A method for monitoring real-time junction temperature of a power module multi-chip parallel connection

Abstract

The present application relates to the technical field of power electronic equipment technology, provide a kind of power module multicore parallel real-time junction temperature monitoring method, comprising the following steps 1, for each parallel IGBT chip in power module, two key temperature monitoring points are determined respectively, step 2, complete temperature sensor assembly power module is placed in equivalent test condition, specified electrical load is applied, to simulate its real operating conditions, step 3, according to the temperature sensitivity coefficient obtained in the calibration process of each temperature sensor, the optical signal collected in step 2 is converted into temperature value, step 4, a large number of original junction temperature data obtained in step 3 are processed and deeply analyzed, not only the junction temperature data is output in the present application, more through subsequent data processing, extract a plurality of key characteristic parameters including maximum junction temperature, chip temperature difference, temperature imbalance degree, thermal time constant, transient temperature change rate extreme.

Description

Technical Field

[0001] This invention relates to the field of power electronic equipment technology, and more specifically to a method for real-time monitoring of the junction temperature of a power module with multiple chips connected in parallel. Background Technology

[0002] Power modules, especially insulated-gate bipolar transistor (IGBT) modules with multi-chip parallel structures, are core components of modern power electronic equipment (such as new energy converters and industrial motor drives). Their reliability directly determines the overall performance and lifespan. Junction temperature is the most critical parameter affecting the reliability of power devices; excessively high junction temperatures or excessive temperature fluctuations can lead to permanent failures such as material aging, bond wire detachment, and even chip burnout.

[0003] In multi-chip parallel applications, due to factors such as manufacturing process deviations, layout asymmetry, and uneven parasitic parameters, the current distribution among the chips is often uneven, leading to significant differences in junction temperatures. This uneven temperature distribution limits the overall performance of the module to the temperature of the hottest chip, severely restricting its performance and creating a potential safety hazard of thermal runaway. Therefore, real-time, accurate, and multi-point synchronous monitoring of the junction temperature of parallel chips is crucial for assessing module health, optimizing control, and balancing performance and safety.

[0004] Existing junction temperature measurement methods, such as thermocouples and infrared thermal imaging, have significant limitations in practical applications. Thermocouples have slow response speeds, present electrical insulation challenges, and are difficult to deploy at multiple points within the package. While infrared thermal imaging can visually display the temperature field, it requires opening the package for measurement, making it unsuitable for equipment in actual operation, and it is susceptible to surface emissivity. Electrical parameter methods (such as TSEP) do not require module modification, but their calibration is complex, their accuracy is greatly affected by electrical stress, and they struggle to distinguish the temperatures of multiple parallel chips.

[0005] To address the aforementioned problems, we propose a method for real-time junction temperature monitoring of multiple chips connected in parallel in power modules. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for real-time junction temperature monitoring of multiple parallel chips in power modules. This method overcomes the deficiencies of existing technologies by integrating online electrothermal measurement, a dynamic thermal model (predictive), and fiber optic sensing to construct a multi-redundant system, achieving full lifecycle monitoring of the junction temperature of multiple chips.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for real-time junction temperature monitoring of a multi-chip parallel power module includes the following steps:

[0009] Step 1: For each parallel IGBT chip in the power module, determine two key temperature monitoring points, namely the IGBT cell region and the anti-parallel diode region; select a temperature sensor, securely install it at each chip monitoring point, and ensure reliable thermal contact between the temperature sensor and the chip surface.

[0010] Step 2: Place the power module with the temperature sensor assembly under equivalent test conditions and apply the specified electrical load to simulate its real operating conditions. Collect the optical signals fed back by each temperature sensor in real time through the fiber optic demodulator and record the change of its wavelength over time.

[0011] Step 3: Based on the temperature sensitivity coefficient obtained by each temperature sensor during the calibration process, convert the optical signal collected in Step 2 into a temperature value to generate real-time junction temperature data of each IGBT chip in different functional areas.

[0012] Step 4: Process and perform in-depth analysis on the large amount of raw junction temperature data obtained in Step 3, extract key features characterizing the thermal behavior of the module, and automatically generate a structured test report that can be used for engineering decision-making.

[0013] Preferably, the measurement point positioning method in step 1 specifically includes: with the power module in the open state, a preliminary non-contact scan of the chip surface to which the working current is applied is performed using infrared thermal imaging to obtain the temperature field distribution and hot spot areas. Subsequently, combined with finite element thermal simulation analysis, the detailed heat flow distribution and thermal coupling effect inside the chip are calculated in the model, thereby accurately locating the highest temperature point in the IGBT cell region caused by the current crowding effect and the heat concentration area in the diode region caused by conduction loss. Finally, a high-precision dedicated positioning fixture is used to align and install the temperature sensor at the predetermined position jointly determined by simulation and experiment, and a thermally conductive adhesive with high thermal conductivity is used for filling and fixing.

[0014] Preferably, in step 2, the real-time acquisition process of the fiber optic demodulator uses a high-speed demodulator to synchronously acquire the optical signals of all temperature sensors to ensure that the rapid temperature changes caused by switching transients can be captured, and wavelength division multiplexing is used to allocate an independent wavelength channel for each temperature sensor, thereby transmitting multiple signals in a single optical fiber and accurately distinguishing their sources.

[0015] The demodulator integrates a tunable Fabry-Perot filter, which uses piezoelectric ceramic drive to achieve wavelength scanning and accurately detect the offset of the Bragg wavelength of each temperature sensor due to thermal expansion.

[0016] During the data acquisition process, the correspondence between the wavelength shift sequence of each temperature sensor and time is recorded in real time. The data structure can be represented as follows: Where n is the number of chips, m is the number of sampling points, and the load current waveform is recorded simultaneously. The expression is With mother-money voltage waveform The expression is ;in For the corresponding Two monitoring locations / parameters for each chip; for The load current value at any given time; for The load voltage value at any given time; for Time of the first The wavelength offset of each temperature sensor is used to provide a complete background of electro-thermal coupling data, facilitating subsequent analysis of the relationship between junction temperature and electrical operating conditions.

[0017] Preferably, the process of converting the optical signal into a temperature value in step 3 is achieved by the following formula: ,in, This refers to the change in temperature. This is the temperature sensitivity coefficient; For constant terms, This is the strain sensitivity coefficient; The change in strain;

[0018] For sensors that have been specially encapsulated and cured to ensure no relative displacement between them and the measured object, thus making the strain effect negligible, the formula can be simplified to: ,in, This is the real-time temperature value; For reference temperature; The wavelength is measured in real time; Reference wavelength;

[0019] Based on this transformation relationship, an independent temperature-time series is generated for each temperature sensor. Subsequently, the data were categorized according to chip number and functional area type: for each IGBT chip, its cell temperature sequence was compiled. ) and diode region temperature sequence This ultimately forms a real-time junction temperature dataset for each chip in different functional areas.

[0020] Preferably, step 4 specifically includes: data filtering: smoothing the raw temperature data using a Savitzky-Golay filter.

[0021] Steady-state junction temperature extraction: During the load stabilization phase, the average value of temperature data within a sufficiently long time window is taken as the steady-state junction temperature value under that operating condition. The calculation formula is: ,in, This represents the total number of sampling points within the stable time window. The temperature value at the j-th sampling point

[0022] Transient temperature change analysis: Extracting the rate of temperature change during switching processes The calculation formula is: ,in For time values; This is the temperature value;

[0023] Temperature distribution consistency calculation: To evaluate the current sharing and heat dissipation uniformity when multiple chips are connected in parallel, the standard deviation of temperature in the same functional area of ​​all chips at the same time is calculated. The expression is: ,in, For the first Temperature at each measurement point, and the temperature value at each measurement point. Let be the average temperature at all measurement points. This represents the number of measurement points.

[0024] Preferably, step 4 further extracts the following key features from the processed data:

[0025] Maximum junction temperature : The highest junction temperature of all chips in all functional regions under steady-state operation;

[0026] Maximum temperature difference between chips Defined as the difference between the highest and lowest temperatures at the same time across all measured points; the expression is: This reflects the extreme differences in temperature distribution within the module.

[0027] Temperature imbalance A normalized index characterizing the uniformity of temperature distribution between chips, calculated using the following formula: ,in This represents the average junction temperature.

[0028] Thermal time constant :Through the exponential function T The temperature rise response curve of the chip was obtained by nonlinear fitting. A is the thermal time constant, representing how quickly the chip temperature reaches a steady state from its initial state; B is the steady-state temperature, and A reflects the magnitude of temperature change.

[0029] Extreme values ​​of the rate of temperature change ‎: The maximum absolute value of the rate of temperature change over time during the entire test.

[0030] Preferably, the final output of step 4 integrates all analysis results, calculated key parameters, and various curves and charts according to a predefined standard template format to generate a complete structured junction temperature test report.

[0031] This invention provides a method for real-time junction temperature monitoring of multiple chips connected in parallel in a power module. It offers the following advantages: by arranging sensors in two key functional areas (cell area and diode area) of each IGBT chip and utilizing wavelength division multiplexing (WDM) technology, signals from all monitoring points can be synchronously transmitted in a single optical fiber. This enables synchronous and real-time direct measurement of the junction temperature of all key chips and key areas within the multi-chip parallel module, completely solving the problems of inaccurate and incomplete measurements inherent in traditional methods.

[0032] The sensor uses optical signals as a medium, is immune to strong electromagnetic interference, and is suitable for operation inside high-voltage, high-current power modules. Combined with the Savitzky-Golay filtering signal processing algorithm, it can accurately extract weak temperature feature signals from noise, resulting in accurate and reliable measurement results.

[0033] This invention not only outputs junction temperature data, but also extracts a number of key characteristic parameters through subsequent data processing, including maximum junction temperature, inter-chip temperature difference, temperature imbalance, thermal time constant, and extreme values ​​of transient temperature change rate. These parameters provide comprehensive and in-depth quantitative data support for thermal management strategies (such as active thermal control) for module thermal design verification, health status assessment, and life prediction.

[0034] By accurately identifying hotspots and quantifying the degree of temperature distribution unevenness, direct improvement directions are provided for the packaging design, layout optimization, and parallel current sharing of power modules, thereby improving product reliability and power density from the source. Attached Figure Description

[0035] Figure 1 This is a flowchart of the method of the present invention;

[0036] Figure 2 This is a flowchart of the algorithm for step 4 of the present invention; Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] See attached document Figures 1-2As shown, a method for real-time junction temperature monitoring of a multi-chip parallel power module mainly involves measuring the operating status of the power module and outputting junction temperature measurement data. In a multi-chip parallel module, uneven temperature distribution can lead to permanent device damage if the junction temperature exceeds a threshold. Real-time junction temperature measurement is used to balance performance and safety, and includes the following steps:

[0039] Step 1: For each parallel IGBT chip in the power module, determine two key temperature monitoring points: the IGBT cell region and the anti-parallel diode region; select a temperature sensor (fiber Bragg grating), and securely install it at each chip monitoring point, ensuring reliable thermal contact between the temperature sensor and the chip surface; to ensure the accuracy and real-time performance of temperature transmission.

[0040] Step 2: Place the power module with assembled temperature sensors under equivalent test conditions (such as a dual-pulse test platform or an inverter circuit simulating actual operation), apply a specified electrical load (including specific voltage, current and switching frequency) to simulate its real operating conditions, and collect the optical signals fed back by each temperature sensor in real time through an optical fiber demodulator to record the change of its wavelength over time.

[0041] Step 3: Based on the temperature sensitivity coefficient obtained by each temperature sensor during the calibration process, convert the optical signal collected in Step 2 into a temperature value to generate real-time junction temperature data of each IGBT chip in different functional areas.

[0042] Step 4: Process and perform in-depth analysis on the large amount of raw junction temperature data obtained in Step 3, extract key features characterizing the thermal behavior of the module, and automatically generate a structured test report that can be used for engineering decision-making.

[0043] The measurement point positioning method in step 1 specifically includes: with the power module uncovered, a preliminary non-contact scan of the chip surface to which the operating current is applied is performed using infrared thermal imaging to obtain the temperature field distribution and hot spot areas. Subsequently, combined with finite element thermal simulation analysis, the detailed heat flow distribution and thermal coupling effect inside the chip are calculated in the model, thereby accurately locating the highest temperature point in the IGBT cell region caused by the current crowding effect and the heat concentration area in the diode region caused by conduction loss. Finally, a high-precision dedicated positioning fixture is used to align and install the temperature sensor at the predetermined position jointly determined by simulation and experiment, and a thermally conductive adhesive with high thermal conductivity (such as silver paste or epoxy resin) is used for filling and fixing to maximize the heat transfer efficiency from the chip to the temperature sensor and reduce measurement delay and error.

[0044] The real-time acquisition process of the fiber optic demodulator in step 2 uses a high-speed demodulator (sampling rate ≥ 1 kHz) to synchronously acquire the optical signals of all temperature sensors to ensure that the rapid temperature changes caused by switching transients can be captured. Wavelength division multiplexing (WDM) is used to allocate an independent wavelength channel for each temperature sensor, thereby transmitting multiple signals in a single optical fiber and accurately distinguishing their sources.

[0045] The demodulator integrates a tunable Fabry-Perot filter, which uses piezoelectric ceramic drive to achieve wavelength scanning and accurately detect the offset of the Bragg wavelength of each temperature sensor due to thermal expansion.

[0046] During the data acquisition process, the correspondence between the wavelength shift sequence of each temperature sensor and time is recorded in real time. The data structure can be represented as follows: Where n is the number of chips (e.g., the number of IGBT modules), and m is the number of sampling points (timestamps), while simultaneously recording the load current waveform. The expression is With mother-money voltage waveform The expression is ;in For the corresponding Two monitoring locations / parameters for each chip; for The load current value at any given time; for The load voltage value at any given time; for Time of the first The wavelength offset of each temperature sensor is used to provide a complete background of electro-thermal coupling data, facilitating subsequent analysis of the relationship between junction temperature and electrical operating conditions.

[0047] Wavelength division multiplexing (WDM) technology allows multiple temperature sensors to share the same fiber optic channel, distinguishing signals by different Bragg wavelengths and enabling synchronous monitoring of multiple chips.

[0048] The optical signals fed back from each temperature sensor are acquired in real time by an optical fiber demodulator. The process of converting the optical signals in step 3 into temperature values ​​is achieved by the following formula: ,in, This refers to the change in temperature. This is the temperature sensitivity coefficient; For constant terms, This is the strain sensitivity coefficient; The formula takes into account the cross-sensitivity effect of temperature and strain on the wavelength of the temperature sensor.

[0049] For sensors that have been specially encapsulated and cured to ensure no relative displacement between them and the measured object, thus making the strain effect negligible, the formula can be simplified to: ,in, This is the real-time temperature value; For reference temperature; The wavelength is measured in real time; Reference wavelength;

[0050] Based on this transformation relationship, an independent temperature-time series is generated for each temperature sensor. Subsequently, the data were categorized according to chip number and functional area type: for each IGBT chip, its cell temperature sequence was compiled. ) and diode region temperature sequence This ultimately creates a real-time junction temperature dataset for each chip in different functional areas, providing a data foundation for subsequent analysis.

[0051] Step 4 specifically includes: Data filtering: The original temperature data is smoothed using a Savitzky-Golay filter. The Savitzky-Golay filter is a digital filter based on local polynomial least squares fitting. Its advantage is that it can effectively suppress random noise while better preserving the original signal characteristics in the temperature response curve, such as peak value, width and other key thermal feature information, and avoid introducing obvious phase distortion.

[0052] Steady-state junction temperature extraction: During the load stabilization phase (typically requiring current fluctuations to be controlled within ±2%), the average value of temperature data within a sufficiently long time window is taken as the steady-state junction temperature value under this operating condition. The calculation formula is: ,in, This represents the total number of sampling points within the stable time window. The temperature value at the j-th sampling point

[0053] Transient temperature change analysis: Extracting the rate of temperature change during switching processes (such as turning on and off). The calculation formula is: ,in For time values; This is the temperature value;

[0054] Temperature distribution consistency calculation: To evaluate the current sharing and heat dissipation uniformity when multiple chips are connected in parallel, the standard deviation of temperature in the same functional area (e.g., all cell regions) of all chips at the same time is calculated. The expression is: ,in, For the first Temperature at each measurement point, and the temperature value at each measurement point. Let be the average temperature at all measurement points. This represents the number of measurement points.

[0055] Step 4 further extracts the following key features from the processed data.

[0056] Maximum junction temperature The highest junction temperature of all chips in all functional areas under steady-state operation is the most direct indicator for evaluating module reliability.

[0057] Maximum temperature difference between chips Defined as the difference between the highest and lowest temperatures at the same time across all measured points; the expression is: This reflects the extreme differences in temperature distribution within the module and is an important basis for judging whether there is a serious thermal imbalance.

[0058] Temperature imbalance A normalized index characterizing the uniformity of temperature distribution between chips, calculated using the following formula: ,in This represents the average junction temperature; the smaller the value, the more uniform the temperature distribution.

[0059] Thermal time constant :Through the exponential function T The temperature rise response curve of the chip was obtained by nonlinear fitting. A is the thermal time constant, representing how quickly the chip temperature reaches a steady state from its initial state; B is the steady-state temperature, and A reflects the magnitude of temperature change.

[0060] Extreme values ​​of the rate of temperature change The maximum absolute value of the rate of temperature change over time during the entire test process; this parameter reflects the most drastic degree of dynamic temperature change and is closely related to the switching losses and transient thermal stress of power devices.

[0061] The final output of step 4 integrates all analysis results, calculated key parameters, and various graphs (such as temperature-time curves, temperature distribution cloud maps, current-temperature superimposed waveforms, etc.) according to a predefined standard template format to generate a complete structured junction temperature test report (usually output as a PDF document). This report provides direct and reliable data support for the thermal design verification, condition monitoring, and safety assessment of power modules.

[0062] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time monitoring of the junction temperature of a multi-chip parallel power module, characterized in that, Includes the following steps: Step 1: For each parallel IGBT chip in the power module, determine two key temperature monitoring points, namely the IGBT cell region and the anti-parallel diode region; select a temperature sensor, securely install it at each chip monitoring point, and ensure reliable thermal contact between the temperature sensor and the chip surface. Step 2: Place the power module with the temperature sensor assembly under equivalent test conditions and apply the specified electrical load to simulate its real operating conditions. Collect the optical signals fed back by each temperature sensor in real time through the fiber optic demodulator and record the change of its wavelength over time. Step 3: Based on the temperature sensitivity coefficient obtained by each temperature sensor during the calibration process, convert the optical signal collected in Step 2 into a temperature value to generate real-time junction temperature data of each IGBT chip in different functional areas. Step 4: Process and perform in-depth analysis on the large amount of raw junction temperature data obtained in Step 3, extract key features characterizing the thermal behavior of the module, and automatically generate a structured test report that can be used for engineering decision-making. Step 4 specifically includes: data filtering: smoothing the raw temperature data using a Savitzky-Golay filter. Steady-state junction temperature extraction: During the load stabilization phase, the average value of temperature data within a sufficiently long time window is taken as the steady-state junction temperature value under that operating condition. The calculation formula is: ,in, This represents the total number of sampling points within the stable time window. Let j be the temperature value of the j-th sampling point; Transient temperature change analysis: Extracting the rate of temperature change during switching processes The calculation formula is: ,in For time values; This is the temperature value; Temperature distribution consistency calculation: To evaluate the current sharing and heat dissipation uniformity when multiple chips are connected in parallel, the standard deviation of temperature in the same functional area of ​​all chips at the same time is calculated. The expression is: ,in, For the first Temperature at each measurement point, and the temperature value at each measurement point. Let be the average temperature at all measurement points. The number of measurement points; Step 4 further extracts the following key features from the processed data. Maximum junction temperature : The highest junction temperature of all chips in all functional regions under steady-state operation; Maximum temperature difference between chips Defined as the difference between the highest and lowest temperatures at the same time across all measured points; the expression is: This reflects the extreme differences in temperature distribution within the module; Temperature imbalance A normalized index characterizing the uniformity of temperature distribution between chips, calculated using the following formula: ,in This represents the average junction temperature. Thermal time constant :Through the exponential function T The temperature rise response curve of the chip was obtained by nonlinear fitting. A is the thermal time constant, representing how quickly the chip temperature reaches a steady state from its initial state; B is the steady-state temperature, and A reflects the magnitude of temperature change. Extreme values ​​of the rate of temperature change ‎: The maximum absolute value of the rate of temperature change over time during the entire test.

2. The method for monitoring the real-time junction temperature of a multi-chip parallel power module according to claim 1, characterized in that, The measurement point positioning method in step 1 specifically includes: with the power module in the open state, a preliminary non-contact scan of the chip surface to which the operating current is applied is performed using infrared thermal imaging to obtain the temperature field distribution and hot spot areas. Subsequently, combined with finite element thermal simulation analysis, the detailed heat flow distribution and thermal coupling effect inside the chip are calculated in the model, thereby accurately locating the highest temperature point in the IGBT cell region caused by the current crowding effect and the heat concentration area in the diode region caused by conduction loss. Finally, a high-precision dedicated positioning fixture is used to align and install the temperature sensor at the predetermined position jointly determined by simulation and experiment, and a thermally conductive adhesive with high thermal conductivity is used for filling and fixing.

3. The method for monitoring the real-time junction temperature of a multi-chip parallel power module according to claim 1, characterized in that, The real-time acquisition process of the fiber optic demodulator in step 2 uses a high-speed demodulator to synchronously acquire the optical signals of all temperature sensors to ensure that the rapid temperature changes caused by switching transients can be captured. Wavelength division multiplexing is used to allocate an independent wavelength channel for each temperature sensor, thereby transmitting multiple signals in a single optical fiber and accurately distinguishing their sources. The demodulator integrates a tunable Fabry-Perot filter, which uses piezoelectric ceramic drive to achieve wavelength scanning and accurately detect the offset of the Bragg wavelength of each temperature sensor due to thermal expansion. During the data acquisition process, the correspondence between the wavelength shift sequence of each temperature sensor and time is recorded in real time. The data structure can be represented as follows: Where n is the number of chips, m is the number of sampling points, and the load current waveform is recorded simultaneously. The expression is With mother-money voltage waveform The expression is ;in For the corresponding Two monitoring locations / parameters for each chip; for The load current value at any given time; for The load voltage value at any given time; for Time of the first Wavelength offset of each temperature sensor; This provides a complete background of electro-thermal coupling data, facilitating subsequent analysis of the relationship between junction temperature and electrical operating conditions.

4. The method for monitoring the real-time junction temperature of a power module with multiple chips in parallel according to claim 1, characterized in that, The process of converting the optical signal into a temperature value in step 3 is achieved by the following formula: ,in, This refers to the change in temperature. This is the temperature sensitivity coefficient; For constant terms, The strain sensitivity coefficient; The change in strain; For sensors that have been specially encapsulated and cured to ensure no relative displacement between them and the measured object, thus making the strain effect negligible, the formula can be simplified to: ,in, This is the real-time temperature value; For reference temperature; The wavelength is measured in real time; Reference wavelength; Based on this transformation relationship, an independent temperature-time series is generated for each temperature sensor. Subsequently, the data were categorized according to chip number and functional area type: for each IGBT chip, its cell temperature sequence was compiled. ) and diode region temperature sequence This ultimately forms a real-time junction temperature dataset for each chip in different functional areas.

5. The method for monitoring the real-time junction temperature of a power module with multiple chips in parallel according to claim 1, characterized in that, The final output of step 4 integrates all analysis results, calculated key parameters, and various curves and charts according to a predefined standard template format to generate a complete structured junction temperature test report.

Images