IGBT module aging characteristic decoupling parameter monitoring method based on temperature characteristic

Abstract

The application discloses an IGBT module aging characteristic decoupling parameter monitoring method based on temperature characteristics, and specifically comprises the following steps: establishing an IGBT-radiator system thermal network model, iteratively calculating the steady-state temperature difference of each loss period starting node and the steady-state temperature difference of the half loss period node, and constructing an aging characteristic parameter; combining the IGBT module thermal resistance, the saturation voltage drop failure standard, deducing an aging factor calculation formula, and establishing a mapping relationship between the aging characteristic parameter and the aging degree; fixing the working condition, measuring the junction temperature curve of the healthy module under the periodic loss, and calculating the aging characteristic parameter value of the healthy state; measuring the junction temperature curve of any module with different aging degrees under the periodic loss, and calculating the aging characteristic parameter value of the module; and finally calculating the heat dissipation path and the bonding wire aging level of the measured module. The aging characteristic decoupling parameter extraction is more simple and convenient, has small invasion and low cost, decouples the failure mode, and is more comprehensive in representation.

Description

Technical Field

[0001] This invention belongs to the field of reliability assessment technology for power electronic devices, and specifically relates to a method for monitoring decoupling parameters of IGBT module aging characteristics based on temperature characteristics. Background Technology

[0002] As a high-power switching device, the Insulated-Gate Bipolar Transistor (IGBT) is a fundamental core component of converters, and its reliability directly affects the safe operation of the equipment. However, due to long-term operation under varying conditions and harsh environments, the reliability of IGBTs faces significant challenges. Related industrial statistics show that IGBT failures account for more than 25% of all converter system failures. In field applications, due to the difference in thermal expansion coefficients between adjacent material layers, thermal cycling shocks caused by losses generate mechanical stress between the material layers. Long-term mechanical stress leads to a series of aging phenomena, including solder layer cracks, thermal grease voids, and bond wire separation. Aging affects the electrical characteristics and thermal behavior of the module, reducing its reliability and causing it to fail. Therefore, timely knowledge of module aging is crucial for IGBT reliability analysis. Monitoring aging-related parameters to determine the degree of module degradation is an important method in power device reliability research.

[0003] In recent years, several aging characteristic parameters for monitoring heat dissipation path degradation and bond wire separation have been proposed. Based on their source, these parameters can be mainly divided into thermal parameters and electrical parameters. Tang Shengxue, Ma Qiang, Chen Dong, and others used the module thermal resistance increase rate as an aging characteristic parameter to address the chip solder layer void failure problem. By pre-extracting transient thermal resistance curves and calculating the thermal resistance increase rate, they obtained its relationship with the degree of solder layer void growth. Although the thermal resistance increase rate can characterize the aging process of chip solder layer voids, extracting transient thermal resistance curves during experiments requires interrupting system operation, making online monitoring difficult. Furthermore, it does not cover other possible failure modes, limiting its practical engineering application. For the bond wire aging problem, Sun Pengju, Wang Haibo, Gong Can, and others selected short-circuit current as the aging characteristic quantity of IGBT module bond wires. By establishing a module equivalent resistance network model encompassing the bond wires, they analyzed the impact of bond wire aging on the resistance network model, obtaining the mapping relationship between short-circuit current and aging; however, they did not consider its coupling with temperature. Dai Jian, Ge Xinglai, Yang Xu, and others used collector-emitter saturation on-state voltage drop as a characteristic parameter of bond wire aging, monitoring the correlation between the number of bond wires detached due to aging and the collector-emitter saturation on-state voltage drop. Compared to the former, this method dynamically and adaptively eliminates the influence of junction temperature by constructing a reference dataset, achieving temperature decoupling. However, it requires adding a more complex monitoring circuit to pre-construct the reference dataset, which is highly invasive and costly. In contrast to the aforementioned single failure mode aging monitoring methods, M. Du, Q. Kong, Z. Ouyang, and others, targeting solder layer and bond wire coupling failures, divided the module's collector-emitter saturation on-state voltage drop into three parts based on a voltage separation method: collector-emitter threshold voltage, chip on-state voltage, and package voltage, using these as aging characteristic parameters for joint aging monitoring. Compared to single aging mode monitoring, this method is more comprehensive, decoupling and separating bond wire aging and solder layer aging failure modes, resulting in clearer identification, but it still requires adding a more complex measurement circuit.

[0004] In summary, current aging characteristic parameters and monitoring methods often suffer from problems such as difficulty in parameter extraction, limited aging mode identification, complex monitoring methods, difficulty in online monitoring, high invasiveness, and high cost. Therefore, developing an aging characteristic parameter that is easy to extract and can decouple aging modes is of great significance for the reliability analysis of IGBT modules. Summary of the Invention

[0005] To achieve low-intrusion, easy-to-extract, and multi-failure-mode-identifiable online monitoring of IGBT aging, this invention provides a method for monitoring decoupling parameters of IGBT module aging characteristics based on temperature characteristics.

[0006] The present invention provides a method for monitoring decoupling parameters of IGBT module aging characteristics based on temperature characteristics, comprising the following steps:

[0007] Step 1: Calculate the periodic losses of the IGBT and extract the power loss curve; establish a thermal network model of the IGBT-heatsink system and calculate the instantaneous temperature difference ΔT of the IGBT at the beginning of each loss cycle. low (i) The instantaneous temperature difference ΔT between the half-loss period node and the half-loss period node high (i) The steady-state temperature difference ΔT at the starting node of each loss cycle is obtained through iterative calculation. low and the steady-state temperature difference ΔT at the half-loss periodic node high Combining the heat dissipation path and loss calculation formula of the IGBT module, ΔT high With ΔT low Divide by ΔT to construct aging characteristic parameter fun1, and then divide ΔT by ΔT. low With ΔT high Add them together to construct the aging characteristic parameter fun2.

[0008] Step 2: Combining the thermal resistance and saturation voltage drop failure criteria of IGBT modules, derive the calculation formulas for aging degree factors Δfun1 and Δfun2, and establish the mapping relationship between aging characteristic parameters fun1 and fun2 and aging.

[0009] Step 3: Under fixed operating conditions, measure the junction temperature curve of the health module under cycle loss and extract ΔT. high With ΔT low Calculate the aging characteristic parameter values ​​of the health status, namely fun 1-h and fun 2-h Measure the junction temperature curves of any module with different aging levels under cycle loss, and extract ΔT. high With ΔT low The aging characteristic parameter value of this module is calculated to be fun. 1-ag and fun 2-ag .

[0010] Step 4: Based on the fun obtained in Step 3 1-h fun 2-h fun 1-ag and fun 2-ag Calculate Δfun1 and Δfun2; based on the mapping relationship in step 2, obtain the heat dissipation path and bonding wire aging level of the module under test from Δfun1 and Δfun2.

[0011] Furthermore, due to circuit control limitations in step 1, the power loss in the module is half-sine, and the average switching loss P of the module is... switch and on-state loss P conduct The calculation formula is:

[0012]

[0013] Among them, I c For collector current, The power factor angle is M, the modulation ratio is f. sw V is the switching frequency. DC For DC voltage, T j Junction temperature; K V C is the voltage correction factor. T E is the temperature correction factor. on+off For switching energy; I ref V ref and T ref These are the reference values ​​for current, voltage, and temperature, respectively; R ce and V ce The junction temperatures T obtained from the table are respectively j The corresponding collector-emitter resistance and saturation voltage.

[0014] Total loss of IGBT module P loss for:

[0015]

[0016] Where n is 1, 2, 3, ..., t is time, and ω is the loss angular frequency.

[0017] Based on the principle of equal impulse, the power loss of a half-sine waveform can be equivalent to a square wave, and the calculation formula is as follows:

[0018]

[0019] Where P is the square wave amplitude equivalent to the loss waveform.

[0020] Commonly used thermal network models include Foster thermal networks and Cauer thermal networks. However, Cauer thermal networks require calculations based on specific module types, materials, and dimensions, making their creation process complex and limiting their adaptability. Therefore, Foster thermal networks are used to model IGBT modules and heat sink systems. Based on heat transfer theory, the IGBT thermal response caused by power loss is expressed as:

[0021]

[0022] Where, ΔT initial It is the initial temperature difference of the IGBT, and the time constant τ. th Thermal resistance R th and heat capacity C th The product of P loss For the power loss of the IGBT module, ΔT j(t) represents the real-time temperature difference of the IGBT; when the temperature difference of the IGBT module has not reached a steady state, the temperature difference between the starting node and the half-loss cycle node in the IGBT temperature characteristics changes iteratively with the loss cycle; the instantaneous temperature difference ΔT of the calculated node is extracted. low (i) and ΔT high (i) Where i is the number of loss cycles; ΔT low (i) and ΔT high (i) The calculation formula is as follows:

[0023]

[0024] Where f is the loss frequency; assuming the node temperature difference reaches a steady state after the nth cycle, ΔT can be obtained through iterative calculation using the above formula for calculating the instantaneous node temperature difference. low and ΔT high The expression is as follows:

[0025]

[0026] steady-state temperature difference ΔT at nodes low and ΔT high This can be obtained by monitoring the IGBT temperature profile; aging of the bond wire increases the bonding resistance, and aging of the solder layer leads to a shrinkage of the heat dissipation path, affecting R. th and P loss Size; according to the above expression, P loss R th and τ th Both will affect ΔT low and ΔT high The size cannot be determined by ΔT low and ΔT high The changes determine the failure mode; therefore, the aging characteristic parameters fun1 and fun2 are constructed, with the following expressions:

[0027]

[0028] From equation (7), it can be derived that for the aging characteristic parameter fun1, the steady-state ΔT is calculated. low and ΔT high Comparing the two, we can eliminate the interference of P and obtain only the result related to τ. th Parameters related to f; when f is constant, fun1 can uniquely represent τ. th The change in heat capacity during aging is negligible; therefore, fun1 can solely reflect R. th The changes in P and R; and for the aging characteristic parameter fun2, according to the formula analysis, it is affected by P and R. thThe impact; given that τ can be determined in advance via fun1 th R is determined by the time constant calculation formula. th Then, based on the formula for calculating fun2, the known quantity R is excluded. th This allows for a single determination of the change in P; according to the aging mechanism, bond wire aging mainly affects the magnitude of power loss, while solder layer aging shrinks the heat dissipation path, increasing R. th Therefore, based on the above evaluation strategy, fun1 and fun2 can be used together to characterize and evaluate the aging of heat dissipation paths and bond lines.

[0029] Furthermore, in step 2, in the study of IGBT module aging failure, the general aging evaluation criterion is thermal resistance R. th Rise by 20%, V ce Increase by 5%. Based on the aging characteristic parameter fun1, assuming that after aging, R th The value of the aging characteristic parameter fun1 is increased from k1 times the value of the healthy state to fun1. 1-h Turning into aged fun 1-ag Then we have:

[0030]

[0031] Therefore:

[0032]

[0033] Where Δfun1 is the aging factor of the aging characteristic parameter fun1; according to the above formula, comparing fun 1-ag and fun 1-h And Δfun1, which gives the degree of aging of the heat dissipation path; then, for the aging characteristic parameter fun2, since the change in thermal resistance caused by the aging of the heat dissipation path has already been reflected and corrected in fun1, therefore, R is assumed in fun2. th This is a constant and does not characterize the aging of the heat dissipation path, but rather the V value caused by the aging of the bond wire. ce The increase is reflected in P. Based on the square wave loss amplitude calculated using the loss calculation formula and the impulse equality, and the expression for the aging characteristic parameter fun2, we have:

[0034]

[0035] Among them, a1, b1, and b2 are related to V ce The irrelevant value is a constant, which can be calculated according to equation (1); assuming that after the bond line ages, V ce The value of the aging characteristic parameter fun2 increases from the fun value in the healthy state to k2 times. 2-h Turning into aged fun 2-ag We can deduce that:

[0036]

[0037] Among them, according to fun 2-ag The expression can be further simplified to obtain the relationship between health status and aging status as follows:

[0038]

[0039] Where Δfun2 is the aging factor of the aging characteristic parameter fun2; and V ce The value for the health status is a constant and can be obtained from the module manual; a1 can be calculated based on the circuit operating conditions; R th It is a constant; when k2 is 1, Δfun2 is 0, indicating the module is healthy; when k2 is 1.05, Δfun2 reaches its maximum value, indicating the module is aging and failing. Therefore, according to fun... 2-ag With fun 2-h The difference between them can characterize the aging level of the module bond wires.

[0040] Furthermore, in step 3, the temperature curves of the healthy module and the aging module under test are measured under periodic wear. This can be done without interrupting the normal operation of the modules, and can be calculated based on the temperature sensor integrated into the module itself, or measured using a thermometer.

[0041] The beneficial technical effects of this invention are as follows:

[0042] (1) The IGBT aging characteristic parameters proposed in this invention only require monitoring the temperature of the IGBT module. The monitoring point is single, the monitoring is convenient, and the cost is low.

[0043] (2) The IGBT aging characteristic parameters proposed in this invention can be monitored in real time without interrupting the module operation, and have a wider range of applications.

[0044] (3) The IGBT aging characteristic parameters proposed in this invention can identify two main aging modes, decouple the characterization of aging, and monitor more comprehensively. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the three-phase inverter topology and the module under test in an embodiment of the present invention.

[0046] Figure 2 This is a schematic diagram illustrating the equivalent of square wave periodic loss and half-sine periodic loss in an embodiment of the present invention.

[0047] Figure 3 The diagram shows an example of a heat sink and an equivalent thermal resistance diagram of the heat sink fins in the embodiment of the present invention.

[0048] Figure 4The temperature difference curves at the starting node and half-loss cycle node of the loss cycle obtained in this embodiment of the invention are shown.

[0049] Figure 5 The figures show the junction temperature and thermal resistance changes before and after aging of the simulated heat dissipation path in this embodiment of the invention.

[0050] Figure 6 The figures show the changes in the aging characteristic parameters fun1 and fun2 before and after aging of the simulated heat dissipation path in this embodiment of the invention.

[0051] Figure 7 The figures shown are curves illustrating the changes in junction temperature and loss before and after aging of the bond wire in an embodiment of the present invention.

[0052] Figure 8 The curves showing the changes in characteristic parameters fun1 and fun2 after the simulated bonding wire aging in this embodiment of the invention are shown.

[0053] Figure 9 This is a flowchart of the decoupling parameters and aging monitoring method for IGBT module aging characteristics based on temperature characteristics according to the present invention. Detailed Implementation

[0054] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0055] This invention discloses a method for monitoring decoupling parameters of IGBT module aging characteristics based on temperature characteristics. The IGBT module analyzed is an Infineon-FP25R12W2T4 IGBT module, applied in an SPWM-modulated inverter to verify the aging characteristic parameters fun1 and fun2. Specifically, it includes the following steps:

[0056] Step 1: Calculate the periodic losses of the IGBT and extract the power loss curve; establish a thermal network model of the IGBT-heatsink system and calculate the instantaneous temperature difference ΔT of the IGBT at the beginning of each loss cycle. low (i) The instantaneous temperature difference ΔT between the half-loss period node and the half-loss period node high (i) The steady-state temperature difference ΔT at the starting node of each loss cycle is obtained through iterative calculation. low and the steady-state temperature difference ΔT at the half-loss periodic node high Combining the heat dissipation path and loss calculation formula of the IGBT module, ΔT high With ΔT low Divide by ΔT to construct aging characteristic parameter fun1, and then divide ΔT by ΔT. low With ΔT high Add them together to construct the aging characteristic parameter fun2.

[0057] The three-phase full-bridge inverter uses SPWM control, with the control signal driven by the TMS320F28335 chip integrated on the DSP development board. The inverter's operating frequency is set to 0.1Hz. Due to the symmetry of the three-phase inverter, the IGBT module in the lower arm of phase C is selected as the device under test (DUT). Figure 1 As shown in Table 1, during the module's operation, the electrical signals required for IGBT module losses are acquired using IMC's Devices data acquisition system. The losses and corresponding junction temperatures are calculated, and a laser thermometer is used to measure the junction temperature of the module under test. Table 1 shows the inverter's operating conditions.

[0058] Table 1 Operating conditions of three-phase inverter

[0059]

[0060]

[0061] Due to circuit control limitations, the power loss in the module is typically half-sine. The formulas for calculating the average switching loss and conduction loss of this module are as follows:

[0062]

[0063] Among them, I c For collector current, The power factor angle is M, the modulation ratio is f. sw V is the switching frequency. DC For DC voltage, T j This is the junction temperature. (K) V C is the voltage correction factor. T E is the temperature correction factor. on+off For switching energy. I ref V ref and T ref These are reference values ​​for current, voltage, and temperature, respectively. R ce and V ce The junction temperatures T obtained from the table are respectively j The corresponding collector-emitter resistance and saturation voltage. The total loss of the IGBT module is:

[0064]

[0065] Where n is 1, 2, 3, ..., t is time, and ω is the loss angular frequency. For example... Figure 2 According to the principle of equal impulse, the power loss of a half-sine waveform can be equivalent to a square wave, and the calculation formula is as follows:

[0066]

[0067] Where P is the square wave amplitude equivalent to the loss waveform.

[0068] Commonly used thermal network models include Foster thermal networks and Cauer thermal networks. However, Cauer thermal networks require calculations based on specific module types, materials, and dimensions, making the modeling process complex and limiting their adaptability. Therefore, Foster thermal networks are used to model the IGBT module and heatsink system. Each fin in the Infineon-FP25R12W2T4 IGBT module heatsink is considered a heat dissipation unit, without considering heat capacity. Figure 3 As shown. Based on the material and forced air cooling formula, the thermal resistance R of the fins in the heat sink is... th The calculation formula is as follows:

[0069]

[0070] Where S is the surface area of ​​the fins, L is the fin length, q is an empirical coefficient, v is the wind speed, and υ and Pr are the kinematic viscosity of air and Prandtl's constant, respectively. For simplicity, the remaining parts of the radiator and the IGBT module's thermal network are modeled using a Foster thermal network. According to heat transfer theory, the IGBT thermal response caused by power loss is expressed as:

[0071]

[0072] Where, ΔT initial It is the initial temperature difference of the IGBT, and the time constant τ. th Thermal resistance R th and heat capacity C th The product of P loss For the power loss of the IGBT module, ΔT j (t) represents the real-time temperature difference of the IGBT. For example... Figure 4 When the temperature difference of the IGBT module has not reached a steady state, the temperature difference between the starting node and the half-loss cycle node in the IGBT temperature characteristics changes iteratively with the loss cycle. The instantaneous temperature difference ΔT between the calculated nodes is extracted. low (i) and ΔT high (i) ΔT, where i is the number of loss cycles. low (i) and ΔT high (i) The calculation formula is as follows:

[0073]

[0074] Where f is the loss frequency. Assuming the node temperature difference reaches a steady state after the nth cycle, ΔT can be obtained through iterative calculation using the above formula for calculating the instantaneous node temperature difference.low and ΔT high The expression is as follows:

[0075]

[0076] ΔT low and ΔT high This can be obtained by monitoring the IGBT temperature profile and the ambient reference temperature. Existing research indicates that bond wire aging increases bonding resistance, and solder layer aging leads to a shrinkage of the heat dissipation path, affecting R... th and P loss Size. According to the above expression, P loss R th and τ th Both will affect ΔT low and ΔT high The size cannot be determined by ΔT low and ΔT high The changes determine the failure mode. Therefore, the aging characteristic parameters fun1 and fun2 are constructed, with the following expressions:

[0077]

[0078] From the formula, we can derive that for the aging characteristic parameter fun1, the steady-state ΔT is calculated. low and ΔT high Comparing the two, we can eliminate the interference of P and obtain only the result related to τ. th Parameters related to f. When f is constant, fun1 can uniquely represent τ. th The change in heat capacity during aging is very small and can be ignored; therefore, fun1 can solely reflect R. th The changes. As for the aging characteristic parameter fun2, according to the formula analysis, it is affected by P and R. th The impact. Given that τ can be determined in advance via fun1. th R is determined by the time constant calculation formula. th Then, based on the formula for calculating fun2, the known quantity R is excluded. th This allows for a single determination of the change in P. According to the aging mechanism, bond wire aging primarily affects power loss; solder layer aging shrinks the heat dissipation path, increasing R. th Therefore, based on the above evaluation strategy, fun1 and fun2 can be used together to characterize and evaluate the aging of heat dissipation paths and bond lines.

[0079] Step 2: Combining the thermal resistance and saturation voltage drop failure criteria of IGBT modules, derive the calculation formulas for aging factors Δfun1 and Δfun2, and establish the mapping relationship between aging characteristic parameters fun1 and fun2 and aging.

[0080] In studies of IGBT module aging failure, the common aging evaluation criterion is thermal resistance R. th Rise by 20%, V ce Increase by 5%. Based on the aging characteristic parameter fun1, assuming that after aging, R th The value of the aging characteristic fun1 is increased from k1 times the value of the healthy state to fun1. 1-h Turning into aged fun 1-ag Then we have:

[0081]

[0082] Therefore:

[0083]

[0084] Where Δfun1 is the aging factor of the aging characteristic parameter fun1. Based on the above formula, compare fun... 1-ag and fun 1-h And by calculating Δfun1, the aging level of the heat dissipation path can be obtained. Next, for the aging characteristic parameter fun2, since the change in thermal resistance caused by the aging of the heat dissipation path has already been reflected and corrected in fun1, R is assumed to be less significant in fun2. th This is a constant and does not characterize aging, such as aging of bond lines, etc. ce The increase is reflected in P. Based on the square wave loss amplitude calculated using the loss calculation formula and the impulse equality, and the expression for the aging characteristic parameter fun2, we have:

[0085]

[0086] Among them, a1, b1, and b2 are related to V ce The irrelevant value is a constant and can be calculated according to Equation 1. Assuming the bond line ages, V... ce The value of fun2, representing the aging characteristic, increases by a factor of k2 from the value of fun in the healthy state. 2-h Turning into aged fun 2-ag We can deduce that:

[0087]

[0088] Among them, according to fun 2-ag The expression can be further simplified to obtain the relationship between health status and aging status as follows:

[0089]

[0090] Where Δfun2 is the aging factor of the aging characteristic parameter fun2. And V ceThe value for the health status is a constant and can be obtained from the module manual. a1 can be calculated based on the circuit operating conditions. R th This is a constant. When k2 is 1, Δfun2 is 0, indicating the module is healthy. When k2 is 1.05, Δfun2 reaches its maximum value, indicating the module is aging and failing. Therefore, according to fun... 2-ag With fun 2-h The difference between them can characterize the aging of the module bond lines.

[0091] Step 3: Under fixed operating conditions, measure the temperature curve of the health module under periodic losses and extract ΔT. high and ΔT low Calculate the aging characteristic parameter values ​​of the health status, namely fun 1-h and fun 2-h Measure the temperature curve of any IGBT module under cycle loss and extract ΔT. low and ΔT high The aging characteristic parameter value of this module is calculated to be fun. 1-ag and fun 2-ag .

[0092] Studies have shown that aging of the heat dissipation path can increase thermal resistance by 20%. For IGBT module cooling systems, reducing the airflow will increase the thermal resistance of the heat sink. Therefore, for the entire module, reducing the airflow can simulate the aging effect of the heat dissipation path.

[0093] In this experiment, the cooling system airflow velocity was reduced to simulate the degradation of the module's heat dissipation path. At the start of the experiment, the inverter's operating conditions were kept constant. After the tested module's thermal response reached a steady state, at 2900 seconds, the cooling system airflow velocity was reduced to simulate the aging of the IGBT module's heat dissipation path. The module junction temperature was measured, and the aging characteristic parameters fun1 and fun2 curves were calculated.

[0094] according to Figure 5 After reducing the cooling system fan speed, the IGBT module junction temperature rose slowly until it reached a new steady-state temperature. The module temperature was higher than before the fan speed was reduced. Furthermore, the module's thermal resistance increased by approximately 20% after the fan speed reduction, consistent with the aging of the actual heat dissipation path. During heat dissipation path aging, R... th As the value increases, the heat capacity remains essentially unchanged, τ th When the switching frequency f increases while the switching frequency f remains constant, according to the expression for fun1, fun1 should decrease. However, for fun2, R... th When the power loss remains essentially unchanged, fun2 should increase. However, according to... Figure 6 The curves of aging characteristic parameters fun1 and fun2 show that fun1 decreases significantly after 2900s, while fun2 increases, which is consistent with the theoretical derivation.

[0095] Maintaining constant inverter operating conditions and cooling system fan speed, cut the IGBT module bond wires to simulate bond wire aging, and treat this module as an aged module. Start the inverter, acquire the required electrical signals, calculate the module losses, and measure the junction temperature. Figure 7 As shown by the solid line. Subsequently, under the same circuit topology and electrical parameter conditions, newly manufactured IGBT modules of the same model were tested. Based on the data collected from the healthy module under these test conditions, the corresponding losses and junction temperatures of the healthy module were calculated, as follows. Figure 7 As shown by the dashed line. Based on the measured junction temperature of the aging module, the variation curves of the aging characteristic parameters fun1 and fun2 are extracted and calculated, as follows. Figure 8 As shown by the solid line. Next, based on the junction temperature curve of the health module, calculate fun1 and fun2, as follows. Figure 8 As shown by the dashed line.

[0096] according to Figure 7 The junction temperature trends of the aged module and the healthy module were similar, but the aged module was slightly higher overall. Regarding losses, the higher loss values ​​of the aged module compared to the healthy module confirm that cutting the bond wires increases module losses. According to... Figure 8 The aging characteristic parameter fun1 shows little change when comparing the aged and healthy modules. However, compared to fun1, the difference in fun2 between the healthy and aged modules is significant. According to the expressions for fun1 and fun2, since bond wire aging primarily affects module wear, fun1 remains relatively constant, while fun2 is more significantly affected by module wear. Therefore, as wear increases, fun2 also increases, consistent with theoretical derivation.

[0097] By combining theory and experiment, it can be verified that the effect of bond line aging on fun1 is negligible, and fun2 can independently characterize the aging of bond lines. Therefore, the identification of aging modes and the characterization of aging levels are decoupled. Furthermore, based on the changes in the parameter values ​​of fun1 and fun2, multiple failure modes can be identified.

[0098] Step 4: Based on the fun obtained in Step 2 1-h fun 2-h fun 1-ag and fun 2-ag Calculate Δfun1 and Δfun2; based on the mapping relationship in step 3, obtain the heat dissipation path and bond line aging level from Δfun1 and Δfun2.

[0099] Depend on Figure 9 As shown, for modules used in actual field applications, they can be tested in advance, and the aging characteristic parameter values ​​fun under healthy conditions can be calculated based on the definitions of fun1 and fun2. 1-h fun 2-hThen, based on the electrical signals collected during field operations and the known thermal parameters of the module, the losses and junction temperature are calculated, and ΔT is extracted. high and ΔT low The aging characteristic parameter value fun of this module was calculated. 1-ag and fun 2-ag Calculate Δfun1 and Δfun2, and determine whether it is within the safe range Δfun. 1-th Δfun 2-th Finally, the aging of the module is characterized by the values ​​of Δfun1 and Δfun2.

Claims

1. A method for monitoring aging characteristic decoupling parameters of an IGBT module based on temperature characteristics, characterized in that, Comprising the following steps: Step 1: Calculate the cycle loss of IGBT, extract the power loss curve; establish the thermal network model of IGBT-heat sink system, calculate the instantaneous temperature difference ΔT of IGBT at the starting node of each loss cycle low (i) and the instantaneous temperature difference ΔT of the half loss cycle node high (i) , the steady-state temperature difference ΔT of each loss cycle starting node is obtained by iterative calculation low and the steady-state temperature difference ΔT of the half loss cycle node high ; combined with the heat dissipation path of IGBT module and the loss calculation formula, ΔT high is divided by ΔT low , the aging characteristic parameter fun1 is constructed, ΔT low is added to ΔT high , and the aging characteristic parameter fun2 is constructed; Step 2: combining IGBT module thermal resistance, saturation voltage drop failure standard, deriving aging degree factor fun1 and fun2 calculation formula, establishing the mapping relationship between aging characteristic parameters fun1 and fun2 and aging; Step 3: Fix the working condition, measure the junction temperature curve of the healthy module under the cycle loss, extract ΔT high With ΔT low , calculate the aging characteristic parameter value of the health state, respectively fun 1-h and fun 2-h ; Measure the junction temperature curve of any module with different aging degrees under the cycle loss, extract ΔT high With ΔT low , calculate the aging characteristic parameter value of the module fun 1-ag and fun 2-ag ; Step 4: Calculate Δfun1 and Δfun2 according to fun 1-h , fun 2-h , fun 1-ag and fun 2-ag obtained in step 3; obtain the heat dissipation path of the measured module and the bonding wire aging level according to the mapping relationship in step 2 and Δfun1 and Δfun2.

2. The method according to claim 1, characterized in that, The power loss in the module is half-sine, which is affected by the circuit control in step 1, and the average switching loss P of the module switch and the on-state loss P conduct The calculation formula is: where I c is the collector current, is the power factor angle, M is the modulation ratio, f sw is the switching frequency, V DC is the DC voltage, T j is the junction temperature; K V is the voltage correction coefficient, C T is the temperature correction coefficient, E on+off is the switching energy; I ref , V ref and T ref are the reference values of current, voltage and temperature respectively; R ce and V ce are the junction temperature T j corresponding to the collector-emitter resistance and the saturation voltage respectively; IGBT module total loss P loss is: Wherein, n is 1, 2, 3, …, t is time, ω is loss angle frequency; According to the principle of impulse equality, the power loss of half-sine waveform can be equivalent to square wave, and the calculation formula is as follows: Wherein, P is the amplitude of the equivalent square wave of the loss waveform; IGBT module and heat sink system are modeled by Foster thermal network, and according to the heat transfer theory, the thermal response of IGBT caused by power loss is expressed as: Wherein, ΔT initial is the initial temperature difference of IGBT, time constant τ th is the product of thermal resistance R th and heat capacity C th , P loss is the power loss of IGBT module, ΔT j (t) is the real-time temperature difference of IGBT; when the temperature difference of IGBT module does not reach steady state, the temperature difference between the starting node of loss period and the half loss period node in the temperature characteristics of IGBT changes with the iteration of loss period; the instantaneous temperature difference ΔT low (i) and ΔT high (i) , wherein i is the number of loss period; ΔT low (i) and ΔT high (i) The calculation formula is as follows: where f is the loss frequency; assuming that the node temperature difference reaches a steady state after the nth cycle, then according to the above node instantaneous temperature difference calculation formula, the iterative calculation can obtain AT low and AT high , expressed as follows: Node steady state temperature difference ΔT low And ΔT high Can be obtained by monitoring the IGBT temperature curve; the bonding wire aging increases the bonding resistance, and the solder layer aging causes the heat dissipation path to shrink, affecting R th And P loss Size; according to the above expression, P loss , R th And τ th All affect the size of ΔT low And ΔT high , unable to determine the failure mode by the change of ΔT low And ΔT high Therefore, the aging characteristic parameters fun1 and fun2 are constructed, and the expressions are as follows: From formula (7), it can be obtained that, for the aging characteristic parameter fun1, the ΔT at steady state is calculated low and ΔT high , both of which can exclude the interference of P and obtain a parameter only related to τ th and f; in the case of a certain f, fun1 can single-reflect the change of τ th ; and since the change of heat capacity during the aging process is very small, it can be ignored, so fun1 can single-reflect the change of R th ; and for the aging characteristic parameter fun2, according to the formula analysis, it is affected by P and R th ; in view of the fact that τ th can be determined in advance via fun1, R th is determined according to the time constant calculation formula, and P is single-determined according to the fun2 calculation formula by excluding the known quantity R th ; according to the aging mechanism, the bonding wire aging mainly affects the power loss, the solder layer aging shrinks the heat dissipation path, and increases R th , so according to the above evaluation strategy, fun1 and fun2 can be combined to characterize and evaluate the heat dissipation path and the bonding wire aging.

3. The method of claim 2, wherein the temperature-dependent IGBT module aging characteristic decoupling parameter monitoring method is characterized by, In step 2, based on the aging characteristic parameter funl, it is assumed that R th becomes k1 times the value of fun 1-h at the time of the healthy state, and the value of the aging characteristic parameter funl after aging becomes fun 1-ag , then there is: Therefore, we have: Wherein, Δfun1 is the aging factor of the aging characteristic parameter fun1; according to the above formula, fun 1-ag and fun 1-h and Δfun1, the aging degree of the heat dissipation path can be obtained; then, for the aging characteristic parameter fun2, since the change of thermal resistance caused by the aging of the heat dissipation path has been reflected and corrected in fun1, fun2 assumes R th as a constant, which does not represent the aging of the heat dissipation path, and the V ce rise caused by the aging of the bonding wire is reflected in P, according to the loss calculation formula and the impulse equal calculation, the square wave loss amplitude is obtained, and the aging characteristic parameter fun2 expression has: Wherein, a1, b1, b2 are values irrelevant to V ce , are constants, which can be calculated according to formula (1); assuming that the bonding wire is aged, V ce increases by k2 times, and the value of the aging characteristic parameter fun2 changes from fun 2-h in the healthy state to fun 2-ag in the aged state, it can be derived that: wherein, according to the expression of fun 2-ag Further simplification can be obtained, the relationship between the health status and the aging state is: Wherein, Δfun2 is the aging factor of the aging characteristic parameter fun2; and V ce is the value of the health state, is a constant, which can be obtained according to the module manual; a1 can be calculated according to the circuit working condition, R th is a constant; when k2 is 1, Δfun2 is 0, at this time the module is healthy, when k2 is 1.05, Δfun2 is the maximum value, at this time the module is aging failure; therefore, according to the difference between fun 2-ag and fun 2-h , the aging level of the module bonding wire can be represented.

Images