Direct current power recycling method based on junction temperature closed loop control

Abstract

The application belongs to the technical field of power device testing, and particularly relates to a direct current power cycle method based on junction temperature closed loop control. The method comprises the following steps: obtaining a calibration relationship between a small current source drain voltage drop and a junction temperature of a power module to be tested; during large current heating, determining an estimated junction temperature according to a real-time case temperature, a real-time heating power and an online effective thermal resistance, and shutting down the heating loop when the estimated junction temperature reaches a reference junction temperature boundary; determining a feedback junction temperature from the small current source drain voltage drop after the shutdown, and determining an original effective thermal resistance according to the feedback junction temperature, the case temperature at the shutdown moment and the thermal power before the shutdown; performing weighted recursion on the original effective thermal resistance and the online effective thermal resistance to obtain a next round online effective thermal resistance; and taking a target maximum junction temperature as a reference, determining a next round reference junction temperature boundary according to a feedback junction temperature deviation. The application can update the junction temperature estimation parameters and the heating shutdown boundary according to the cycle thermal state, and reduce the maximum junction temperature drift caused by the fixed thermal resistance or the fixed shutdown condition.

Description

Technical Field

[0001] This invention relates to the field of power device testing, and more specifically to a DC power cycling method based on junction temperature closed-loop control. Background Technology

[0002] DC power cycling is used to evaluate the reliability of the power device package interconnect structure and heat transfer path under cyclic thermal stress by repeatedly turning the power module under test on and off to generate heat and cool it. Existing DC power cycling typically uses fixed on-time, fixed cooling time, case temperature threshold or pre-set thermal resistance parameters to control the heating process, and determines the heating termination time based on the cycling current, on-state voltage drop or case temperature.

[0003] This type of processing method fails to consistently reflect the actual thermal state of the power module under test in different cycles. The real-time high-current source drain voltage drop of the power module under test changes with junction temperature and device state. Mounting contact state, heat dissipation boundary, and package degradation also change the heat transfer relationship between junction temperature and case temperature, causing the preset thermal resistance parameter to gradually deviate from the actual value. At the same time, there is a thermal response lag between case temperature and junction temperature. Controlling shutdown based solely on case temperature or fixed heating time cannot directly characterize the highest junction temperature of the current cycle. As a result, the same control parameter may produce different junction temperature rises in different cycles, causing the highest junction temperature to gradually deviate from the target value; if the deviation is too large, it may exceed the allowable junction temperature, while if the deviation is too small, the applied thermal stress will be insufficient. If the junction temperature feedback after shutdown is not used to correct the thermal resistance parameter and subsequent heating shutdown conditions, the thermal state changes formed in the previous cycle cannot be passed to the next cycle, making it difficult to maintain the consistency of the highest junction temperature across cycles. Therefore, it is necessary to solve the problem of how to dynamically determine the junction temperature and heating shutdown conditions based on the current thermal state when the on-state voltage drop, heat dissipation conditions, and heat transfer path change with the cycle. Summary of the Invention

[0004] This invention provides a DC power cycling method based on junction temperature closed-loop control, which is used to at least solve the problem of how to dynamically determine the junction temperature and update the heating shutdown conditions under the condition that the on-state voltage drop and heat transfer relationship of the power module under test change with the cycle, so as to control the highest junction temperature in each cycle.

[0005] This invention provides a DC power cycling method based on junction temperature closed-loop control, the method comprising: The calibration relationship between the low-current source leakage voltage drop and junction temperature of the power module under test is obtained. The power module under test is heated with a high current, and the estimated junction temperature is determined based on the real-time case temperature, real-time heating power, and online effective thermal resistance. The heating circuit is shut off when the estimated junction temperature reaches the reference junction temperature boundary. The feedback junction temperature is determined by the hot-state low-current source leakage voltage drop after shutdown and the calibration relationship. The original effective thermal resistance is determined by the feedback junction temperature, the case temperature at shutdown time, and the hot-state power before shutdown. The original effective thermal resistance and the online effective thermal resistance are weighted and recursively calculated to obtain the next round of online effective thermal resistance for the next round of junction temperature estimation. Using the target maximum junction temperature as a benchmark, the next round of reference junction temperature boundary for the next round of heating shutdown is determined based on the deviation between the feedback junction temperature and the target maximum junction temperature.

[0006] In one possible implementation, obtaining the calibration relationship includes: acquiring multiple calibration points under fixed low current and fixed gate conditions, each calibration point including a calibration temperature and a low current source drain voltage drop collected at the calibration temperature; and generating the calibration relationship based on piecewise linear interpolation of adjacent calibration points.

[0007] In one possible implementation, before applying high-current heating to the power module under test, the heating circuit is kept off, the cold-state low-current source leakage voltage drop is obtained, the minimum junction temperature is determined based on the cold-state low-current source leakage voltage drop and the calibration relationship, and the power module under test is applied to the power module under test with high current heating when the minimum junction temperature is lower than the preset minimum junction temperature target.

[0008] In one possible implementation, determining the real-time heating power includes: obtaining the real-time high-current source leakage voltage drop, and determining the real-time heating power by multiplying the cyclic high-current setpoint by the real-time high-current source leakage voltage drop; determining the estimated junction temperature includes: adding the product of the real-time heating power and the online effective thermal resistance to the real-time case temperature to obtain the estimated junction temperature.

[0009] In one possible implementation, during any sampling period, when the estimated junction temperature reaches the reference junction temperature boundary, the heating circuit is turned off, and the first sampled value collected after the heating circuit is turned off and located within the preset small current source drain voltage drop range is determined as the hot small current source drain voltage drop.

[0010] In one possible implementation, the case temperature at the shutdown moment is the real-time case temperature when the heating circuit is shut down, and the hot-state power before shutdown is the real-time heating power before the heating circuit is shut down; the determination of the original effective thermal resistance includes: the ratio of the difference between the feedback junction temperature and the case temperature at the shutdown moment to the hot-state power before shutdown is determined as the original effective thermal resistance.

[0011] In one possible implementation, the weighted recursion includes: using the difference obtained by subtracting the thermal resistance update coefficient as the weight of the online effective thermal resistance, using the thermal resistance update coefficient as the weight of the original effective thermal resistance, and performing a weighted summation of the online effective thermal resistance and the original effective thermal resistance to obtain a candidate online effective thermal resistance, which is used to determine the online effective thermal resistance in the next round.

[0012] In one possible implementation, starting from the first time the self-feedback junction temperature enters a preset convergence range centered on the target highest junction temperature, a range check is performed on the candidate online effective thermal resistance. When the candidate online effective thermal resistance is within the preset effective thermal resistance range, the candidate online effective thermal resistance is determined as the online effective thermal resistance for the next round. When the candidate online effective thermal resistance exceeds the preset effective thermal resistance range, the online effective thermal resistance is determined as the online effective thermal resistance for the next round. When the candidate online effective thermal resistances obtained in three consecutive cycles all exceed the preset effective thermal resistance range, the DC power cycling is stopped.

[0013] In one possible implementation, determining the next round reference junction temperature boundary includes: determining the difference between the target maximum junction temperature and the feedback junction temperature as the control deviation, calculating the product of the reference junction temperature correction coefficient and the control deviation, and determining the sum of the product and the target maximum junction temperature as the next round reference junction temperature boundary, and limiting the next round reference junction temperature boundary between the preset lower limit of the reference junction temperature and the preset upper limit of the reference junction temperature.

[0014] In one possible implementation, the number of power modules under test is two or three, and each power module under test adopts an independent calibration relationship, target maximum junction temperature, online effective thermal resistance, and reference junction temperature boundary. The power modules under test are heated with high current in a preset cycle sequence, and the online effective thermal resistance and reference junction temperature boundary of each power module under test are updated respectively to obtain the next round of online effective thermal resistance and the next round of reference junction temperature boundary of each power module under test.

[0015] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: By establishing a calibration relationship between the leakage voltage drop of the small current source of the measured power module and its junction temperature, the leakage voltage drop of the hot small current source obtained after shutdown can be converted into a feedback junction temperature, giving the junction temperature feedback a clear source of electrical parameters. The estimated junction temperature is determined jointly by real-time case temperature, real-time heating power, and online effective thermal resistance, with the shutdown condition being that the estimated junction temperature reaches the reference junction temperature boundary. This allows the heating control to simultaneously reflect the case temperature, actual power input, and current heat transfer state. The original effective thermal resistance is determined by the feedback junction temperature, the case temperature at shutdown time, and the hot-state power before shutdown, quantifying the heat transfer relationship of this round. By weighted recursion of the original effective thermal resistance and the online effective thermal resistance, the thermal state of this round can be introduced into the next round of junction temperature estimation, limiting parameter mutations caused by single-round measurement fluctuations. By correcting the reference junction temperature boundary of the next round based on the target highest junction temperature, the heating shutdown condition is adjusted round by round according to the feedback junction temperature deviation. Overall, the junction temperature estimation parameters and heating shutdown boundary are updated by loop feedback, which can reduce the drift of the maximum junction temperature caused by fixed thermal resistance and fixed shutdown conditions, and keep the thermal stress control of each DC power cycle continuous. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is an experimental graph showing the sensitivity of the feedback junction temperature parameter of CAB016M12FM3 under the conditions of a target maximum junction temperature of 100°C and an initial online effective thermal resistance of 1K / W in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application.

[0018] In this embodiment, the junction temperature closed-loop control takes the power module under test as the processing object, and uses the calibration relationship between the drain voltage drop of the small current source and the junction temperature, the real-time case temperature, the real-time heating power, the online effective thermal resistance, the reference junction temperature boundary, and the target maximum junction temperature as control inputs. During high-current heating, the DC power loop platform continuously acquires the real-time case temperature and real-time heating power, and determines the estimated junction temperature by combining it with the online effective thermal resistance. The relationship between the estimated junction temperature and the reference junction temperature boundary is used to control the heating loop shutdown. After the heating loop is shut down, the DC power loop platform acquires the hot-state drain voltage drop of the small current source, determines the feedback junction temperature through the calibration relationship, and then determines the original effective thermal resistance using the feedback junction temperature, the case temperature at the shutdown time, and the hot-state power before shutdown. The original effective thermal resistance is used to update the online effective thermal resistance for the next round, and the deviation between the feedback junction temperature and the target maximum junction temperature is used to determine the reference junction temperature boundary for the next round. This forms the processing basis for passing the junction temperature estimation parameters and heating shutdown conditions from the current loop to the next loop.

[0019] like Figure 1 As shown, a DC power cycling method based on junction temperature closed-loop control includes the following processing steps.

[0020] Before the DC power cycling begins, the DC power cycling platform establishes a calibration relationship between the drain voltage drop of the low-current source and the junction temperature for the power module under test. During the calibration phase, the low-current and gate conditions are kept constant, and the drain voltage drop of the low-current source is collected at multiple calibration temperatures. Each calibration temperature and its corresponding drain voltage drop of the low-current source are used to form a calibration point.

[0021] The DC power cycling platform sorts multiple calibration points according to the calibration temperature and uses piecewise linear interpolation between adjacent calibration points to generate calibration relationships. During the operation phase, when collecting the leakage voltage drop of a cold-state small current source or a hot-state small current source, the DC power cycling platform calls the calibration relationship to determine the junction temperature, so that the calibration results enter the minimum junction temperature determination, feedback junction temperature determination, and subsequent parameter update stages.

[0022] In one embodiment, the DC power cycling platform performs calibration for each power module under test (DUT). The DUT is connected to a low-current temperature sensing loop, with the low current maintained at approximately 100 mA, and the fixed gate condition remains constant across all calibration temperatures. The low current setting should ensure that the self-heating generated by the DUT is insufficient to alter the current calibration temperature, while keeping the low-current source drain voltage drop within the voltage acquisition range. The fixed gate condition is used to eliminate the influence of gate state changes on the low-current source drain voltage drop, ensuring that data obtained at different calibration temperatures have the same measurement basis.

[0023] The power module under test is placed in a temperature-adjustable calibration environment. The calibration temperatures are set in ascending or descending order, covering the junction temperature range used in DC power cycling. When the power module under test reaches and stabilizes at the current calibration temperature, the DC power cycling platform collects the leakage voltage drop of the small current source, and combines the current calibration temperature and the current leakage voltage drop of the small current source to form a calibration point.

[0024] After completing data acquisition at one calibration temperature, switch to the next calibration temperature and repeat the acquisition until multiple calibration points are formed. If acquisition fails or the leakage voltage drop of the small current source exceeds the voltage acquisition range, the current data is not written to the calibration point, and the fixed small current and fixed gate conditions are maintained for re-acquisition.

[0025] Multiple calibration points are sorted according to their calibration temperatures and stored independently for each power module under test. A piecewise linear interpolation interval is defined between any two adjacent calibration points. During operation, when the drain voltage drop of the small current source is obtained, the DC power cycling platform locates the piecewise linear interpolation interval containing the drain voltage drop and determines the current junction temperature using the calibration temperatures at both ends of the interval and the drain voltage drop of the small current source. The conversion relationship between the drain voltage drop of the small current source and the junction temperature is expressed as:

[0026] in, This refers to the junction temperature, expressed in degrees Celsius. This is the leakage voltage drop of a small current source, in volts. This is a calibration relationship between the drain voltage drop of a small current source and the junction temperature, determined by multiple calibration points and piecewise linear interpolation. This calculated relationship is used to convert the acquired drain voltage drop of the small current source into the junction temperature.

[0027] When the leakage voltage drop of a small current source lies between two adjacent calibration points, piecewise linear interpolation only uses the calibration points at both ends of the current interval. When the leakage voltage drop of the small current source coincides with a calibration point, the calibration temperature at that calibration point is directly used. Different power modules under test use their own calibration relationships; the calibration point of one power module under test is not used for another.

[0028] After the calibration relationship is generated, the junction temperature determination stage of the DC power cycle begins. The drain voltage drop of the cold-state small current source is converted into the minimum junction temperature through the calibration relationship. The minimum junction temperature is used to determine whether to enter high-current heating. The drain voltage drop of the hot-state small current source is converted into the feedback junction temperature through the same calibration relationship. The feedback junction temperature is used to determine the original effective thermal resistance and the reference junction temperature boundary for the next cycle. Thus, the calibration relationship forms a continuous data transfer relationship between cooling release, feedback junction temperature determination, and subsequent parameter updates.

[0029] Before the start of the k-th round of high-current heating, the DC power cycling platform reads the online effective thermal resistance and reference junction temperature boundary of the power module under test, and obtains the real-time case temperature and real-time high-current source drain voltage drop according to the sampling period after the heating circuit is turned on. The real-time high-current source drain voltage drop and the cycle high-current setpoint together form the real-time heating power, and the real-time case temperature, real-time heating power and online effective thermal resistance together form the estimated junction temperature.

[0030] The estimated junction temperature is compared with the reference junction temperature boundary in each sampling period. High-current heating is maintained until the estimated junction temperature reaches the reference junction temperature boundary; then the heating circuit is immediately shut off. After the heating circuit is shut off, it switches to low-current temperature measurement mode, forming a hot-state low-current source drain voltage drop, which is then transmitted to the feedback junction temperature determination stage.

[0031] In one embodiment, before the start of the k-th round of high-current heating, the DC power cycling platform keeps the heating circuit off and puts the power module under test in a low-current temperature measurement state. The low-current temperature measurement state uses the same low-current and fixed gate conditions as the calibration relationship establishment phase, so that the cold-state low-current source drain voltage drop can be directly entered into the calibration relationship for junction temperature conversion.

[0032] After the voltage acquisition channel obtains the cold-state low-current source drain voltage drop of the power module under test, the DC power cycling platform calls the calibration relationship bound to the power module under test to determine the minimum junction temperature before the start of the k-th cycle. The minimum junction temperature is indicated by the symbol... The unit is Celsius. The minimum junction temperature is not a real-time control quantity during heating, but rather a measure used to determine whether the power module under test has cooled down to a temperature that allows it to enter the next round of high-current heating.

[0033] The preset minimum junction temperature target is written before the start of the DC power cycle. The preset minimum junction temperature target is determined based on the minimum junction temperature required by the test, the allowable operating temperature range of the power module under test, and the junction temperature range covered by the calibration relationship. It is used to constrain the start conditions of each high-current heating cycle. The same set of DC power cycles uses a fixed preset minimum junction temperature target; when it is necessary to change the test temperature boundary, it is reconfigured before the start of a new set of DC power cycles.

[0034] When multiple power modules are under test, each power module under test saves its preset minimum junction temperature target. The current power module under test only calls the preset minimum junction temperature target and calibration relationship corresponding to itself.

[0035] When the minimum junction temperature is lower than the preset minimum junction temperature target, the DC power cycling platform confirms that the cooling release condition is met, turns on the heating circuit of the currently tested power module, and starts high-current heating. When the minimum junction temperature is not lower than the preset minimum junction temperature target, the heating circuit remains off, the DC power cycling platform continues cooling, and re-acquires the cold-state low-current source drain voltage drop and minimum junction temperature in subsequent sampling cycles.

[0036] The judgment uses the lowest junction temperature formed in the current sampling period, does not accumulate the lowest junction temperatures of multiple sampling periods, and does not replace the judgment of the lowest junction temperature with a fixed cooling time.

[0037] When the cold-state small current source leakage voltage drop acquisition fails, exceeds the preset small current source leakage voltage drop range, or cannot form the minimum junction temperature through calibration relationship, the DC power cycling platform shuts down all outputs and stops the current DC power cycle, and abnormal data is not included in the minimum junction temperature judgment.

[0038] The high-current heating release result generated after the minimum junction temperature meets the preset minimum junction temperature target is transmitted to the heating stage to control the heating circuit to switch from the off state to the on state.

[0039] In one embodiment, after high-current heating begins, the DC power cycling platform acquires the real-time high-current source drain voltage drop and real-time case temperature according to a set sampling period. The cycling high-current setpoint is configured before the start of the current DC power cycle and remains unchanged during the current heating phase.

[0040] The high current setting value can be determined based on the rated current of the power module under test, the test conditions of DC power cycling, and the output capability of the DC power cycling platform. It is used to limit the current input during high current heating and participate in the real-time heating power calculation.

[0041] The real-time heating power is not calculated using the real-time current acquisition, but rather determined by the product of the cyclic high current setpoint and the real-time high current source leakage voltage drop. The real-time heating power at time t in the k-th cycle is calculated using the following formula:

[0042] In the formula, The real-time heating power at time t in the k-th round is expressed in watts. This is the setting value for high-current circulation, in amperes. Let be the real-time high-current source drain voltage drop at time t in the k-th round, in volts. This calculation incorporates the change in the current on-state voltage drop of the power module under test into the real-time heating power, so that the real-time heating power is updated according to the on-state of the power module under test, and serves as the input for estimating the junction temperature rise.

[0043] After the heating circuit is turned on, the DC power cycling platform acquires the real-time high-current source leakage voltage drop according to the set sampling period. The preset high-current source leakage voltage drop range is determined based on the on-state voltage drop range of the power module under test and the input range of the voltage acquisition channel under the set high-current cycling value, and is used to determine whether the real-time high-current source leakage voltage drop exceeds the limit. When the real-time high-current source leakage voltage drop exceeds the preset high-current source leakage voltage drop range or sampling fails, the abnormal value is not used to determine the real-time heating power, the DC power cycling platform shuts down all outputs and stops the current DC power cycle.

[0044] The real-time shell temperature is obtained from the shell temperature acquisition channel within the same sampling period. The online effective thermal resistance is the online effective thermal resistance formed after the previous round; the online effective thermal resistance used in the k-th heating round is indicated by the symbol... This indicates that, before the initial cycle establishes the effective online thermal resistance of the previous cycle, the initial effective online thermal resistance configured before the start of the DC power cycle is used. The initial effective online thermal resistance is determined based on the stable effective online thermal resistance formed by a DC power cycle performed on a similar power module under the same installation conditions and heat dissipation boundaries, and is used to establish an estimated junction temperature in the first cycle.

[0045] The estimated junction temperature at time t in the k-th cycle of the DC power cycling platform is determined by the following formula:

[0046] In the formula, The estimated junction temperature at time t in the k-th round is given in degrees Celsius. The real-time case temperature at time t in the kth round is in degrees Celsius. The case temperature protection threshold is determined based on the allowable case temperature of the power module under test and the test safety requirements. It is used to determine whether the real-time case temperature and the case temperature at the turn-off time exceed the allowable operating range. When the real-time case temperature exceeds the case temperature protection threshold, the DC power cycling platform shuts down all outputs and stops the current DC power cycling. The data of the current sampling period will not be included in the estimated junction temperature determination and subsequent parameter update stages. This refers to the real-time heating power. This represents the effective online thermal resistance formed after the end of the previous cycle, expressed in Kelvin per watt.

[0047] The product of real-time heating power and online effective thermal resistance characterizes the temperature rise of the junction region of the measured power module relative to the casing, with the real-time casing temperature serving as the temperature reference. The estimated junction temperature generated in each sampling cycle is immediately entered into the reference junction temperature boundary comparison, without accumulation across sampling cycles.

[0048] In one embodiment, after the DC power cycling platform generates an estimated junction temperature in each sampling cycle, it compares the estimated junction temperature with the reference junction temperature boundary used in the current k-th round. The reference junction temperature boundary in the k-th round is marked with a symbol... It indicates that the unit is Celsius.

[0049] The initial cycle uses the reference junction temperature boundary configured before the start of the DC power cycle, and subsequent cycles use the reference junction temperature boundary formed in the previous cycle. The reference junction temperature boundary is used to define the high-current heating termination condition of the current cycle and is not taken as the actual highest junction temperature of the power module under test.

[0050] The turn-off condition between the estimated junction temperature and the reference junction temperature boundary is determined by the following formula:

[0051] In the formula, The estimated junction temperature at time t in the k-th round. This is the reference junction temperature boundary for the k-th round.

[0052] When the estimated junction temperature is lower than the reference junction temperature boundary, the heating circuit remains connected, the DC power cycling platform enters the next sampling cycle, reacquires the real-time high current source drain voltage drop and real-time case temperature, and updates the real-time heating power and estimated junction temperature.

[0053] If the estimated junction temperature in any sampling period is equal to or higher than the reference junction temperature boundary, the current comparison result immediately triggers the heating circuit shutdown. The shutdown decision is made on a single basis; it is not required that the shutdown condition be met in multiple consecutive sampling periods, nor is it averaged among multiple estimated junction temperatures.

[0054] After the heating circuit shutdown command is issued, the high-current heating of the currently measured power module stops, and the DC power circulation platform switches the voltage acquisition state to the low-current temperature measurement state. The program does not include a separate delay between the shutdown command, the heating circuit disconnection, and the low-current temperature measurement.

[0055] The program state writing, analog voltage acquisition task release and reconstruction, and acquisition card sampling result in a physical delay of hundreds of microseconds. The DC power cyclic platform does not set a fixed sampling window, does not read the window peak or window mean, and does not perform explicit delay compensation.

[0056] Once the analog voltage acquisition task enters the acquisition state, the first sampled value acquired after the heating circuit is turned off and located within the preset small current source leakage voltage drop range will be determined as the hot small current source leakage voltage drop. Sampled values ​​that do not form an effective voltage or exceed the preset small current source leakage voltage drop range during task switching will not participate in the feedback junction temperature determination.

[0057] The preset small current source leakage voltage drop range is determined based on the voltage range covered by the calibration relationship and the input range of the voltage acquisition channel. It is used to determine whether the cold-state and hot-state small current source leakage voltage drops can enter the junction temperature determination stage.

[0058] Since the physical delay from the high current turn-off to the acquisition of the hot-state small current source drain voltage drop is on the order of hundreds of microseconds, the feedback junction temperature determined by the hot-state small current source drain voltage drop combined with the calibration relationship is used to approximately characterize the highest junction temperature in this round, and to determine the original effective thermal resistance and the reference junction temperature boundary for the next round.

[0059] When the analog voltage acquisition task reconstruction fails, the hot-state small current source leakage voltage drop acquisition fails, or the hot-state small current source leakage voltage drop exceeds the preset small current source leakage voltage drop range, the DC power cycling platform shuts down all outputs and stops the current DC power cycling. Abnormal sampling values ​​do not enter the feedback junction temperature determination stage.

[0060] The leakage voltage drop of the normally formed hot small current source is transferred to the feedback junction temperature determination stage and serves as the data basis for the original effective thermal resistance and the next round of reference junction temperature boundary calculation.

[0061] After the heating circuit is turned off, the DC power cycling platform keeps the current power module under test in a low-current temperature measurement state, receives the leakage voltage drop of the hot low-current source after the shutdown, and calls the calibration relationship between the leakage voltage drop of the low-current source bound to the current power module under test and the junction temperature to determine the feedback junction temperature of this round.

[0062] The DC power circulation platform synchronously retains the real-time case temperature when the heating circuit is turned off and the real-time heating power before turn-off, respectively forming the case temperature at the time of turn-off and the hot-state power before turn-off. The difference between the feedback junction temperature and the case temperature at the time of turn-off reflects the junction-case temperature rise near the turn-off point in this cycle. The DC power circulation platform determines the ratio of this difference to the hot-state power before turn-off as the original effective thermal resistance.

[0063] The original effective thermal resistance is not involved in the heating shutdown that has already been completed in this round, but is transferred to the online effective thermal resistance update stage, serving as the basis for updating the junction temperature estimation parameters in the next round.

[0064] In one embodiment, when the estimated junction temperature in the k-th round reaches the reference junction temperature boundary, the DC power cycling platform issues a heating circuit shutdown command and retains the real-time high current source drain voltage drop and real-time heating power obtained before the shutdown command takes effect.

[0065] After switching from voltage acquisition to low-current temperature measurement, the first sampled value acquired after shutdown and falling within the preset low-current source drain voltage drop range is determined as the k-th round hot-state low-current source drain voltage drop. The DC power cycling platform inputs the hot-state low-current source drain voltage drop into the calibration relationship between the low-current source drain voltage drop and junction temperature, and the feedback junction temperature is determined according to the following formula:

[0066] In the formula, The k-th feedback junction temperature is expressed in degrees Celsius. The leakage voltage drop of the hot-state small current source is collected after the heating circuit is turned off in the kth round, in volts. This establishes the calibration relationship between the leakage voltage drop and junction temperature of a small current source. The feedback junction temperature is used to characterize the junction temperature state near the heating circuit turn-off and is incorporated into the initial effective thermal resistance determination process.

[0067] The shell temperature at the time of shutdown is represented by a symbol. This indicates that the DC power cycling platform saves the real-time case temperature within the corresponding sampling period when the heating circuit is turned off as the case temperature at the time of shutdown, so that the feedback junction temperature and the case temperature at the time of shutdown use the same cycle of thermal state data.

[0068] Hot power before shutdown is represented by the symbol This is indicated and determined by the circulating high current setpoint and the real-time high current source leakage voltage drop collected before the heating circuit is turned off:

[0069] In the formula, The hot power before the k-th round of shutdown is expressed in watts. This is the setting value for high-current circulation, in amperes. This represents the real-time leakage voltage drop of the high-current source, measured in volts, collected before the heating circuit is shut down in the kth round. The hot-state power before shutdown reflects the actual power input of the measured power module before shutdown; the power value after the heating circuit is shut down is not used in the calculation of the original effective thermal resistance.

[0070] The DC power cycling platform uses the feedback junction temperature, the case temperature at turn-off, and the hot-state power before turn-off to determine the initial effective thermal resistance of the k-th cycle. The calculation relationship is as follows:

[0071] In the formula, The original effective thermal resistance in the kth round is expressed in Kelvin per watt. This is the junction temperature for the k-th round of feedback; The shell temperature at the k-th turn-off time; This represents the hot power before the k-th round of shutdown.

[0072] The difference between the feedback junction temperature and the case temperature at the time of shutdown represents the junction-case temperature rise near the shutdown point. The ratio of the junction-case temperature rise to the hot power before shutdown is used to characterize the effective heat transfer relationship formed under the combined effect of the current installation state, heat dissipation boundary, and heat transfer path of the measured power module.

[0073] If the leakage voltage drop of the hot-state small current source exceeds the preset small current source leakage voltage drop range, the case temperature exceeds the case temperature protection threshold at the time of shutdown, the real-time leakage voltage drop of the large current source collected before shutdown exceeds the preset large current source leakage voltage drop range, or the hot-state power before shutdown cannot be determined or is zero, the DC power cycle platform shuts down all outputs and stops the current DC power cycle, and the data of this round does not enter the online effective thermal resistance update stage.

[0074] When all inputs are valid, the original effective thermal resistance is saved in association with the current round number and passed to the weighted recursion stage to determine the online effective thermal resistance for the next round. The original effective thermal resistance is only used as an update input after the current round ends and does not replace the online effective thermal resistance already used in the current heating stage, in order to maintain the round correspondence between the calculated parameters and the sampled data.

[0075] After the k-th heating cycle ends and the initial effective thermal resistance is formed, the DC power cycle platform reads the online effective thermal resistance used in the k-th heating cycle, uses the initial effective thermal resistance as a new input reflecting the heat transfer state of this cycle, uses the online effective thermal resistance as a historical input formed by the previous cycle, and performs weighted recursion according to the thermal resistance update coefficient.

[0076] A weighted recursive method is used to generate candidate online effective thermal resistances. The DC power cycling platform determines the next round of online effective thermal resistances based on whether range verification is enabled and the relationship between the candidate online effective thermal resistances and the preset effective thermal resistance range. The next round of online effective thermal resistances is associated with and saved to the currently measured power module, and participates in the junction temperature estimation calculation during the (k+1)th round of high-current heating. This ensures that the original effective thermal resistance formed by the feedback junction temperature, the case temperature at the turn-off time, and the hot-state power before turn-off is included in the next round of junction temperature estimation.

[0077] In one embodiment, after the initial effective thermal resistance of the k-th round is formed, the DC power cycling platform reads the online effective thermal resistance of the power module under test. The online effective thermal resistance used for high-current heating in the k-th round is the online effective thermal resistance currently stored at the beginning of the k-th round. This indicates that if the candidate online effective thermal resistance fails the range check in the previous round, the online effective thermal resistance used will still be the online effective thermal resistance currently saved at the start of round k; the original effective thermal resistance formed in round k will be adopted. This indicates that the two thermal resistances use the same unit and belong to the same power module under test; the original effective thermal resistance or online effective thermal resistance is not exchanged between different power modules under test.

[0078] Weighted recursion is performed according to the following formula:

[0079] in, The candidate online effective thermal resistance is formed by the kth round of recursion, in Kelvin per watt; The online effective thermal resistance used for heating in the kth round, in Kelvin per watt; The original effective thermal resistance in the kth round is expressed in Kelvin per watt. The thermal resistance update coefficient is dimensionless. As the weight of the effective thermal resistance in the online environment, This represents the weight of the original effective thermal resistance.

[0080] The weighted recursive relation uses the online effective thermal resistance to retain the heat transfer state of the previous cycle, and uses the original effective thermal resistance to introduce the heat transfer change reflected by the feedback junction temperature in the kth round.

[0081] The thermal resistance update factor is configured before the start of the DC power cycle and saved separately for each power module under test. When the thermal resistance update factor is small, the online effective thermal resistance changes slowly between adjacent cycles, and the measurement fluctuations in the original effective thermal resistance of a single cycle have little impact on the junction temperature estimation of the next cycle. When the thermal resistance update factor is large, the proportion of the original effective thermal resistance in the k-th cycle is higher, and the candidate online effective thermal resistance can follow the changes in installation status, heat dissipation boundary, or heat transfer path more quickly. The setting of the thermal resistance update factor needs to be a trade-off between the update speed and the fluctuations between cycles.

[0082] In a set of parameter sensitivity tests, thermal resistance update coefficients of 0.15, 0.25, 0.35, and 0.5 were compared. When the thermal resistance update coefficient was 0.15, the feedback junction temperature entered the preset convergence range more slowly; when the thermal resistance update coefficient was 0.5, the speed of entering the preset convergence range was faster, but the error and fluctuation in the stable cycle increased. For engineering configuration, 0.2 can be used as the initial value of the thermal resistance update coefficient, and it can be adjusted according to the feedback junction temperature change and online effective thermal resistance fluctuation of the power module under test in a pre-performed DC power cycle.

[0083] After completing the weighted summation, The candidate online effective thermal resistance is passed to the range verification stage. If range verification has not yet been enabled or the candidate online effective thermal resistance passes the range verification, the candidate online effective thermal resistance is determined as the online effective thermal resistance for the next round, and is used in the (k+1)th round. Participate in the calculation of estimated junction temperature.

[0084] In one embodiment, the DC power cycling platform uses whether the feedback junction temperature enters a preset convergence range for the first time as the activation condition for candidate online effective thermal resistance range verification. The preset convergence range is set with the target maximum junction temperature as the center, indicating that the feedback junction temperature has entered a stable control phase near the target maximum junction temperature. The preset convergence range can be set according to the maximum allowable junction temperature deviation of DC power cycling; in the current embodiment, the preset convergence range is a temperature range formed by being 2 degrees Celsius above and below the target maximum junction temperature.

[0085] Before the feedback junction temperature first enters the preset convergence range, the online effective thermal resistance is in the initial adjustment phase. The DC power cycling platform does not perform range verification on candidate online effective thermal resistances, but instead uses the weighted recursive candidate online effective thermal resistances as the next round's online effective thermal resistances. When the feedback junction temperature first enters the preset convergence range, the DC power cycling platform begins range verification, and continues to perform range verification in subsequent cycles after candidate online effective thermal resistances are formed, starting from the cycle in which the feedback junction temperature first enters the preset convergence range.

[0086] The preset effective thermal resistance range is configured by the user before the DC power cycle begins. The preset effective thermal resistance range can be determined based on the results of a pre-performed DC power cycle of the same type of power module under the same installation conditions and heat dissipation boundaries. After the pre-performed DC power cycle enters a stable cycle, the online effective thermal resistance formed by multiple cycles is recorded, and the lower and upper boundaries of the preset effective thermal resistance range are set based on the stable online effective thermal resistance.

[0087] The preset effective thermal resistance range is used to limit the thermal resistance value that can be included in the next round of junction temperature estimation, so as to avoid sudden changes in the estimated junction temperature in the next round due to occasional feedback junction temperature, case temperature at the time of shutdown, or abnormal hot power before shutdown.

[0088] When a candidate online effective thermal resistance (TDR) is within the preset effective TDR range, the DC power cycling platform determines the candidate TDR as the next round's online effective TDR and associates it with the current power module under test and the (k+1)th round. When a candidate TDR is below the lower boundary or above the upper boundary, the candidate TDR is not written into the next round's control parameters. The DC power cycling platform retains the online effective TDR used for heating in the kth round and determines the retained online effective TDR as the currently saved online effective TDR at the start of the (k+1)th round. Candidate online effective TDRs exceeding the preset effective TDR range are not included in the junction temperature estimation calculation for the (k+1)th round.

[0089] The DC power cycling platform records the number of cycles that continuously exceed the preset effective thermal resistance range. When a candidate online effective thermal resistance exceeds the preset effective thermal resistance range in one cycle, the number of cycles continuously exceeding the preset effective thermal resistance range is incremented by one; when the candidate online effective thermal resistance is within the preset effective thermal resistance range, the number of cycles is reset to zero; when the candidate online effective thermal resistances formed in three consecutive cycles all exceed the preset effective thermal resistance range, the DC power cycling platform determines that the online effective thermal resistance update process has a continuous abnormality, stops the DC power cycling, and keeps the heating circuit off.

[0090] The online effective thermal resistance determined before the shutdown will not be used for new high-current heating. If the candidate online effective thermal resistance does not exceed the preset effective thermal resistance range for three consecutive cycles, the DC power cycle will continue to run, and the online effective thermal resistance for the next cycle will be determined according to the range verification results.

[0091] After the high-current heating of the kth round ends and the feedback junction temperature is formed, the DC power cycling platform reads the target maximum junction temperature of the current power module under test, compares the feedback junction temperature with the target maximum junction temperature, and forms a control deviation to characterize the direction and magnitude of the temperature deviation in the current round.

[0092] The control deviation and the reference junction temperature correction factor jointly participate in determining the next round of reference junction temperature boundary, ensuring that the next round of reference junction temperature boundary is always calculated around the target maximum junction temperature, rather than continuously accumulating on the current reference junction temperature boundary. The formed next round of reference junction temperature boundary is associated with and stored in the current power module under test, and compared with the estimated junction temperature during the next round of high-current heating, so as to trigger the heating circuit shutdown when the estimated junction temperature reaches the next round of reference junction temperature boundary.

[0093] In one embodiment, after the k-th round of feedback junction temperature is established, the DC power cycling platform reads the target maximum junction temperature. The target maximum junction temperature is determined by the test task, the allowable junction temperature of the power module under test, and the coverage range of the calibration relationship. The control deviation between the k-th round of feedback junction temperature and the target maximum junction temperature is determined according to the following formula:

[0094] In the formula, The control deviation for the kth round is expressed in degrees Celsius. The target maximum junction temperature, in degrees Celsius; The feedback junction temperature is denoted as k-th round, in degrees Celsius. When the control deviation is positive, the feedback junction temperature is lower than the target maximum junction temperature; when the control deviation is negative, the feedback junction temperature is higher than the target maximum junction temperature.

[0095] The next reference junction temperature boundary is determined according to the following formula:

[0096] In the formula, This is the reference junction temperature boundary for the (k+1)th round, in degrees Celsius. This is a dimensionless reference junction temperature correction factor. The reference junction temperature correction factor is used to adjust the degree of influence of control deviation on the reference junction temperature boundary in the next round.

[0097] The reference junction temperature boundary for the next round is determined using a non-cumulative method. Each round uses the target maximum junction temperature as the calculation benchmark, and control deviations are not accumulated on the reference junction temperature boundary. When the feedback junction temperature is lower than the target maximum junction temperature, the reference junction temperature boundary for the next round increases; when the feedback junction temperature is higher than the target maximum junction temperature, the reference junction temperature boundary for the next round decreases. The resulting reference junction temperature boundary for the next round is used for comparison between the estimated junction temperature and the heating loop shutdown in the next round.

[0098] like Figure 2 As shown, in the parameter sensitivity test of the CAB016M12FM3 power module under test, the target maximum junction temperature was set to 100 degrees Celsius, the cyclic high current setting was set to 50 amps, and the initial online effective thermal resistance was set to 1 Kelvin per watt. Figure 2 The fixed thermal resistance update factor Ka in part a is 0.25. The feedback junction temperature changes are examined when the reference junction temperature correction factor Kp is 0, 0.1, 0.2, 0.3 and 0.4 respectively. Figure 2 In part b, the fixed reference junction temperature correction factor Kp is 0.2, and the feedback junction temperature changes are examined when the thermal resistance update factor Ka is 0.15, 0.25, 0.35 and 0.5 respectively.

[0099] Under each parameter combination, the feedback junction temperature gradually approaches the target maximum junction temperature from the initial temperature. When the thermal resistance update coefficient increases, the number of cycles in which the feedback junction temperature enters the preset convergence range decreases, but the inter-cycle fluctuations in the steady-state phase increase. When the reference junction temperature correction coefficient changes, the correction magnitude of the control deviation by the reference junction temperature boundary in the next cycle changes accordingly, causing the feedback junction temperature to exhibit different convergence speeds, steady-state deviations, and steady-state maximum overshoots.

[0100] To Figure 2 The steady-state control results of each parameter combination were quantitatively compared. The steady-state mean absolute error, steady-state mean error, steady-state standard deviation, steady-state maximum overshoot, and the number of cycles that first entered the preset convergence range were statistically analyzed. The statistical results are shown in Table 1.

[0101] In this embodiment, the last 20 cycles of each parameter combination are defined as the steady-state statistical phase. The steady-state mean absolute error is the average of the absolute values ​​of the difference between the feedback junction temperature and the target maximum junction temperature during the steady-state statistical phase. The steady-state mean error is the average of the difference between the feedback junction temperature and the target maximum junction temperature. The steady-state standard deviation is the standard deviation of the feedback junction temperature during the steady-state statistical phase. The steady-state maximum overshoot is the maximum difference between the feedback junction temperature and the target maximum junction temperature. The cycle cycle in which the feedback junction temperature first enters the preset convergence range is the cycle cycle in which the feedback junction temperature first enters the range of 2 degrees Celsius above and below the target maximum junction temperature.

[0102] Table 1 Steady-state statistics of parameter sensitivity test for CAB016M12FM3

[0103] As shown in Table 1, when the thermal resistance update coefficient Ka is fixed at 0.25, the steady-state mean absolute errors corresponding to the reference junction temperature correction coefficients Kp of 0.3 and 0.4 are 1.020 degrees Celsius and 1.021 degrees Celsius, respectively. Among them, when the reference junction temperature correction coefficient Kp is 0.4, it enters the preset convergence range for the first time in the 9th round, and the steady-state standard deviation when the reference junction temperature correction coefficient Kp is 0.3 is 1.214 degrees Celsius.

[0104] When the reference junction temperature correction factor Kp is fixed at 0.2, the thermal resistance update factor Ka increases from 0.15 to 0.5. The number of cycles required for the first entry into the preset convergence range is advanced from the 20th cycle to the 6th cycle. The steady-state mean absolute error increases from 1.277 degrees Celsius to 1.391 degrees Celsius, and the steady-state standard deviation increases from 1.567 degrees Celsius to 1.818 degrees Celsius. Increasing the thermal resistance update factor can shorten the number of cycles required for the feedback junction temperature to enter the preset convergence range, but the fluctuation of the feedback junction temperature in the steady-state phase increases accordingly.

[0105] Therefore, the configuration of the reference junction temperature correction factor and the thermal resistance update factor needs to be determined in conjunction with the allowable number of convergence cycles, steady-state deviation, inter-cycle fluctuations, and steady-state maximum overshoot. The reference junction temperature correction factor can be configured in the range of 0.2 to 0.4, and the thermal resistance update factor can be selected based on the feedback junction temperature change and online effective thermal resistance fluctuation of the power module under test.

[0106] The calculation results are compared with the preset lower and upper limits of the reference junction temperature. These two boundaries are jointly determined by the allowable junction temperature of the power module under test, the target maximum junction temperature, the calibration coverage area, heat dissipation capacity, and protection conditions. If the calculated result is lower than the preset lower limit, the preset lower limit is used; if it is higher than the preset upper limit, the preset upper limit is used; and if it falls between the two, the calculation result is retained. The reference junction temperature boundary for the next round after the limits are met is directly used for the next round of high-current heating without changing the reference junction temperature correction factor.

[0107] In one embodiment, the DC power cycling platform connects two or three power modules under test (DUTs), and configures calibration relationships, target maximum junction temperature, online effective thermal resistance, and reference junction temperature boundaries for each DUT. Each parameter is saved separately for each DUT. The feedback junction temperature, next-cycle online effective thermal resistance, and next-cycle reference junction temperature boundary formed for any DUT in the current cycle are only written to that DUT and are not used for estimating junction temperatures or shutting down heating circuits for other DUTs.

[0108] The preset cycle sequence is determined based on the power modules under test participating in the DC power cycle and their connection order. When there are two power modules under test, the DC power cycle platform performs high-current heating sequentially only between the two power modules under test; when there are three power modules under test, the DC power cycle platform performs high-current heating sequentially between all three power modules under test. The preset cycle sequence is fixed before the DC power cycle begins, ensuring that each power module under test repeatedly enters high-current heating according to the same cycle structure.

[0109] When the power module under test enters high-current heating, the heating circuits of other power modules under test remain off. The DC power cycling platform uses the calibration relationship, online effective thermal resistance, and reference junction temperature boundary of the current power module under test to determine the estimated junction temperature of the current power module under test, and shuts off the current heating circuit when the estimated junction temperature reaches the reference junction temperature boundary. The drain voltage drop of the hot-state small current source formed after shutdown is converted into a feedback junction temperature through the calibration relationship of the current power module under test. The feedback junction temperature is then used to determine the original effective thermal resistance of the current power module under test, the online effective thermal resistance of the next round, and the reference junction temperature boundary of the next round.

[0110] After updating the parameters of the current power module under test, the DC power cycling platform associates and saves the next round of online effective thermal resistance and the next round of reference junction temperature boundary with the current power module under test, and then switches to the next power module under test according to the preset cycling sequence. When the next power module under test starts high-current heating, it reads its own saved online effective thermal resistance and reference junction temperature boundary, but does not read the update results generated by the previous power module under test.

[0111] When each power module under test in the preset cycle sequence has completed one high-current heating, hot-state low-current source leakage voltage drop acquisition, and parameter update, the current cycle ends. Upon entering the next cycle, each power module under test adopts the next cycle's online effective thermal resistance and reference junction temperature boundary determined for itself in the previous cycle as its new online effective thermal resistance and reference junction temperature boundary. Thus, each power module under test maintains independent junction temperature estimation, heating circuit shutdown, and parameter update processes while sharing the DC power cycle platform.

[0112] When operating with three power modules under test, the DC power cycling platform can execute in the order of the first, second, and third power modules under test. When operating with two power modules under test, the heating circuit of the power module under test not participating in the current DC power cycle remains off. Its calibration relationship, target maximum junction temperature, online effective thermal resistance, and reference junction temperature boundary are not included in the current calculation and are not changed by the parameter updates of other power modules under test.

[0113] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A DC power cycling method based on junction temperature closed-loop control, characterized in that, The method includes: Obtain the calibration relationship between the low-current source leakage voltage drop and junction temperature of the power module under test; The power module under test is heated with a large current. The estimated junction temperature is determined based on the real-time case temperature, real-time heating power and online effective thermal resistance. The heating circuit is turned off when the estimated junction temperature reaches the reference junction temperature boundary. The feedback junction temperature is determined by the hot-state small current source leakage voltage drop after shutdown and the calibration relationship, and the original effective thermal resistance is determined by the feedback junction temperature, the case temperature at shutdown time, and the hot-state power before shutdown. The original effective thermal resistance and the online effective thermal resistance are weighted and recursively calculated to obtain the next round of online effective thermal resistance for the next round of junction temperature estimation; Using the target maximum junction temperature as a reference, the next round reference junction temperature boundary for the next round of heating shutdown is determined based on the deviation between the feedback junction temperature and the target maximum junction temperature.

2. The method according to claim 1, characterized in that, The acquisition of the calibration relationship includes: Multiple calibration points were obtained under fixed low current and fixed gate conditions. Each calibration point included a calibration temperature and a low current source drain voltage drop collected at the calibration temperature. The calibration relationship is generated by piecewise linear interpolation based on adjacent calibration points.

3. The method according to claim 1, characterized in that, Before applying high-current heating to the power module under test, the heating circuit is kept off, the leakage voltage drop of the cold-state small current source is obtained, the minimum junction temperature is determined based on the leakage voltage drop of the cold-state small current source and the calibration relationship, and the power module under test is applied high-current heating when the minimum junction temperature is lower than the preset minimum junction temperature target.

4. The method according to claim 1, characterized in that, The determination of the real-time heating power includes: The real-time high-current source leakage voltage drop is obtained, and the product of the circulating high-current set value and the real-time high-current source leakage voltage drop is determined as the real-time heating power. The determination of the estimated junction temperature includes: The estimated junction temperature is obtained by adding the product of the real-time heating power and the online effective thermal resistance to the real-time case temperature.

5. The method according to claim 1, characterized in that, During any sampling period, when the estimated junction temperature reaches the reference junction temperature boundary, the heating circuit is turned off, and the first sampled value collected after the heating circuit is turned off and located within the preset small current source leakage voltage drop range is determined as the hot small current source leakage voltage drop.

6. The method according to claim 1, characterized in that, The shell temperature at the shutdown time is the real-time shell temperature when the heating circuit is shut off, and the hot power before shutdown is the real-time heating power before the heating circuit is shut off; The determination of the original effective thermal resistance includes: The ratio of the difference between the feedback junction temperature and the case temperature at the turn-off time to the hot power before turn-off is determined as the original effective thermal resistance.

7. The method according to claim 1, characterized in that, The weighted recursion includes: The difference obtained by subtracting the thermal resistance update coefficient from the online effective thermal resistance is used as the weight of the online effective thermal resistance, and the thermal resistance update coefficient is used as the weight of the original effective thermal resistance. The online effective thermal resistance and the original effective thermal resistance are weighted and summed to obtain the candidate online effective thermal resistance. The candidate online effective thermal resistance is used to determine the online effective thermal resistance in the next round.

8. The method according to claim 7, characterized in that, From the moment the feedback junction temperature first enters the preset convergence range centered on the target highest junction temperature, the candidate online effective thermal resistance is checked for range. When the candidate online effective thermal resistance is within the preset effective thermal resistance range, the candidate online effective thermal resistance is determined as the online effective thermal resistance for the next round. When the candidate online effective thermal resistance exceeds the preset effective thermal resistance range, the online effective thermal resistance is determined as the online effective thermal resistance for the next round. When the candidate online effective thermal resistances obtained in three consecutive cycles all exceed the preset effective thermal resistance range, the DC power cycle is stopped.

9. The method according to claim 1, characterized in that, The determination of the next round of reference junction temperature boundary includes: The difference between the target maximum junction temperature and the feedback junction temperature is determined as the control deviation. The product of the reference junction temperature correction coefficient and the control deviation is calculated, and the sum of the product and the target maximum junction temperature is determined as the reference junction temperature boundary for the next round. The reference junction temperature boundary for the next round is restricted to between the preset lower limit and the preset upper limit of the reference junction temperature.

10. The method according to claim 1, characterized in that, The number of the power modules under test is two or three, and each power module under test adopts an independent calibration relationship, target maximum junction temperature, online effective thermal resistance and reference junction temperature boundary; Each of the power modules under test is heated with a large current in a preset cycle sequence, and the online effective thermal resistance and reference junction temperature boundary of each power module under test are updated respectively to obtain the next round of online effective thermal resistance and the next round of reference junction temperature boundary of each power module under test.

Images