6; the eight thermal conductivities are denoted as thermal conductivities G. 1_3 Thermal conductivity G 2_4 Thermal conductivity G 3_5 Thermal conductivity G 4_6 Thermal conductivity G 3_4 Thermal conductivity G 5_6 Thermal conductivity G 5_0 Thermal conductivity G 6_0 The six heat capacities are denoted as heat capacity C. i i = 1, 2, 3, 4, 5, 6; the two input loss current sources are respectively denoted as input loss current source P. T and input loss current source P D .
[0060] The compact thermal network model is a two-dimensional Cauer thermal network structure, divided into three layers: temperature nodes N1 and N2 form the first layer, temperature nodes N3 and N4 form the second layer, and temperature nodes N5 and N6 form the third layer. Temperature nodes N1, N3, and N5 are aligned in the vertical direction, while temperature nodes N2, N4, and N6 are aligned in the other vertical direction.
[0061] With input loss current source P T Starting from P, we form an IGBT substring. Specifically, the input loss current source P... T One end is grounded, and the other end is connected to the temperature node N1. Thermal conductivity G 1_3 Located between temperature node N1 and temperature node N3, thermal conductivity G 3_5 Located between temperature junctions N3 and N5, thermal conductivity G 5_0 One end of the capacitor is connected to temperature node N5, and the other end is grounded. One end of the heat capacitors C1, C3, and C5 is connected to temperature nodes N1, N3, and N5 respectively, and the other end is grounded. The input loss current source P... D A Diode substring is formed starting from the input loss current source P. D One end is grounded, and the other end is connected to the temperature junction N2. Thermal conductivity G 2_4 Located between temperature node N2 and temperature node N4, thermal conductivity G 4_6 Located between temperature node N4 and temperature node N6, thermal conductivity G 6_0 One end of the heat capacity C2 is connected to temperature node N6, and the other end is grounded. One end of the heat capacity C2, heat capacity C4, and heat capacity C6 are connected to temperature nodes N2, N4, and N6 respectively, and the other end is grounded.
[0062] Thermal conductivity G 3_4 The thermal conductivity G is set between the temperature node N3 of the IGBT substring and the temperature node N4 of the Diode substring. 5_6 It is set between the temperature node N5 of the IGBT substring and the temperature node N6 of the Diode substring.
[0063] Figure 1 This is a flowchart illustrating the method for defining the thermal decoupling boundary conditions according to the present invention. As shown in the diagram, the specific steps of the method for determining the thermal decoupling boundary conditions are as follows:
[0064] Step 1: Establish the state-space equations of the compact thermal network model.
[0065] Let the compact thermal network model be denoted as thermal network model M1, and the state-space equation of thermal network model M1 is expressed as follows:
[0066]
[0067] Where t is time, and A is the coefficient matrix of the heat network model M1, its expression is:
[0068]
[0069] B is the input matrix of the thermal network model M1, and its expression is:
[0070]
[0071] C is the output matrix of the heat network matrix M1, and its expression is:
[0072] C = [0 1 0 0 0 0];
[0073] T is the node temperature matrix of the thermal network matrix M1, and its expression is:
[0074] T = [T1 T2 T3 T4 T5 T6] T ;
[0075] P is the input loss matrix of the thermal network matrix M1, and its expression is:
[0076] P = [P] T P D ] T .
[0077] In this embodiment, the thermal conductivity and heat capacity of the compact thermal network model are shown in the table below.
[0078]
[0079] Step 2: Establish the transfer functions of heat network models M1 and M2.
[0080] Step 2.1, establish the transfer function and transfer function
[0081] Based on the state-space equation of the heat network model M1 obtained in step 1, the transfer function expression of the heat network model M1 in the complex frequency domain is obtained as follows:
[0082]
[0083] Where s is the Laplacian operator and I is the identity matrix. The input loss current source P of the thermal network model M1 T The transfer function to the node temperature T2, The input loss current source P of the thermal network model M1 D The transfer function to the junction temperature T2.
[0084] Step 2.2, establish the transfer function
[0085] Remove the IGBT substrings and thermal conductivity G from the thermal network model M1. 3_4 Thermal conductivity G 5_6This yields a one-dimensional Cauer structure thermal network model, denoted as thermal network model M2. The structure of thermal network model M2 is shown below. Figure 3 .
[0086] In the complex frequency domain, the input loss current source P of the thermal network model M2 D The transfer function expression to the node temperature T2 is:
[0087]
[0088] Step 3, solve for the coupling coefficient ΔG(w)
[0089] Using the transfer function obtained in equation (1) and transfer function The transfer function obtained in equation (2) Establish coupling coefficient ΔG Z The expression for (w) is:
[0090]
[0091] Where j is the complex unit, and w is the coupling coefficient ΔG Z The angular frequency of (w).
[0092] Step 4: Solve the thermal decoupling boundary conditions
[0093] Using the transfer function obtained in equation (1) The coupling coefficient ΔG obtained in equation (3) Z (w), the expression for solving the thermal decoupling boundary conditions is established as follows:
[0094]
[0095] In the formula, L -1 {} represents the inverse Laplace transform, ΔG Z The coupling coefficient ΔG Z The magnitude of (w) and ΔG Z =|ΔG Z (w)|,P T (s) represents the input loss current source P. T The Laplace transform form, P D (s) represents the input loss current source P. D The Laplace transform form, E D The given error.
[0096] To verify the effectiveness of the present invention, simulation verification was performed.
[0097] The simulation parameters are set as follows:
[0098] P T=75sin(30t)+75|sin(30t)|
[0099] P D =60sin(30t)+60|sin(30t)|
[0100] Coupling coefficient ΔG Z The angular frequency of (w) is set to 30 rad / s, and the given error E D Set to 2℃, the given parameters satisfy the expression for the thermal decoupling boundary conditions. Calculate the node temperatures of the compact thermal network model M1 and the independent thermal network model in MATLAB, and extract the node temperature (Diode chip junction temperature) of temperature node N2 for comparison.
[0101] Figure 4 To compare the results in a graph, from Figure 4 It is evident that when the thermal decoupling boundary conditions are met, the temperature curves calculated by the two methods show a high degree of agreement, with a maximum error of 1.5℃, which is less than the given error. In this case, the compact thermal network model can be simplified into an independent thermal network model for calculation.
[0102] The simulation parameters are set as follows:
[0103] P T =80sin(314t)+80|sin(314t)|
[0104] P D =70sin(314t)+70|sin(314t)|
[0105] Coupling coefficient ΔG Z The angular frequency of (w) is set to 314 rad / s, and the given error E D The temperature is set to 2℃, and the given parameters do not satisfy the expression for the thermal decoupling boundary conditions. The node temperatures of the compact thermal network model M1 and the independent thermal network model are calculated in MATLAB, and the node temperature (Diode chip junction temperature) of temperature node N2 is extracted and compared.
[0106] Figure 5 To compare the results in a graph, from Figure 4 It is evident that when the thermal decoupling boundary conditions are not met, the temperature curves calculated by the two methods do not match well, with a maximum error of 4℃, which is greater than the given error. In this case, the compact thermal network model cannot be simplified into an independent thermal network model for calculation.
Claims
1. A method for determining the thermal decoupling boundary conditions of a compact thermal network model, wherein the compact thermal network model is used to describe the heat transfer behavior of one IGBT chip and one Diode chip in an IGBT power module, and considers the thermal coupling effect between the two chips; The compact thermal network model includes 6 temperature nodes, 2 input loss current sources, 8 thermal conductors, and 6 heat capacities; the 6 temperature nodes are denoted as temperature node N. i i = 1, 2, 3, 4, 5, 6, temperature node N i The temperature is denoted as the junction temperature T. i i = 1, 2, 3, 4, 5, 6; the eight thermal conductivities are denoted as thermal conductivities G. 1_3 Thermal conductivity G 2_4 Thermal conductivity G 3_5 Thermal conductivity G 4_6 Thermal conductivity G 3_4 Thermal conductivity G 5_6 Thermal conductivity G 5_0 Thermal conductivity G 6_0 The six heat capacities are denoted as heat capacity C. i i = 1, 2, 3, 4, 5, 6; the two input loss current sources are respectively denoted as input loss current source P. T and input loss current source P D ; The compact thermal network model is a two-dimensional Cauer thermal network structure, divided into three layers: temperature nodes N1 and N2 form the first layer, temperature nodes N3 and N4 form the second layer, and temperature nodes N5 and N6 form the third layer. Temperature nodes N1, N3, and N5 are aligned in the vertical direction, while temperature nodes N2, N4, and N6 are aligned in the other vertical direction. With input loss current source P T Starting from P, we form an IGBT substring. Specifically, the input loss current source P... T One end is grounded, and the other end is connected to the temperature node N1. Thermal conductivity G 1_3 Located between temperature node N1 and temperature node N3, thermal conductivity G 3_5 Located between temperature junctions N3 and N5, thermal conductivity G 5_0 One end of the capacitor is connected to temperature node N5, and the other end is grounded. One end of the heat capacitors C1, C3, and C5 is connected to temperature nodes N1, N3, and N5 respectively, and the other end is grounded. The input loss current source P... D A Diode substring is formed starting from the input loss current source P. D One end is grounded, and the other end is connected to the temperature junction N2. Thermal conductivity G 2_4 Located between temperature node N2 and temperature node N4, thermal conductivity G 4_6 Located between temperature node N4 and temperature node N6, thermal conductivity G 6_0 One end of the heat capacity C2 is connected to temperature node N6, and the other end is grounded. One end of the heat capacity C2, heat capacity C4, and heat capacity C6 are connected to temperature node N2, temperature node N4, and temperature node N6 respectively, and the other end is grounded. Thermal conductivity G 3_4 The thermal conductivity G is set between the temperature node N3 of the IGBT substring and the temperature node N4 of the Diode substring. 5_6 It is set between the temperature node N5 of the IGBT substring and the temperature node N6 of the Diode substring; Its features are, The specific steps for determining the thermal decoupling boundary conditions are as follows: Step 1: Establish the state-space equations of the compact thermal network model. Let the compact thermal network model be denoted as thermal network model M1, and the state-space equation of thermal network model M1 is expressed as follows: Where t is time, and A is the coefficient matrix of the heat network model M1, its expression is: B is the input matrix of the thermal network model M1, and its expression is: C is the output matrix of the heat network matrix M1, and its expression is: C＝[0 1 0 0 0 0]； T is the node temperature matrix of the thermal network matrix M1, and its expression is: T=[T1 T2 T3 T4 T5 T6] T ; P is the input loss matrix of the thermal network matrix M1, and its expression is: P=[P T P D ] T ; Step 2: Establish the transfer functions of heat network models M1 and M2. Step 2.1, establish the transfer function and transfer function Based on the state-space equation of the heat network model M1 obtained in step 1, the transfer function expression of the heat network model M1 in the complex frequency domain is obtained as follows: Where s is the Laplacian operator and I is the identity matrix. The input loss current source P of the thermal network model M1 T The transfer function to the node temperature T2, The input loss current source P of the thermal network model M1 D The transfer function to the node temperature T2; Step 2.2, establish the transfer function Remove the IGBT substrings and thermal conductivity G from the thermal network model M1. 3_4 Thermal conductivity G 5_6 A one-dimensional Cauer thermal network model is obtained, denoted as thermal network model M2; in the complex frequency domain, the input loss current source P of thermal network model M2 is... D The transfer function expression to the node temperature T2 is: Step 3, solve for the coupling coefficient ΔG(w) Using the transfer function obtained in equation (1) and transfer function The transfer function obtained in equation (2) Establish coupling coefficient ΔC Z The expression for (w) is: Where j is the complex unit, and w is the coupling coefficient ΔG Z angular frequency of (w); Step 4: Solve the thermal decoupling boundary conditions Using the transfer function obtained in equation (1) The coupling coefficient ΔG obtained in equation (3) Z (w), the expression for solving the thermal decoupling boundary conditions is established as follows: In the formula, L -1 {} denotes the inverse Laplace transform, ΔG z The coupling coefficient ΔG z The magnitude of (w) and ΔG Z =|ΔG Z (w)|,P T (s) represents the input loss current source P. T The Laplace transform form, P D (s) represents the input loss current source P. D The Laplace transform form, E D Given the error.