Method for predicting junction temperature of silicon carbide MOSFET and anti-parallel diode

Abstract

The invention relates to a silicon carbide MOSFET (Metal-Oxide-Semiconductor Field Effect Transistor) and anti-parallel diode junction temperature prediction method, which comprises the following steps: placing a frequency converter power module on a heating plate for heating, and monitoring the temperature of the heating plate in real time through a PTC thermistor sensor; naturally cooling to Tjmax-5 DEG C, and carrying out a pulse test; applying a current vector pulse sequence through a control panel, and repeating current pulse adjustment; when the temperature of the heating plate is reduced every 5 DEG C, testing is repeated; repeating the steps until the temperature of the heating plate is naturally cooled to 25 DEG C, and completing the last test; and training a support vector by using the data set to obtain a junction temperature prediction model of the silicon carbide MOSFET and the anti-parallel diode, thereby obtaining a junction temperature prediction model of the silicon carbide MOSFET and the anti-parallel diode. According to the method, the influence of aging on the accuracy of junction temperature prediction is considered, the calibration test is periodically repeated after the junction temperature prediction model works for a certain hour, and the junction temperature prediction model under two working hours is utilized to realize the full-cycle junction temperature prediction under the intermediate working duration.

Description

Technical Field

[0001] The invention belongs to the technical field of silicon carbide MOSFET, and in particular relates to a method for predicting the junction temperature of a silicon carbide MOSFET and an anti-parallel diode used in a frequency converter. Background Art

[0002] Industrial electrification requires power electronic systems to improve power density, partial load efficiency, reduce costs and extend equipment life. To this end, silicon carbide SiC power devices provide significant performance advantages with innovative packaging and assembly technologies, but high costs and reliability issues still exist. Power inverters require advanced thermal management and accurate temperature models to avoid shortened life or equipment failure due to overheating.

[0003] Semiconductor chips are usually packaged inside modules, and it is difficult to directly observe their junction temperature. At present, scholars have conducted in-depth research on the extraction and detection methods of junction temperature of power devices and proposed a variety of solutions, which can be mainly summarized into four categories: optical method, thermal network method, physical contact method and temperature-sensitive parameter method. The temperature-sensitive parameter method has attracted widespread attention due to its suitability for online measurement, fast response speed and low cost, and has gradually become a research hotspot.

[0004] The physical mechanism of the temperature-sensitive parameter method is that some physical parameters inside the device change with the junction temperature. For example, the intrinsic carrier concentration and carrier lifetime increase with the increase of temperature, while the carrier mobility decreases with the increase of temperature. These changes cause the electrical parameters of the device to shift, which is manifested externally as changes in parameters such as on-resistance / voltage drop, switching delay, and voltage / current change rate. These electrical parameters affected by junction temperature are called temperature-sensitive parameters. Different temperature-sensitive parameter methods have their own advantages and disadvantages in terms of sensitivity, linearity, and robustness. It is necessary to select appropriate temperature-sensitive parameters for junction temperature extraction based on the specific device and actual operating conditions.

[0005] Among the many temperature-sensitive parameters, those directly affected by junction temperature changes are called direct temperature-sensitive parameters, including threshold voltage, on-resistance / voltage drop, Miller capacitance, and transconductance, etc. Indirect temperature-sensitive parameters are affected by one or more direct temperature-sensitive parameters, including turn-on / turn-off delay, short-circuit current, voltage / current change rate, etc.

[0006] Although there are many junction temperature estimation techniques, they have not been widely used in industry, especially the junction temperature prediction of anti-parallel diodes, which is critical for device health monitoring. Monitoring the temperature of all power electronic devices, including diodes, can improve system reliability, enable dynamic adjustment of inverter peak current without increasing the risk of failure, and facilitate more in-depth diagnostics.

[0007] Traditional methods such as optical temperature measurement and thermistor temperature measurement are difficult to apply in power modules due to high cost or reliability issues. The combination of temperature sensors and lumped parameter thermal network (LPTN) models, although low cost, its model parameters are easily affected by time, resulting in inaccurate junction temperature estimation. The calibration of the temperature-sensitive parameter method is usually performed offline at the device level, using dedicated laboratory equipment such as power cycling. The calibration data is later used for temperature estimation on the power board, where additional parasitic effects of the board are difficult to consider. It is also necessary to pay expensive testing fees or purchase related experimental equipment, which has high time and economic costs, making it difficult to be effectively promoted and applied. Summary of the invention

[0008] In the present invention, the temperature sensitive electrical parameters (TSEP) of SiC MOSFET and anti-parallel diode are calibrated directly on the power board of the target inverter, without the need for additional dedicated equipment, using the existing current and voltage measurement system on the inverter. This means that if the measurement errors are consistent over time, they will not affect the accuracy of the temperature estimation. On this basis, the present invention takes into account the influence of the degree of device aging in the junction temperature prediction, repeats the calibration test regularly after a certain number of hours of work, and uses the junction temperature prediction model under two working times to achieve full-cycle junction temperature prediction under intermediate working time.

[0009] The present invention is achieved through the following technical solutions:

[0010] The present invention provides a method for predicting the junction temperature of a silicon carbide MOSFET and an anti-parallel diode, comprising the following steps:

[0011] (1) Place the inverter power module on the heating plate, turn on the heating plate, set the temperature controller to preheat to the specified Tj_max temperature, and monitor the temperature of the heating plate in real time through the PTC thermistor sensor. When the temperature of the PTC thermistor sensor reaches the target temperature, stop heating. At this time, the reading of the PTC thermistor sensor can represent the junction temperature of the MOSFET and the anti-parallel diode;

[0012] (2) Use the PTC thermistor sensor to continue monitoring the temperature of the heating plate. When it cools naturally to Tj_max-5°C, prepare to start the pulse test. Send the first group of short current pulses in the spatial vector direction of phase a. The amplitude of the pulse current gradually increases from 0.1*ID_max to ID_max. Ensure that the current increases step by step according to the predetermined value, and record the conduction voltage drop of the silicon carbide MOSFET and the anti-parallel diode of the inverter;

[0013] (3) At a temperature of Tj_max-5°C, apply the same current pulse sequence to the other five space vector directions respectively, repeat the current pulse adjustment in step (2), ensure that the duration and amplitude of each current pulse are consistent, and during the application of each set of current pulses, record the exact temperature measured by the PTC thermistor sensor at each pulse;

[0014] (4) After testing the current pulse of Tj_max-5°C, allow the heating plate to continue to cool naturally, monitor the temperature displayed by the PTC thermistor sensor, and repeat the pulse test in steps (2) and (3) every time the temperature of the heating plate drops to Tj_max-10°C, Tj_max-15°C, and so on, at intervals of 5°C, to ensure that the duration and amplitude of the current pulse remain consistent. When testing at each temperature level, continue to record the exact temperature measured by the PTC thermistor sensor and the data corresponding to the current pulse;

[0015] (5) Repeat step (4) until the temperature of the heating plate cools naturally to 25°C. At 25°C, complete the last current pulse test and record the final data;

[0016] (6) Through the above experimental steps, a data set of junction temperature related to on-state voltage and current can be obtained. By using the data set to train the support vector, a junction temperature prediction model of silicon carbide MOSFET and anti-parallel diode can be obtained.

[0017] In a preferred embodiment of the present invention, in step (2), the duration of the short current pulse does not exceed 150 microseconds, and the interval between each pulse is 0.25 seconds.

[0018] In order to ensure the accuracy of the prediction results, in step (6), data sets under multiple working hours are obtained through discrete sampling, and then these data sets are used for real-time prediction;

[0019] Furthermore, the real-time prediction method is as follows: firstly, two adjacent sets of data are used for prediction, and then the prediction results are weighted according to the difference in usage time:

[0020]

[0021] In the formula, t1 and t2 represent the working time of the silicon carbide device when collecting data, t represents the actual working time of the current device, and f t1 represents the junction temperature prediction model under the working time of t1, f t2 It represents the junction temperature prediction model under the working time of t2.

[0022] The beneficial effects of the present invention are:

[0023] The present invention is based on the method for estimating the temperature of SiC MOSFET and extends it to estimating the temperature of the anti-parallel diode in the device. Although the diode temperature is usually ignored in MOSFET inverters because it is considered not to be a critical factor, this is not always the case, especially in inverters where the anti-parallel diode is designed to run continuously. The proposed diode junction temperature monitoring function does not require any hardware cost, and the VCE monitoring circuit for SiC MOSFET junction temperature prediction can still be reused for the on-state voltage test of the diode; the temperature sensitive electrical parameters (TSEP) of the SiC MOSFET and the anti-parallel diode can be calibrated directly on the main circuit of the inverter without the need for complex experiments such as power cycling or the use of other high-cost equipment, and the additional circuit required for the VON measurement system is simple and low-cost; the junction temperature prediction model used in the present invention does not involve complex calculations or models that are difficult to adjust, and the non-ideality of the on-board voltage measurement system (such as offset and gain errors) will not affect the junction temperature estimator; and the present invention takes into account the influence of aging on the accuracy of junction temperature prediction, regularly repeats the calibration test after a certain number of hours of work, and uses the junction temperature prediction model under two working times to achieve full-cycle junction temperature prediction under intermediate working time. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a schematic structural diagram of a junction temperature estimator of the present invention;

[0025] Figure 2 is the V of the silicon carbide MOSFET of the present invention CE Measurement circuit diagram;

[0026] Figure 3 It is a schematic diagram of the data acquisition platform of the present invention;

[0027] Figure 4 It is a flow chart of the method for predicting the junction temperature of a silicon carbide MOSFET and an anti-parallel diode of the present invention;

[0028] Figure 5 It is the space vector diagram of the frequency converter of the present invention. DETAILED DESCRIPTION

[0029] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0030] The junction temperature estimator proposed in the present invention utilizes the relationship between the junction temperature of the MOSFET and its on-resistance. Thanks to the online junction temperature estimator, the inverter can fully utilize the safe operating area (SOA) of the power MOSFET without the risk of thermal failure. The present invention improves the existing MOSFET temperature estimation method, including the temperature estimation of the anti-parallel diode, while minimizing the software complexity.

[0031] The functional diagram of the junction temperature estimator for MOSFET and diode is shown in Figure 1 As shown in the figure, the two junction temperature estimators have the same principle, which is to collect the on-state voltage Von and the on-state current Ion for junction temperature prediction, and input them into the fitting surface of a trained support vector machine model to obtain the junction temperature estimation Tj of the device. These support vector machine models, which are unique to each device, are trained with data obtained in advance through self-calibration tests performed directly on the inverter.

[0032] V of SiC MOSFET CE Measuring circuit such as Figure 2 As shown in the figure, in each PWM cycle, the saturation voltage of the MOSFET is measured, and the on-state voltage of the anti-parallel diode can still be measured by multiplexing this circuit. This circuit can block high voltage and prevent switching transients. In the worst-case switching situation, the clamping diode will limit the voltage amplitude. The measurement circuit is equipped for each of the six MOS tubes in the inverter.

[0033] The six MOSFETs and anti-parallel diodes in the inverter are respectively in the current range of 0.1*I D_max to I D_max Ampere and Junction Temperature Range 25°C to T j_max To identify the following, use the following acquisition process. D_max and T j_max They are the maximum on-state current and maximum junction temperature of SiC MOS devices respectively.

[0034] The present invention provides a method for predicting the junction temperature of a silicon carbide MOSFET and an anti-parallel diode based on the following example: Figure 3 In the data acquisition platform shown, the inverter power module 2 is placed on the heating plate 1, the PTC thermistor sensor 3 is placed on the inverter power module 2, the PTC thermistor sensor 3 and the heating plate 1 are respectively connected to the controller 4, the PTC thermistor sensor 3 can monitor the temperature of the heating plate 1 and the inverter power module 2 in real time, the controller 4 can collect voltage, current and temperature signals in real time, and can also generate waves to control the current pulses of the SiC power module.

[0035] like Figure 4As shown, the method for predicting the junction temperature of a silicon carbide MOSFET and an anti-parallel diode of the present invention specifically comprises the following steps:

[0036] (1) First, turn on the heating plate and set the temperature controller to preheat to the specified T j_max Temperature. The PTC thermistor sensor monitors the temperature of the heating plate in real time to ensure that it reaches T j_max When the temperature measured by the PTC thermistor sensor reaches the target temperature, the heating plate is immediately turned off and external heating is stopped. At this time, the heating plate cools down naturally. Due to the slow heat dissipation, the junction temperature of all devices will remain close to the temperature of the heating plate. Therefore, the reading of the PTC thermistor sensor can represent the junction temperature of the SiC MOSFET and the anti-parallel diode.

[0037] (2) Use the PTC thermistor sensor to continue monitoring the temperature of the heating plate and wait for it to cool naturally to T j_max At -5°C, the pulse test is ready to start. Figure 5 As shown, the first group of short current pulses are emitted in the direction of the space vector of phase a, with a duration of no more than 150 microseconds and an interval of about 0.25 seconds between each pulse. The amplitude of the pulse current should be from 0.1*I D_max Gradually increase to I D_max , ensuring that the current increases step by step according to the predetermined value, and recording the conduction voltage drop of the silicon carbide MOSFET and the anti-parallel diode of the specific power device. The duration of the short pulse is controlled to be very short to ensure that the junction temperature does not rise rapidly due to the current pulse and can still be represented by the temperature of the PTC thermistor.

[0038] (3) In T j_max At -5℃, Figure 5 As shown, the same current pulse sequence is applied to the other five space vector directions (phases b, c, d, e, and f). Repeat the current pulse adjustment in step (2) to ensure that the duration and amplitude of each current pulse are consistent. During the application of each set of current pulses, the exact temperature measured by the PTC thermistor sensor during each pulse is recorded to facilitate subsequent analysis of performance changes at different junction temperatures.

[0039] (4) After testing T j_max After the -5℃ current pulse, let the heating plate continue to cool naturally and monitor the temperature displayed by the PTC thermistor sensor. j_max -10℃、T j_maxRepeat the pulse tests in steps (2) and (3) at 5°C intervals, such as -15°C, and so on. Ensure that the amplitude and duration of the current pulse remain consistent at each temperature level to prevent temperature fluctuations from affecting the test results. Continue to record the exact temperature and current pulse data measured by the PTC thermistor sensor at each temperature level.

[0040] (5) Repeat step (4) until the temperature of the heating plate cools naturally to 25°C. At 25°C, complete the last current pulse test and record the final data. Ensure that the data of all temperature levels are fully recorded to provide a basis for subsequent device calibration and thermal performance analysis.

[0041] The assumption that the junction temperature is equal to the heating plate temperature is verified as follows. A single current pulse with a duration of less than 150 microseconds causes a junction temperature increase of less than 2°C in the MOSFET. An idle time of 0.25 seconds before the next pulse resets the remaining temperature perturbation. Therefore, the pulses are considered to be thermally independent of each other. When the heating plate approaches the low temperature range, the total test time can be consistently reduced by using forced cooling during the test.

[0042] (6) Through the above experimental steps, a data set of junction temperature related to on-state voltage and current can be obtained. By using this data set to train the support vector, a junction temperature prediction model for silicon carbide MOS and anti-parallel diode can be obtained.

[0043] The electrical and thermal characteristics of power semiconductors tend to change as components age due to multiple degradation phenomena such as bond wire lifting, die delamination and cracking, oxide degradation and solder fatigue. Both MOSFETs and diodes experience degradation of their conduction characteristics during their service life, which is reflected in an increase in on-resistance (i.e. on-voltage / on-current).

[0044] Since the proposed junction temperature estimator is based on component conduction characteristics obtained during calibration tests performed at the beginning of the inverter life, the accuracy of the junction temperature will deteriorate as the component age increases.

[0045] As a precaution, one way to maintain the accuracy of the junction temperature estimator is to repeat the calibration test regularly after a certain number of working hours. However, this test method cannot guarantee online real-time updates, so data under multiple working hours can be obtained through discrete sampling, and then these data can be used for prediction.

[0046] The real-time prediction method is as follows: First, two sets of adjacent data are used for prediction, and then the prediction results are weighted according to the difference in usage time:

[0047]

[0048] In the formula, t1 and t2 represent the working time of the silicon carbide device when collecting data, t represents the actual working time of the current device, and f t1 represents the junction temperature prediction model under the working time of t1, f t2 It represents the junction temperature prediction model under the working time of t2.

[0049] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for predicting junction temperature of silicon carbide MOSFET and anti-parallel diode, characterized in that: The steps include: (1) Place the inverter power module on the heating plate, turn on the heating plate, set the temperature controller to preheat to the specified Tj_max temperature, and monitor the temperature of the heating plate in real time through the PTC thermistor sensor. When the temperature of the PTC thermistor sensor reaches the target temperature, stop heating. At this time, the reading of the PTC thermistor sensor can represent the junction temperature of the MOSFET and the anti-parallel diode; (2) Use the PTC thermistor sensor to continue monitoring the temperature of the heating plate. When it cools naturally to Tj_max-5°C, prepare to start the pulse test. Send the first group of short current pulses in the spatial vector direction of phase a. The amplitude of the pulse current gradually increases from 0.1*ID_max to ID_max. Ensure that the current increases step by step according to the predetermined value, and record the conduction voltage drop of the silicon carbide MOSFET and the anti-parallel diode of the inverter; (3) At a temperature of Tj_max-5°C, apply the same current pulse sequence to the other five space vector directions respectively, repeat the current pulse adjustment in step (2), ensure that the duration and amplitude of each current pulse are consistent, and during the application of each set of current pulses, record the exact temperature measured by the PTC thermistor sensor at each pulse; (4) After testing the current pulse of Tj_max-5°C, allow the heating plate to continue to cool naturally, monitor the temperature displayed by the PTC thermistor sensor, and repeat the pulse test in steps (2) and (3) every time the temperature of the heating plate drops to Tj_max-10°C, Tj_max-15°C, and so on, at intervals of 5°C, to ensure that the duration and amplitude of the current pulse remain consistent. When testing at each temperature level, continue to record the exact temperature measured by the PTC thermistor sensor and the data corresponding to the current pulse; (5) Repeat step (4) until the temperature of the heating plate cools naturally to 25°C. At 25°C, complete the last current pulse test and record the final data; (6) Through the above experimental steps, a data set of junction temperature related to on-state voltage and current can be obtained. By using the data set to train the support vector, a junction temperature prediction model of silicon carbide MOSFET and anti-parallel diode can be obtained.

2. The method for predicting junction temperature of silicon carbide MOSFET and anti-parallel diode according to claim 1, characterized in that: In step (2), the duration of the short current pulse does not exceed 150 microseconds, and the interval between each pulse is 0.25 seconds.

3. The method for predicting junction temperature of silicon carbide MOSFET and anti-parallel diode according to claim 2, characterized in that: In step (6), data sets under multiple working hours are obtained through discrete sampling, and then these data sets are used for real-time prediction.

4. The method for predicting junction temperature of silicon carbide MOSFET and anti-parallel diode according to claim 3, characterized in that: The real-time prediction method is as follows: first, two adjacent sets of data are used for prediction, and then the prediction results are weighted according to the difference in usage time: In the formula, t1 and t2 represent the working time of the silicon carbide device when collecting data, t represents the actual working time of the current device, and f t1 represents the junction temperature prediction model under the working time of t1, f t2 It represents the junction temperature prediction model under the working time of t2.

Images