WO2021006288A1 - Transient thermal characteristic analysis device, analysis method, and program - Google Patents

Transient thermal characteristic analysis device, analysis method, and program Download PDF

Info

Publication number
WO2021006288A1
WO2021006288A1 PCT/JP2020/026665 JP2020026665W WO2021006288A1 WO 2021006288 A1 WO2021006288 A1 WO 2021006288A1 JP 2020026665 W JP2020026665 W JP 2020026665W WO 2021006288 A1 WO2021006288 A1 WO 2021006288A1
Authority
WO
WIPO (PCT)
Prior art keywords
time
series data
time series
transient thermal
power module
Prior art date
Application number
PCT/JP2020/026665
Other languages
French (fr)
Japanese (ja)
Inventor
崇平 福永
舟木　剛
史樹 加藤
Original Assignee
国立大学法人大阪大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 国立大学法人大阪大学 filed Critical 国立大学法人大阪大学
Publication of WO2021006288A1 publication Critical patent/WO2021006288A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N25/00—Investigating or analyzing materials by the use of thermal means
    • G01N25/18—Investigating or analyzing materials by the use of thermal means by investigating thermal conductivity
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/26—Testing of individual semiconductor devices

Definitions

  • the present invention relates to a technique for analyzing transient thermal characteristics of a power module.
  • power modules power semiconductor devices applied to power conversion circuits and power semiconductor modules in which one or more of them are mounted and modularized (hereinafter, collectively referred to as power modules). ) Is becoming more sophisticated. On the other hand, the amount of heat generated has become a problem with such high functionality. For example, during the switching operation of a power conversion circuit, the loss generated in the power module becomes heat, which causes the temperature of the power device and peripheral components such as capacitors and reactors to rise. As a result, if the temperature difference environment between the operation and the stop occurs repeatedly, the joint portion of the power module and the like may deteriorate.
  • the thermal resistance called structural function is calculated from the time-series response (time-series data) of the junction temperature measured using the temperature dependence (temperature parameter) of the electrical characteristics of the power module.
  • a thermal equivalent circuit consisting of a heat capacity is used.
  • the transient thermal characteristics measurement method (JESD51-14) according to the JEDEC (US Joint Electronic Device Commission) standard shown in Non-Patent Document 1. )It has been known.
  • FIG. 8 shows a conventional analysis algorithm based on the JESD51-14 measurement method.
  • this analysis algorithm first, in the transient temperature measurement step, time-series data of the junction temperature of the power module is sampled at equal intervals in a linear time domain, for example, in units of microseconds (step S101). Next, using a moving average digital filter, noise filtering processing in the linear time domain is performed, and noise removal processing included in the time-series data of the junction temperature is performed (step S103). Next, the time-series data of the junction temperature from which noise has been removed is converted into a logarithmic time domain (step S105).
  • the data is resampled by thinning out, interpolation, or the like so that the data is evenly spaced in the logarithmic time domain.
  • numerical differentiation is executed as a preprocessing for identifying the parameters of the thermal circuit model by the deconvolution integral (step S107), and then the thermal time constant spectrum is calculated by the deconvolution integral (step S109). ..
  • the parameters of the Foster circuit as an example of a one-dimensional model of the thermal circuit are identified (step S111), and further, using the identified parameters, the structural function is calculated through the Foster-Cauer transformation (step S113). ).
  • Patent Document 1 the heat sink is arranged in the first and second states, the forward voltage is sampled, and the transient thermal resistance value and its differential value are calculated and displayed from the junction temperature obtained in each state.
  • a device for displaying on the unit has been proposed.
  • Patent Document 1 has a description of converting the measured forward voltage into a junction temperature, a method of removing a noise component contained in the measured voltage is not considered.
  • the present invention has been made in view of the above, and transient heat enables more accurate analysis by removing noise components in the frequency domain with respect to the time series data (profile) of the junction temperature of the power module. It provides a characteristic analysis device, an analysis method, and a program.
  • the transient thermal characteristic analysis device is the transient thermal characteristic analysis device for analyzing the transient thermal characteristics of the power module from time series data which are junction temperatures sampled at linear time equal intervals in the heat dissipation process of the power module.
  • a conversion processing means that resamples time series data at unequal intervals, converts it to logarithmic time, and applies numerical differentiation to the converted time series data, and unequal interval Fourier conversion to the time series data that has undergone numerical differentiation.
  • the data is provided with a Fourier conversion means for performing the above, and a noise removing means for performing noise removing processing in the frequency region on the Fourier transformed time series data.
  • the transient thermal characteristic analysis method is a transient thermal characteristic analysis method for analyzing the transient thermal characteristic of the power module from time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module.
  • the program according to the present invention is a program for analyzing the transient thermal characteristics of the power module by a computer from the time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module.
  • Conversion processing means that resamples at unequal intervals, converts them to logarithmic time, and applies numerical differentiation to the converted time series data
  • Fourier conversion means that applies unequal interval Fourier transformation to the time series data that has undergone numerical differentiation.
  • the computer functions as a noise removing means for performing noise removing processing in the frequency region on the Fourier transformed time series data.
  • time-series data which is the junction temperature sampled at regular intervals of linear time in the heat dissipation process of the power module, is obtained, and the transient thermal characteristics of the power module are analyzed from the time-series data of the junction temperature. That is, the time-series data is resampled at unequal intervals by the conversion processing means, converted into logarithmic time, and the converted time-series data is numerically differentiated. Further, the time series data subjected to the numerical differentiation is subjected to an unequal interval Fourier transform by the Fourier transform means. Then, the noise removing means performs noise removing processing in the frequency domain on the Fourier-transformed time-series data.
  • the time-series data of the junction temperature of the power module is obtained by using the temperature dependence of the electrical characteristics of the power semiconductor device to obtain the voltage and current at the time of measurement. Converted from the value.
  • this noise becomes an error factor in the analysis of the transient thermal characteristics. Therefore, instead of the noise filtering in the time domain for the time series data that has been conventionally used, that is, the obtained time series data is once converted into logarithmic hours at unequal intervals without filtering in the time domain, and numerical values are obtained. After differentiation, Fourier transform is performed at unequal intervals to remove noise in the frequency domain.
  • the influence of noise can be further reduced.
  • the originally measured data can be used instead of the conventional method of applying similar data by thinning out or interpolating the data. You can use all of yourself.
  • a part of the period is extracted, and a filter is applied to appropriately remove noise of frequency characteristics that is unnecessary for analysis of transient thermal characteristics peculiar to the period.
  • noise there is also an aspect. According to the above, it is possible to effectively remove noise such as removal of high frequency components and removal of intermediate frequency components in correspondence with the frequency characteristics of noise components.
  • noise can be reduced on the analysis software side, transient thermal characteristic analysis with high reproducibility becomes possible.
  • the transient thermal characteristics of the power module can be analyzed with high accuracy.
  • FIG. 1 It is a block diagram which shows one Embodiment of the transient thermal characteristic analysis apparatus which concerns on this invention. It is a side sectional view which shows an example of a power module. It is a flowchart which shows the transient thermal characteristic analysis algorithm which concerns on this invention.
  • A shows the numerical differentiation of the time series data of the junction temperature
  • B is the analysis method (Example) according to the present invention. It is a figure explaining the noise reduction effect by the conventional analysis method (comparative example).
  • FIG. 1 is a block diagram showing an embodiment of the transient thermal characteristic analysis device according to the present invention.
  • the transient thermal characteristic analysis device 1 samples the time-series data of the junction temperature of the power module 40, performs necessary noise removal processing, and then analyzes and evaluates the transient thermal characteristic of the structural function.
  • the transient thermal characteristic analyzer 1 is typically processed by a computer.
  • the junction temperature measurement site is the junction with the semiconductor in the power module 40.
  • the computer includes a control unit 10 that performs arithmetic processing by a processor (CPU).
  • the transient thermal characteristic analysis device 1 also includes a display unit 21, a storage unit 101, a power supply 31 on the measurement system side, and a driver 32 for power supply.
  • the control unit 10 is connected to the display unit 21, the storage unit 101, and the driver 32, and is also connectable to the junction unit of the power module 40 to be analyzed connected (set) to the measurement system.
  • the display unit 21 displays an image and is provided as needed.
  • the display unit 21 confirmably displays the content set via the input unit of the illustration such as a touch panel, or displays the analysis content in various modes.
  • the storage unit 101 stores the processing program executed by the processor of the control unit 10, and also stores data necessary for executing the processing program, for example, various numerical calculation formulas, various conversion formulas, various functions, and the temperature of the power module 40. Store parameters (temperature dependence), etc.
  • the storage unit 101 also has a work area for temporarily storing data in the process of processing.
  • the power supply 31 constitutes a power supply circuit capable of outputting a large current for heating that temporarily heats the power module 40 to a predetermined temperature and a minute small current for measuring the junction temperature at the time of heat dissipation or cooling.
  • the power supply 31 may be provided with a large current power supply circuit and a small current power supply circuit separately.
  • the driver 32 is composed of an IGBT (insulated gate type bipolar transistor) or the like for switching between the large current and the small current from the power supply 31 and supplying the power module 40 (junction).
  • the driver 32 is driven according to a drive signal from the control unit 10 as described later, and is controlled to output a predetermined current.
  • the control unit 10 takes in the junction unit of the power module 40, for example, the forward voltage and current of the diode via the A / D conversion unit and the like.
  • the control unit 10 obtains time-series data of the junction temperature from the measured voltage and current using temperature parameters.
  • the power module 40 has a semiconductor package 41 mounted on the heat sink 46 via grease 45 as a heat radiating material.
  • the power module 40 includes a tip 42, a die attach 43 and a die pad 44, typically having a junction on the tip 42.
  • the thermal resistance can be adjusted by the material (thermal conductivity), shape and size of the package 41, and the material, shape and size of each member in the heat dissipation path from the chip 42 to the grease 45. Further, the surface condition of each member can also be adjusted.
  • the lead wire 47 may also be considered as a part of the heat dissipation path.
  • the control unit 10 functions as a power supply drive unit 11, a measurement processing unit 12, a calculation unit 13, and a structural function calculation unit 14 by executing a processing program stored in the storage unit 101 with a processor.
  • the calculation unit 13 includes a conversion processing unit 131, a noise removal unit 132, and a thermal time constant spectrum calculation unit 133.
  • the power supply drive unit 11 controls the current level supplied to the power module 40. Upon receiving the start instruction of the analysis algorithm, the power supply drive unit 11 first causes the driver 32 to output a large current, and when the temperature reaches a predetermined temperature, the power supply drive unit 11 shifts to the transient temperature measurement process to reduce the current to a predetermined level. It is to be switched and output.
  • the measurement processing unit 12 Upon receiving the instruction to start the transient temperature measurement process, the measurement processing unit 12 sets the voltage value from the junction unit of the power module 40 at a predetermined cycle, for example, in units of ⁇ seconds, that is, at equal intervals in the linear time region, to the current value at that time. A / D conversion is performed together with, and the reading is performed periodically.
  • the measurement processing unit 12 samples the time-series data of the junction temperature (stored in the storage unit 101) by converting the read voltage value and the current value at that time using a temperature parameter (FIG. 3). See step S1).
  • the measurement processing unit 12 executes preprocessing for processing by the calculation unit 13. Preprocessing includes, for example, thinning process, offset process (initial data cut process, see FIG. 4A), and transient thermal resistance calculation process.
  • the conversion processing unit 131 executes various conversion processes described later.
  • the conversion processing unit 131 performs a process of resampling the time-series data of the junction temperature sampled at equal intervals in the linear time domain at unequal intervals when converting to the data in the logarithmic time domain ( See step S3 in FIG. 3). By resampling at unequal intervals, all measured time-series data can be used, eliminating data omissions and maintaining accuracy.
  • the conversion processing unit 131 executes numerical differentiation processing on the time series data resampled at unequal intervals in the logarithmic time domain (see step S5 in FIG. 3). Since the conversion processing unit 131 uses the acquired time series data itself for the numerical differentiation processing, the accuracy is maintained, and as a result, higher noise reduction performance can be exhibited.
  • the conversion processing unit 131 performs an unequal-interval discrete Fourier transform on the time-series data subjected to numerical differentiation to convert the time-series data into a frequency domain (see step S7 in FIG. 3). Further, the conversion processing unit 131 performs inverse discrete Fourier transform by performing resampling processing at equal intervals in the logarithmic time domain on the time series data for which noise removal processing has been executed by noise filtering (see step S11 in FIG. 3). ).
  • the noise removing unit 132 noise-filters the time-series data subjected to the non-equidistant discrete Fourier transform in the frequency domain (see step S9 in FIG. 3).
  • the noise filtering is preferably a low-pass filter that effectively removes high-frequency component noise such as the white noise mixed from the power module 40 side and the measurement system side when acquiring junction temperature data. ..
  • the Fermi-Dirac filter function can be adopted.
  • the cooling noise is also removed.
  • the bandpass filter can also be applied as having a high frequency removal (lowpass filter) function.
  • the thermal time constant spectrum calculation unit 133 calculates the thermal time constant spectrum by deconvolution integration with respect to the time constant data subjected to the inverse discrete Fourier transform in step S11 of FIG. 3 (see step S13 of FIG. 3).
  • the structural function calculation unit 14 identifies the parameters of the Foster circuit as an example of a one-dimensional model of the thermal circuit from the calculated thermal time constant spectrum (see step S15 in FIG. 3). Further, the structural function calculation unit 14 calculates the structural function (thermal resistance, heat capacity) through the Foster-Cauer conversion using the identified parameters (see step S17 in FIG. 3). From the contents of the obtained structural function, improvements and matters to be improved for each member can be easily identified.
  • the analysis device T3Ster manufactured by Mentor Graphics is applied to the processing portion common to that in FIG. Further, the data of the comparative examples and the examples shown in FIGS. 4 and 5 are created based on the analysis device T3Ster and by changing the processing procedure of the analysis device T3Ster to the procedure of FIG.
  • the power module 40 to be analyzed is set in the transient thermal characteristic analysis device 1.
  • the measurement processing unit 12 receives an instruction to start the transient temperature measurement process, it reads the voltage and current values from the junction unit of the power module 40 at equal intervals in a linear time region of, for example, 1 ⁇ sec, and reads the temperature by A / D conversion.
  • the time-series data of the junction temperature is sampled by converting using the parameters (step S1).
  • the measurement processing unit 12 then executes preprocessing for processing by the calculation unit 13.
  • the conversion processing unit 131 converts the time-series data of the junction temperature obtained in step S1 into data in the logarithmic time domain (step S3).
  • the time series data is resampled as it is, that is, at unequal intervals in the logarithmic time domain. Since the time series data acquired in this way is used, the accuracy is maintained, and as a result, higher noise reduction performance is exhibited.
  • FIG. 4A shows an example of the numerical differentiation of the time-series data of the junction temperature subjected to the pretreatment.
  • the horizontal axis is the logarithmic time
  • the vertical axis is the normalization level of the numerical differentiation.
  • the conversion processing unit 131 performs unequal-interval discrete Fourier transform on the time-series data subjected to numerical differentiation to convert the time-series data into the frequency domain (step S7).
  • the noise removing unit 132 noise-filters the time-series data subjected to the non-equidistant discrete Fourier transform in the frequency domain (step S9).
  • the noise filtering is a low-pass filter that effectively removes high-frequency component noise such as the white noise mixed from the power module 40 side and the measurement system side when acquiring junction temperature data.
  • the conversion processing unit 131 performs inverse discrete Fourier transform by resampling the time-series data from which noise has been removed by noise filtering at equal intervals in the logarithmic time domain (step S11).
  • FIG. 4 (B) shows an example (example) of the result of performing the inverse discrete Fourier transform in step S11 based on FIG. 4 (A).
  • the comparative example is the one to which the conventional analysis algorithm shown in FIG. 8 is applied, and the numerical differentiation of FIG. 4A is directly subjected to the inverse discrete Fourier transform.
  • the graphic shown in the comparative example has a high frequency component corresponding to the noise component remaining, while the graphic shown in the example is a curve in which the residual high frequency component is hardly observed. That is, it can be seen that the noise component is effectively removed.
  • the vertical axis of FIG. 4B is enlarged as compared with that of FIG. 4A.
  • the thermal time constant spectrum calculation unit 133 calculates the thermal time constant spectrum by deconvolution integration with respect to the time constant response subjected to the inverse discrete Fourier transform in step S11 (step S13).
  • FIG. 5 is a diagram showing a thermal time constant spectrum, and the figure of the thermal time constant spectrum shown in the comparative example is generally gentle, and the level change is particularly small on the initial time side.
  • the figure of the thermal time constant spectrum shown in the examples shows a relatively wavy waveform, and in particular, the peak shows a relatively high level, and a thermal time constant spectrum in which noise is further removed is obtained. Furthermore, since the wavy tendency is remarkable (the level change is large) on the initial time side, it can be seen that the noise is removed more as a whole.
  • the structural function calculation unit 14 identifies the parameters of the Foster circuit as a one-dimensional model example of the thermal circuit from the calculated thermal time constant spectrum (step S15), and subsequently, using the identified parameters. , Foster-Cauer conversion is performed to calculate the structural function (thermal resistance, heat capacity) (step S17). Then, the calculated structural function of each member is displayed on the display unit 21.
  • FIG. 7 are numerical values when the sample noise of each level is applied to the analysis device T3Ster manufactured by Mentor Graphics, and the numerical values in the comparative example are based on the implementation of the algorithm in FIG. It is based on the implementation of the algorithm of FIG.
  • the example can remove the noise by about one digit to several times as much as the noise of any level.
  • the actual noise level is considered to be about -40 dB (noise level: 1.0%), but even for noise about 5 times (-26 dB (noise level: 5.0%)), a higher noise removal effect is achieved. Has been obtained.
  • the noise level (amplitude) is used as a parameter, and the data in which uniform noise due to random numbers is superimposed is targeted for analysis.
  • the present invention can include the following aspects.
  • the power module to be analyzed may include not only power modules but also general semiconductor devices in which heat generation is a problem, and LEDs for light emission.
  • the transient thermal characteristic analysis device 1 may be divided into a configuration including the structural function calculation unit 14 and a configuration up to the thermal time constant spectrum calculation and a configuration for performing the structural function calculation.
  • the transient thermal characteristic analysis device 1 does not have to have all of the above-mentioned parts, but has a mode including up to the calculation unit 13 or a mode having at least a configuration up to the noise removing unit 132 of the calculation unit 13. In these cases, the remaining parts up to the structural function calculation unit 14 may be made executable by a separate device.
  • the transient thermal characteristic analysis device may adopt the following aspects. That is, in the transient thermal characteristic analyzer that analyzes the transient thermal characteristics of the power module from the time series data that are the junction temperatures sampled at regular intervals in linear time in the heat dissipation process of the power module, the time series data is converted into logarithmic hours.
  • a conversion processing means that applies numerical differentiation to the converted time-series data
  • a Fourier transform means that performs Fourier transform to the time-series data that has undergone numerical differentiation
  • a frequency region for the Fourier-transformed time-series data It may be provided with a noise removing means for performing the noise removing processing in the above. In this way, by performing the noise removal processing in the frequency domain on the time series data, it is possible to calculate the thermal time constant spectrum with higher accuracy than that of the conventional apparatus.
  • the conversion processing means resamples the time series data at unequal intervals, converts the time series data into logarithmic hours, and applies numerical differentiation to the converted time series data, and the Fourier transform means. May be subjected to an unequal interval Fourier transform on the time series data subjected to the numerical differentiation.
  • the logarithmic and Fourier transformed time series data are set at unequal intervals, more accurate analysis of transient thermal characteristics can be expected as compared with the case of resampling at equal intervals.
  • the present invention relates to a transient thermal characteristic analyzer that analyzes the transient thermal characteristics of the power module from time series data which are junction temperatures sampled at linear time equal intervals in the heat dissipation process of the power module.
  • a conversion processing means that resamples the series data at unequal intervals, converts it to logarithmic time, and applies numerical differentiation to the converted time series data, and an unequal interval Fourier transform to the time series data that has undergone numerical differentiation. It is preferable to provide the Fourier transform means for performing the Fourier transform and the noise removing means for performing the noise removing processing in the frequency region on the Fourier transformed time series data.
  • the present invention does not use the time-series data in the transient thermal characteristic analysis method for analyzing the transient thermal characteristics of the power module from the time-series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module.
  • the time series data are unequally spaced.
  • time-series data of junction temperatures sampled at linear time equal intervals is obtained in the heat dissipation process of the power module, and the transient thermal characteristics of the power module are analyzed from the time-series data of the junction temperature. That is, the time-series data is resampled at unequal intervals by the conversion processing means, converted into logarithmic time, and the converted time-series data is numerically differentiated. Further, the time series data subjected to the numerical differentiation is subjected to an unequal interval Fourier transform by the Fourier transform means. Then, the noise removing means performs noise removing processing in the frequency domain on the Fourier-transformed time-series data.
  • the time-series data of the junction temperature of the power module is obtained by using the temperature dependence of the electrical characteristics of the power semiconductor device to obtain the voltage and current at the time of measurement. Converted from the value.
  • this noise becomes an error factor in the analysis of the transient thermal characteristics. Therefore, instead of the noise filtering in the time domain for the time series data that has been conventionally used, that is, the obtained time series data is once converted into logarithmic hours at unequal intervals without filtering in the time domain, and numerical values are obtained.
  • the Fourier transform is performed at unequal intervals to remove the noise in the frequency domain so that the influence of the noise can be further reduced.
  • the originally measured data can be used instead of the conventional method of applying similar data by thinning out or interpolating the data. You can use all of yourself.
  • a part of the period is extracted, and a filter is applied to appropriately remove noise of frequency characteristics that is unnecessary for analysis of transient thermal characteristics peculiar to the period.
  • noise can be reduced on the analysis software side, transient thermal characteristic analysis with high reproducibility becomes possible.
  • the noise removing means removes high frequency components in the frequency domain. Further, it is particularly preferable that the noise removing means is a low-pass filter. According to such a configuration, it is possible to selectively remove a frequency component, for example, a high frequency component according to the characteristics of noise superimposed when measuring time series data, and in that case, a simple low-pass filter can be applied. preferable.
  • the present invention includes an inverse Fourier transform means that resamples the time series data processed by the noise removing means at equal intervals in logarithmic time and performs an inverse discrete Fourier transform, and a time series data processed by the inverse Fourier transform means. It is preferable to provide a calculation means for calculating the thermal time constant spectrum by performing deconvolution integration. According to this configuration, by removing noise in the frequency domain, it is possible to reduce the influence of noise on the thermal time constant spectrum required for calculating the thermal equivalent circuit parameters. Further, since the measured time series data is data sampled at equal intervals in the linear time domain, the sampling intervals are unequal in the logarithmic time domain.
  • the inverse Fourier transform was used to perform resampling at regular intervals for logarithmic hours. In this way, by obtaining the thermal time constant spectrum in which noise is further suppressed, the structural function can be calculated with high accuracy.
  • Transient thermal characteristic analyzer 10 Control unit 12 Measurement processing unit 13 Calculation unit 131 Conversion processing unit (conversion processing means, Fourier transform means, inverse Fourier transform means) 132 Noise removal unit (noise removal means) 133 Thermal time constant spectrum calculation unit (calculation means) 14 Structural function calculation unit 40 Power module

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Testing Of Individual Semiconductor Devices (AREA)
  • Investigating Or Analyzing Materials Using Thermal Means (AREA)

Abstract

A transient thermal characteristic analysis device (1) analyzes transient thermal characteristics of a power module (40) on the basis of time-sequential data which is junction temperatures sampled at an equal linear time interval in a heat radiation process of the power module (40). The transient thermal characteristic analysis device (1) is provided with: a conversion processing unit (131) that re-samples the time-sequential data at unequal intervals, converts the time-sequential data into logarithmic time, and executes numerical differentiation on the converted time-sequential data; a conversion processing unit (131) that executes unequal interval Fourier transformation on the numerically differentiated time-sequential data; and a noise removal unit (132) that carries out a noise removal process in a frequency region for the Fourier transformed time-sequential data. Accordingly, the transient thermal characteristics of the power module is analyzed more accurately.

Description

過渡熱特性解析装置、解析方法及びプログラムTransient thermal characteristic analyzer, analysis method and program
　本発明は、パワーモジュールの過渡熱特性を解析する技術に関する。 The present invention relates to a technique for analyzing transient thermal characteristics of a power module.
　近年、車載用、電力系用等のパワーエレクトロニクス分野において、電力変換回路に適用されるパワー半導体デバイス及びこれを１又は複数個搭載してモジュール化したパワー半導体モジュール（以下、まとめてパワーモジュールという。）の高機能化が進んでいる。一方、かかる高機能化に伴って発熱量が問題となっている。例えば、電力変換回路のスイッチング動作時に、パワーモジュールにおいて生じる損失が熱となり、パワーデバイス及び周辺部品であるコンデンサやリアクトル等の温度上昇を招く。この結果、動作時と停止時との間の温度差環境が繰り返し発生すると、パワーモジュールの接合部等が劣化する虞がある。また、近年では、負荷急変時や故障時の過電流に起因する瞬時的な発熱を考慮したパワーモジュールの過渡熱特性に対する設計及び評価も求められている。さらに、パワーモジュールのパッケージ技術の進歩に伴って熱特性が改善されて熱等価回路パラメータの値が小さくなっており、より高い精度での解析、評価手法が求められている。 In recent years, in the field of power electronics such as for automobiles and electric power systems, power semiconductor devices applied to power conversion circuits and power semiconductor modules in which one or more of them are mounted and modularized (hereinafter, collectively referred to as power modules). ) Is becoming more sophisticated. On the other hand, the amount of heat generated has become a problem with such high functionality. For example, during the switching operation of a power conversion circuit, the loss generated in the power module becomes heat, which causes the temperature of the power device and peripheral components such as capacitors and reactors to rise. As a result, if the temperature difference environment between the operation and the stop occurs repeatedly, the joint portion of the power module and the like may deteriorate. Further, in recent years, there has been a demand for design and evaluation of transient thermal characteristics of power modules in consideration of instantaneous heat generation due to overcurrent at the time of sudden load change or failure. Further, with the progress of the packaging technology of the power module, the thermal characteristics are improved and the value of the thermal equivalent circuit parameter is reduced, and an analysis and evaluation method with higher accuracy is required.
　過渡熱特性の解析では、パワーモジュールの電気的特性の温度依存性（温度パラメータ）を用いて測定されるジャンクション温度の時系列応答（時系列データ）から算出される、構造関数と呼ばれる熱抵抗と熱容量とからなる熱等価回路を用いる。逆畳み込み積分を用いて過渡熱特性を表す熱等価回路のパラメータを算出する方法として、非特許文献１に示す、JEDEC（米国合同電子デバイス委員会）の規格による過渡熱特性測定法（JESD51-14）が知られている。 In the analysis of transient thermal characteristics, the thermal resistance called structural function is calculated from the time-series response (time-series data) of the junction temperature measured using the temperature dependence (temperature parameter) of the electrical characteristics of the power module. A thermal equivalent circuit consisting of a heat capacity is used. As a method for calculating the parameters of the thermal equivalent circuit representing the transient thermal characteristics using the deconvolution integral, the transient thermal characteristics measurement method (JESD51-14) according to the JEDEC (US Joint Electronic Device Commission) standard shown in Non-Patent Document 1. )It has been known.
　図８は、JESD51-14測定法に基づく従来の解析アルゴリズムを示す。この解析アルゴリズムでは、まず過渡温度測定工程において、パワーモジュールのジャンクション温度の時系列データが、例えばμ秒単位で、線形時間領域で等間隔にサンプリングされる（ステップＳ１０１）。次いで、移動平均デジタルフィルタを用いて、線形時間領域におけるノイズフィルタリング処理が施されて、ジャンクション温度の時系列データに含まれるノイズの除去処理が行われる（ステップＳ１０３）。次いで、ノイズが除去されたジャンクション温度の時系列データに対して対数時間領域への変換処理が行われる（ステップＳ１０５）。データの変換は、対数時間領域で等間隔となるように、データが間引きや補間等によってリサンプリングされる。そして、逆畳み込み積分により熱回路モデルのパラメータを同定するための前処理としての数値微分が実行され（ステップＳ１０７）、続いて、逆畳み込み積分による熱時定数スペクトルの算出が行われる（ステップＳ１０９）。この後、熱回路の一次元モデル例としてのFoster回路のパラメータが同定され（ステップＳ１１１）、さらに、同定されたパラメータを用いて、Foster-Cauer変換を経て構造関数の算出が行われる（ステップＳ１１３）。 FIG. 8 shows a conventional analysis algorithm based on the JESD51-14 measurement method. In this analysis algorithm, first, in the transient temperature measurement step, time-series data of the junction temperature of the power module is sampled at equal intervals in a linear time domain, for example, in units of microseconds (step S101). Next, using a moving average digital filter, noise filtering processing in the linear time domain is performed, and noise removal processing included in the time-series data of the junction temperature is performed (step S103). Next, the time-series data of the junction temperature from which noise has been removed is converted into a logarithmic time domain (step S105). In the data conversion, the data is resampled by thinning out, interpolation, or the like so that the data is evenly spaced in the logarithmic time domain. Then, numerical differentiation is executed as a preprocessing for identifying the parameters of the thermal circuit model by the deconvolution integral (step S107), and then the thermal time constant spectrum is calculated by the deconvolution integral (step S109). .. After that, the parameters of the Foster circuit as an example of a one-dimensional model of the thermal circuit are identified (step S111), and further, using the identified parameters, the structural function is calculated through the Foster-Cauer transformation (step S113). ).
　また、特許文献１には、ヒートシンクを第１、第２の状態に配置して順方向電圧をサンプリングし、各状態で得られたジャンクション温度から過渡熱抵抗値及びその微分値を算出して表示部に表示する装置が提案されている。 Further, in Patent Document 1, the heat sink is arranged in the first and second states, the forward voltage is sampled, and the transient thermal resistance value and its differential value are calculated and displayed from the junction temperature obtained in each state. A device for displaying on the unit has been proposed.
特開２０１９－１５５６４号公報JP-A-2019-15564
　最近では、パワーモジュールのパッケージ技術の進歩によって熱特性が改善され、熱等価回路パラメータの値が小さくなっているため、従来の手法では、要求される精度での評価が困難になってきている。JEDECの規格による解析方法では、具体的な数値処理アルゴリズムは定義されていない。また、当該規格が適用されたMentor Graphics 社製の解析装置T3Ster、及びその解析プログラムであるT3SterMasterによれば、ノイズ除去は、電気的に得られたジャンクション温度の時系列データに対して時間領域で移動平均フィルタリング処理を施したに過ぎない。そのため、ホワイトノイズ等の測定ノイズや量子化誤差が必ずしも効果的に除去できるとはいえない。このような場合、熱等価回路のパラメータの算出においてノイズが相対的に顕在化してしまう。また、ノイズ除去処理で除去しきれなかったノイズ成分が数値微分によって結果的に顕在化してしまう虞もある。さらに、特許文献１に記載の装置は、測定した順方向電圧をジャンクション温度に換算する記載はあるものの、測定電圧に含まれるノイズ成分を除去する方法は考慮されていない。 Recently, due to advances in power module packaging technology, thermal characteristics have been improved and the values of thermal equivalent circuit parameters have become smaller, making it difficult to evaluate with the required accuracy using conventional methods. No specific numerical processing algorithm is defined in the analysis method according to the JEDEC standard. In addition, according to the analysis device T3Ster manufactured by Mentor Graphics, to which the standard is applied, and its analysis program T3SterMaster, noise removal is performed in the time domain with respect to the time-series data of the junction temperature obtained electrically. It is just a moving average filtering process. Therefore, it cannot always be said that measurement noise such as white noise and quantization error can be effectively removed. In such a case, noise becomes relatively apparent in the calculation of the parameters of the thermal equivalent circuit. In addition, there is a possibility that noise components that cannot be completely removed by the noise removal process will eventually become apparent due to numerical differentiation. Further, although the apparatus described in Patent Document 1 has a description of converting the measured forward voltage into a junction temperature, a method of removing a noise component contained in the measured voltage is not considered.
　本発明は、上記に鑑みてなされたもので、パワーモジュールのジャンクション温度の時系列データ（プロファイル）に対して周波数領域でノイズ成分を除去することで、より精度の高い解析を可能にする過渡熱特性解析装置、解析方法及びプログラムを提供するものである。 The present invention has been made in view of the above, and transient heat enables more accurate analysis by removing noise components in the frequency domain with respect to the time series data (profile) of the junction temperature of the power module. It provides a characteristic analysis device, an analysis method, and a program.
　本発明に係る過渡熱特性解析装置は、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析装置において、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段と、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段と、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段とを備えたものである。 The transient thermal characteristic analysis device according to the present invention is the transient thermal characteristic analysis device for analyzing the transient thermal characteristics of the power module from time series data which are junction temperatures sampled at linear time equal intervals in the heat dissipation process of the power module. A conversion processing means that resamples time series data at unequal intervals, converts it to logarithmic time, and applies numerical differentiation to the converted time series data, and unequal interval Fourier conversion to the time series data that has undergone numerical differentiation. The data is provided with a Fourier conversion means for performing the above, and a noise removing means for performing noise removing processing in the frequency region on the Fourier transformed time series data.
　また、本発明に係る過渡熱特性解析方法は、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析方法において、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換工程と、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換工程と、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去工程とを備えたものである。 Further, the transient thermal characteristic analysis method according to the present invention is a transient thermal characteristic analysis method for analyzing the transient thermal characteristic of the power module from time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module. , The conversion step of resampling the time series data at unequal intervals, converting it to logarithmic time, and applying numerical differentiation to the converted time series data, and unequal interval Fourier on the time series data to which the numerical differentiation has been applied. It includes a Fourier conversion step of performing conversion and a noise removing step of performing noise removal processing in the frequency region on the Fourier-transformed time-series data.
　また、本発明に係るプログラムは、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性をコンピュータにより解析するプログラムにおいて、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段、及び前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段として、前記コンピュータを機能させるものである。 Further, the program according to the present invention is a program for analyzing the transient thermal characteristics of the power module by a computer from the time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module. Conversion processing means that resamples at unequal intervals, converts them to logarithmic time, and applies numerical differentiation to the converted time series data, and Fourier conversion means that applies unequal interval Fourier transformation to the time series data that has undergone numerical differentiation. The computer functions as a noise removing means for performing noise removing processing in the frequency region on the Fourier transformed time series data.
　これらによれば、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データが得られ、ジャンクション温度の時系列データからパワーモジュールの過渡熱特性の解析が行われる。すなわち、前記時系列データは、変換処理手段により、不等間隔でリサンプリングされて対数時間に変換され、さらに変換された時系列データに数値微分が施される。また、前記数値微分が施された時系列データは、フーリエ変換手段により、不等間隔フーリエ変換が施される。そして、ノイズ除去手段により、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理が行われる。 According to these, time-series data, which is the junction temperature sampled at regular intervals of linear time in the heat dissipation process of the power module, is obtained, and the transient thermal characteristics of the power module are analyzed from the time-series data of the junction temperature. That is, the time-series data is resampled at unequal intervals by the conversion processing means, converted into logarithmic time, and the converted time-series data is numerically differentiated. Further, the time series data subjected to the numerical differentiation is subjected to an unequal interval Fourier transform by the Fourier transform means. Then, the noise removing means performs noise removing processing in the frequency domain on the Fourier-transformed time-series data.
　例えば、パワーモジュールの過渡熱特性を表す熱等価回路パラメータを求めるために、パワー半導体デバイスの電気特性の温度依存性を利用してパワーモジュールのジャンクション温度の時系列データは、測定時の電圧、電流値から換算される。このとき、電圧、電流値にノイズが重畳すると、このノイズが過渡熱特性の解析において誤差要因となる。そこで、従来用いられていた時系列データに対する時間領域でのノイズフィルタリングの代わりに、すなわち、得られた時系列データを時間領域でフィルタリングせずに、一旦不等間隔で対数時間に変換し、数値微分した後に不等間隔でフーリエ変換を行って、周波数領域においてノイズを除去する。このようにしてノイズの影響を一層低減可能にする。測定された時系列データの線形時間領域から対数時間への変換を不等間隔で行うことで、従来のようなデータの間引きや補間による類似データを適用する方法に代えて、元々測定されたデータ自身を全て使用可能としている。また、測定された時系列データのうちの全域の他、一部期間を抽出し、当該期間に特有な過渡熱特性の解析に対して不要となる周波数特性のノイズを適宜除去するフィルタを適用する態様もある。以上によれば、ノイズ成分の周波数特性に対応させて、例えば高周波成分の除去とか中間周波数成分の除去とかの効果的なノイズ除去が可能となる。しかも、解析ソフトウェア側でノイズ低減が可能となるため、再現性の高い過渡熱特性解析が可能となる。 For example, in order to obtain the thermal equivalent circuit parameters that represent the transient thermal characteristics of a power module, the time-series data of the junction temperature of the power module is obtained by using the temperature dependence of the electrical characteristics of the power semiconductor device to obtain the voltage and current at the time of measurement. Converted from the value. At this time, if noise is superimposed on the voltage and current values, this noise becomes an error factor in the analysis of the transient thermal characteristics. Therefore, instead of the noise filtering in the time domain for the time series data that has been conventionally used, that is, the obtained time series data is once converted into logarithmic hours at unequal intervals without filtering in the time domain, and numerical values are obtained. After differentiation, Fourier transform is performed at unequal intervals to remove noise in the frequency domain. In this way, the influence of noise can be further reduced. By converting the measured time series data from the linear time domain to logarithmic time at unequal intervals, the originally measured data can be used instead of the conventional method of applying similar data by thinning out or interpolating the data. You can use all of yourself. In addition to the entire area of the measured time-series data, a part of the period is extracted, and a filter is applied to appropriately remove noise of frequency characteristics that is unnecessary for analysis of transient thermal characteristics peculiar to the period. There is also an aspect. According to the above, it is possible to effectively remove noise such as removal of high frequency components and removal of intermediate frequency components in correspondence with the frequency characteristics of noise components. Moreover, since noise can be reduced on the analysis software side, transient thermal characteristic analysis with high reproducibility becomes possible.
　本発明によれば、パワーモジュールの過渡熱特性を高精度で解析することができる。 According to the present invention, the transient thermal characteristics of the power module can be analyzed with high accuracy.
本発明に係る過渡熱特性解析装置の一実施形態を示すブロック図である。It is a block diagram which shows one Embodiment of the transient thermal characteristic analysis apparatus which concerns on this invention. パワーモジュールの一例を示す側面断面図である。It is a side sectional view which shows an example of a power module. 本発明に係る過渡熱特性解析アルゴリズムを示すフローチャートである。It is a flowchart which shows the transient thermal characteristic analysis algorithm which concerns on this invention. 対数時間軸でのジャンクション温度の時系列データの微分値を示すタイムチャートで、（Ａ）はジャンクション温度の時系列データの数値微分を示し、（Ｂ）は本発明に係る解析法（実施例）と従来の解析法（比較例）によるノイズ低減効果を説明する図である。In a time chart showing the differential value of the time series data of the junction temperature on the logarithmic time axis, (A) shows the numerical differentiation of the time series data of the junction temperature, and (B) is the analysis method (Example) according to the present invention. It is a figure explaining the noise reduction effect by the conventional analysis method (comparative example). 対数時間軸での熱等価回路パラメータの算出に用いる、実施例と比較例における熱時定数スペクトルを示す図である。It is a figure which shows the thermal time constant spectrum in an Example and a comparative example used for the calculation of a thermal equivalent circuit parameter on a logarithmic time axis. ジャンクション温度の時系列データ（ideal）に重畳する各レベルに対応したサンプルとしての乱数ノイズのパワースペクトルを示す図である。It is a figure which shows the power spectrum of the random number noise as a sample corresponding to each level superposed on the time series data (ideal) of a junction temperature. 図６に示すサンプルノイズを重畳した場合の、実施例と比較例とにおけるノイズ低減効果の違いを示す図表で、数値は真値からのRMS（Root Mean Square）値を示している。In the chart showing the difference in noise reduction effect between the example and the comparative example when the sample noise shown in FIG. 6 is superimposed, the numerical value shows the RMS (Root Mean Square) value from the true value. JESD51-14測定法に基づく従来の過渡熱特性解析アルゴリズムを示すフローチャートである。It is a flowchart which shows the conventional transient thermal characteristic analysis algorithm based on JESD51-14 measurement method.
　図１は、本発明に係る過渡熱特性解析装置の一実施形態を示すブロック図である。過渡熱特性解析装置１は、パワーモジュール４０のジャンクション温度の時系列データをサンプリングし、必要なノイズ除去処理を施した後、構造関数の過渡熱特性を解析、評価する。過渡熱特性解析装置１は、典型的にはコンピュータで処理される。なお、ジャンクション温度の測定部位は、パワーモジュール４０内の半導体との接合部である。また、コンピュータは、プロセッサ（ＣＰＵ）で演算処理を行う制御部１０を備えている。 FIG. 1 is a block diagram showing an embodiment of the transient thermal characteristic analysis device according to the present invention. The transient thermal characteristic analysis device 1 samples the time-series data of the junction temperature of the power module 40, performs necessary noise removal processing, and then analyzes and evaluates the transient thermal characteristic of the structural function. The transient thermal characteristic analyzer 1 is typically processed by a computer. The junction temperature measurement site is the junction with the semiconductor in the power module 40. Further, the computer includes a control unit 10 that performs arithmetic processing by a processor (CPU).
　過渡熱特性解析装置１は、また、表示部２１、記憶部１０１、さらに測定系側の電源３１及び給電用のドライバ３２を備えている。制御部１０は、表示部２１、記憶部１０１及びドライバ３２と接続されると共に、測定系に接続（セット）された解析対象のパワーモジュール４０のジャンクション部と接続可能にされている。 The transient thermal characteristic analysis device 1 also includes a display unit 21, a storage unit 101, a power supply 31 on the measurement system side, and a driver 32 for power supply. The control unit 10 is connected to the display unit 21, the storage unit 101, and the driver 32, and is also connectable to the junction unit of the power module 40 to be analyzed connected (set) to the measurement system.
　表示部２１は、画像を表示するもので、必要に応じて設けられる。表示部２１は、例えばタッチパネル等の図略の入力部を介して設定される内容を確認的に表示したり、解析内容を種々の態様で表示したりする。記憶部１０１は、制御部１０のプロセッサが実行する処理プログラムを記憶すると共に、処理プログラムを実行する上で必要なデータ類、例えば各種数値計算式、各種変換式、各種関数、パワーモジュール４０の温度パラメータ（温度依存性）等を記憶する。記憶部１０１は、また、処理途中のデータを一時的に記憶するワークエリアを備えている。 The display unit 21 displays an image and is provided as needed. The display unit 21 confirmably displays the content set via the input unit of the illustration such as a touch panel, or displays the analysis content in various modes. The storage unit 101 stores the processing program executed by the processor of the control unit 10, and also stores data necessary for executing the processing program, for example, various numerical calculation formulas, various conversion formulas, various functions, and the temperature of the power module 40. Store parameters (temperature dependence), etc. The storage unit 101 also has a work area for temporarily storing data in the process of processing.
　電源３１は、パワーモジュール４０を所定温度に一旦加熱する加熱用の大電流及び放熱乃至冷却時でのジャンクション温度計測用の微小な小電流が出力可能な電源回路を構成している。電源３１は、大電流用電源回路と小電流用電源回路とを個別に設けた態様でもよい。ドライバ３２は、電源３１から前記大電流用と小電流用とを切り替えてパワーモジュール４０（の接合部）に供給するためのIGBT（絶縁ゲート型バイポーラトランジスタ）等からなる。ドライバ３２は、後述するように制御部１０からの駆動信号に従って駆動され、所定の電流を出力するように制御される。 The power supply 31 constitutes a power supply circuit capable of outputting a large current for heating that temporarily heats the power module 40 to a predetermined temperature and a minute small current for measuring the junction temperature at the time of heat dissipation or cooling. The power supply 31 may be provided with a large current power supply circuit and a small current power supply circuit separately. The driver 32 is composed of an IGBT (insulated gate type bipolar transistor) or the like for switching between the large current and the small current from the power supply 31 and supplying the power module 40 (junction). The driver 32 is driven according to a drive signal from the control unit 10 as described later, and is controlled to output a predetermined current.
　制御部１０は、パワーモジュール４０のジャンクション部、例えばダイオードの順方向電圧及び電流をＡ／Ｄ変換部等を経由して取り込む。制御部１０は、計測した電圧及び電流から温度パラメータを用いてジャンクション温度の時系列データを得る。 The control unit 10 takes in the junction unit of the power module 40, for example, the forward voltage and current of the diode via the A / D conversion unit and the like. The control unit 10 obtains time-series data of the junction temperature from the measured voltage and current using temperature parameters.
　ここで、図２を用いて本実施形態に係る解析対象のパワーモジュール４０の構成を説明する。パワーモジュール４０は、本実施形態では、半導体パッケージ４１が放熱材としてのグリース４５を介してヒートシンク４６上に搭載されている。パワーモジュール４０は、チップ４２、ダイアタッチ４３及びダイパッド４４を備え、典型的にはチップ４２上にジャンクション部がある。パッケージ４１の材質（熱伝導性）、形状及びサイズ、さらにチップ４２からグリース４５までの放熱経路における各部材の材質、形状及びサイズによって熱抵抗が調整可能となる。また、各部材の表面状態も調整対象となり得る。なお、リード線４７も放熱経路の一部として考慮すべき場合もある。 Here, the configuration of the power module 40 to be analyzed according to the present embodiment will be described with reference to FIG. In the present embodiment, the power module 40 has a semiconductor package 41 mounted on the heat sink 46 via grease 45 as a heat radiating material. The power module 40 includes a tip 42, a die attach 43 and a die pad 44, typically having a junction on the tip 42. The thermal resistance can be adjusted by the material (thermal conductivity), shape and size of the package 41, and the material, shape and size of each member in the heat dissipation path from the chip 42 to the grease 45. Further, the surface condition of each member can also be adjusted. The lead wire 47 may also be considered as a part of the heat dissipation path.
　制御部１０は、記憶部１０１に記憶されている処理プログラムをプロセッサで実行することで、電源駆動部１１、測定処理部１２、演算部１３及び構造関数算出部１４として機能する。演算部１３は、変換処理部１３１、ノイズ除去部１３２及び熱時定数スペクトル算出部１３３を備えている。 The control unit 10 functions as a power supply drive unit 11, a measurement processing unit 12, a calculation unit 13, and a structural function calculation unit 14 by executing a processing program stored in the storage unit 101 with a processor. The calculation unit 13 includes a conversion processing unit 131, a noise removal unit 132, and a thermal time constant spectrum calculation unit 133.
　電源駆動部１１は、パワーモジュール４０に供給する電流レベルを制御する。電源駆動部１１は、解析アルゴリズムの開始指示を受けると、ドライバ３２に対して、まず大電流を出力させ、所定温度に達した時点で過渡温度測定工程に移行して、所定レベルの小電流に切り替えて出力させるものである。 The power supply drive unit 11 controls the current level supplied to the power module 40. Upon receiving the start instruction of the analysis algorithm, the power supply drive unit 11 first causes the driver 32 to output a large current, and when the temperature reaches a predetermined temperature, the power supply drive unit 11 shifts to the transient temperature measurement process to reduce the current to a predetermined level. It is to be switched and output.
　以降、測定処理部１２から構造関数算出部１４までの説明を、図３に示す解析アルゴリズムを参照して行う。測定処理部１２は、過渡温度測定工程の開始指示を受けると、所定周期、例えばμ秒単位で、すなわち線形時間領域で等間隔にパワーモジュール４０のジャンクション部からの電圧値をその時点の電流値と共にＡ／Ｄ変換して周期的に読み取る。測定処理部１２は、読み取った電圧値及びその時点の電流値を温度パラメータを用いて変換することでジャンクション温度の時系列データのサンプリング（記憶部１０１に記憶）を行うものである（図３のステップＳ１参照）。また、測定処理部１２は、演算部１３による処理のための前処理を実行する。前処理としては、例えば間引き処理、オフセット処理（初期データカット処理、図４（Ａ）参照）、過渡熱抵抗の計算処理がある。 Hereinafter, the explanations from the measurement processing unit 12 to the structural function calculation unit 14 will be given with reference to the analysis algorithm shown in FIG. Upon receiving the instruction to start the transient temperature measurement process, the measurement processing unit 12 sets the voltage value from the junction unit of the power module 40 at a predetermined cycle, for example, in units of μ seconds, that is, at equal intervals in the linear time region, to the current value at that time. A / D conversion is performed together with, and the reading is performed periodically. The measurement processing unit 12 samples the time-series data of the junction temperature (stored in the storage unit 101) by converting the read voltage value and the current value at that time using a temperature parameter (FIG. 3). See step S1). In addition, the measurement processing unit 12 executes preprocessing for processing by the calculation unit 13. Preprocessing includes, for example, thinning process, offset process (initial data cut process, see FIG. 4A), and transient thermal resistance calculation process.
　変換処理部１３１は、後述する種々の変換処理を実行する。本実施形態では、変換処理部１３１は、線形時間領域で等間隔にサンプリングされたジャンクション温度の時系列データを、対数時間領域のデータに変換する際に不等間隔でリサンプリングする処理を行う（図３のステップＳ３参照）。不等間隔でリサンプリングすることで、測定した時系列データを全て用いることができてデータの欠落をなくし、精度を維持している。また、変換処理部１３１は、対数時間領域で不等間隔にリサンプリングされた時系列データに対して数値微分処理を実行する（図３のステップＳ５参照）。なお、変換処理部１３１は、取得した時系列データそのものを数値微分処理に使用するので、精度が維持され、その結果、より高いノイズ低減性能が発揮可能となる。 The conversion processing unit 131 executes various conversion processes described later. In the present embodiment, the conversion processing unit 131 performs a process of resampling the time-series data of the junction temperature sampled at equal intervals in the linear time domain at unequal intervals when converting to the data in the logarithmic time domain ( See step S3 in FIG. 3). By resampling at unequal intervals, all measured time-series data can be used, eliminating data omissions and maintaining accuracy. Further, the conversion processing unit 131 executes numerical differentiation processing on the time series data resampled at unequal intervals in the logarithmic time domain (see step S5 in FIG. 3). Since the conversion processing unit 131 uses the acquired time series data itself for the numerical differentiation processing, the accuracy is maintained, and as a result, higher noise reduction performance can be exhibited.
　また、変換処理部１３１は、数値微分が施された時系列データに対して不等間隔離散フーリエ変換を行って、時系列データを周波数領域に変換する（図３のステップＳ７参照）。さらに、変換処理部１３１は、ノイズフィルタリングでノイズ除去処理が実行された時系列データに対して対数時間領域で等間隔にリサンプリング処理を施して逆離散フーリエ変換を行う（図３のステップＳ１１参照）。 Further, the conversion processing unit 131 performs an unequal-interval discrete Fourier transform on the time-series data subjected to numerical differentiation to convert the time-series data into a frequency domain (see step S7 in FIG. 3). Further, the conversion processing unit 131 performs inverse discrete Fourier transform by performing resampling processing at equal intervals in the logarithmic time domain on the time series data for which noise removal processing has been executed by noise filtering (see step S11 in FIG. 3). ).
　ノイズ除去部１３２は、不等間隔離散フーリエ変換が施された時系列データを周波数領域においてノイズフィルタリングするものである（図３のステップＳ９参照）。ノイズフィルタリングは、本実施形態では、ジャンクション温度データを取得する時に、パワーモジュール４０側及び測定系側から混入する前記したホワイトノイズ等の高周波成分ノイズを効果的に除去するローパスフィルタであることが好ましい。例えば、Fermi-Diracフィルタ関数が採用可能である。また、測定時に冷却器を採用する態様では、冷却ノイズも除去対象とすることが好ましい。なお、バンドパスフィルタも高周波除去（ローパスフィルタ）機能を備えるものとして適用可能である。 The noise removing unit 132 noise-filters the time-series data subjected to the non-equidistant discrete Fourier transform in the frequency domain (see step S9 in FIG. 3). In the present embodiment, the noise filtering is preferably a low-pass filter that effectively removes high-frequency component noise such as the white noise mixed from the power module 40 side and the measurement system side when acquiring junction temperature data. .. For example, the Fermi-Dirac filter function can be adopted. Further, in the embodiment in which the cooler is adopted at the time of measurement, it is preferable that the cooling noise is also removed. The bandpass filter can also be applied as having a high frequency removal (lowpass filter) function.
　熱時定数スペクトル算出部１３３は、図３のステップＳ１１で逆離散フーリエ変換された時定数データに対して、逆畳み込み積分による熱時定数スペクトルの算出を行う（図３のステップＳ１３参照）。 The thermal time constant spectrum calculation unit 133 calculates the thermal time constant spectrum by deconvolution integration with respect to the time constant data subjected to the inverse discrete Fourier transform in step S11 of FIG. 3 (see step S13 of FIG. 3).
　構造関数算出部１４は、算出された熱時定数スペクトルから、熱回路の一次元モデル例としてのFoster回路のパラメータの同定を行う（図３のステップＳ１５参照）。また、構造関数算出部１４は、同定されたパラメータを用いて、Foster-Cauer変換を経て構造関数（熱抵抗、熱容量）の算出を行う（図３のステップＳ１７参照）。得られた構造関数の内容から各部材に対する改善、改良すべき事項が容易に識別可能となる。 The structural function calculation unit 14 identifies the parameters of the Foster circuit as an example of a one-dimensional model of the thermal circuit from the calculated thermal time constant spectrum (see step S15 in FIG. 3). Further, the structural function calculation unit 14 calculates the structural function (thermal resistance, heat capacity) through the Foster-Cauer conversion using the identified parameters (see step S17 in FIG. 3). From the contents of the obtained structural function, improvements and matters to be improved for each member can be easily identified.
　続いて、図３のフローチャートを用いて本発明に係る解析アルゴリズムを説明する。 Subsequently, the analysis algorithm according to the present invention will be described with reference to the flowchart of FIG.
　なお、図３において、図８と共通する処理部分はMentor Graphics 社製の解析装置T3Sterを適用している。また、図４、図５に示す比較例及び実施例の各データは、解析装置T3Sterに基づいて、及び解析装置T3Sterの処理手順を図３の手順に変更して適用して作成されている。 Note that in FIG. 3, the analysis device T3Ster manufactured by Mentor Graphics is applied to the processing portion common to that in FIG. Further, the data of the comparative examples and the examples shown in FIGS. 4 and 5 are created based on the analysis device T3Ster and by changing the processing procedure of the analysis device T3Ster to the procedure of FIG.
　図３において、先ず、解析対象のパワーモジュール４０が過渡熱特性解析装置１にセットされる。測定処理部１２は、過渡温度測定工程の開始指示を受けると、例えば１μ秒という線形時間領域で等間隔にパワーモジュール４０のジャンクション部からの電圧及び電流値をＡ／Ｄ変換して読み取り、温度パラメータを用いて変換することでジャンクション温度の時系列データのサンプリングを行う（ステップＳ１）。また、測定処理部１２は、次いで演算部１３による処理のための前処理を実行する。 In FIG. 3, first, the power module 40 to be analyzed is set in the transient thermal characteristic analysis device 1. When the measurement processing unit 12 receives an instruction to start the transient temperature measurement process, it reads the voltage and current values from the junction unit of the power module 40 at equal intervals in a linear time region of, for example, 1 μsec, and reads the temperature by A / D conversion. The time-series data of the junction temperature is sampled by converting using the parameters (step S1). In addition, the measurement processing unit 12 then executes preprocessing for processing by the calculation unit 13.
　続いて、変換処理部１３１は、ステップＳ１で得られたジャンクション温度の時系列データを対数時間領域のデータに変換する（ステップＳ３）。このとき、時系列データは、そのままで、すなわち対数時間領域において不等間隔のままでリサンプリングされる。このように取得した時系列データそのものを使用するので、精度が維持され、その結果、より高いノイズ低減性能が発揮される。 Subsequently, the conversion processing unit 131 converts the time-series data of the junction temperature obtained in step S1 into data in the logarithmic time domain (step S3). At this time, the time series data is resampled as it is, that is, at unequal intervals in the logarithmic time domain. Since the time series data acquired in this way is used, the accuracy is maintained, and as a result, higher noise reduction performance is exhibited.
　また、変換処理部１３１は、対数時間領域で不等間隔にリサンプリングされた時系列データに対して数値微分処理を実行する（ステップＳ５）。図４（Ａ）には、前記前処理が施されたジャンクション温度の時系列データの数値微分の一例が示されている。図４では、横軸は対数時間、縦軸は数値微分の正規化レベルである。 Further, the conversion processing unit 131 executes numerical differentiation processing on the time series data resampled at unequal intervals in the logarithmic time domain (step S5). FIG. 4A shows an example of the numerical differentiation of the time-series data of the junction temperature subjected to the pretreatment. In FIG. 4, the horizontal axis is the logarithmic time, and the vertical axis is the normalization level of the numerical differentiation.
　次いで、変換処理部１３１は、数値微分が施された時系列データに対して不等間隔離散フーリエ変換を行って、時系列データを周波数領域に変換する（ステップＳ７）。 Next, the conversion processing unit 131 performs unequal-interval discrete Fourier transform on the time-series data subjected to numerical differentiation to convert the time-series data into the frequency domain (step S7).
　次いで、ノイズ除去部１３２は、不等間隔離散フーリエ変換が施された時系列データを周波数領域においてノイズフィルタリングする（ステップＳ９）。ノイズフィルタリングは、本実施形態では、ジャンクション温度データを取得する時に、パワーモジュール４０側及び測定系側から混入する前記したホワイトノイズ等の高周波成分ノイズを効果的に除去するローパスフィルタである。 Next, the noise removing unit 132 noise-filters the time-series data subjected to the non-equidistant discrete Fourier transform in the frequency domain (step S9). In the present embodiment, the noise filtering is a low-pass filter that effectively removes high-frequency component noise such as the white noise mixed from the power module 40 side and the measurement system side when acquiring junction temperature data.
　次いで、変換処理部１３１は、ノイズフィルタリングでノイズが除去された時系列データに対して対数時間領域で等間隔にリサンプリング処理を施して逆離散フーリエ変換を行う（ステップＳ１１）。 Next, the conversion processing unit 131 performs inverse discrete Fourier transform by resampling the time-series data from which noise has been removed by noise filtering at equal intervals in the logarithmic time domain (step S11).
　図４（Ｂ）は、図４（Ａ）を元にして、ステップＳ１１で逆離散フーリエ変換を行った結果の一例（実施例）を示している。なお、図４（Ｂ）において、比較例は、図８に示す従来の解析アルゴリズムを適用したもので、図４（Ａ）の数値微分をそのまま逆離散フーリエ変換したものである。図４（Ｂ）から分かるように、比較例で示す図形にはノイズ成分に相当する高周波成分が残留している一方、実施例で示す図形は、高周波成分の残留が殆ど見られない曲線であり、すなわちノイズ成分が効果的に除去されていることが分かる。なお、図４（Ｂ）の縦軸は、同図（Ａ）に比して拡大されている。 FIG. 4 (B) shows an example (example) of the result of performing the inverse discrete Fourier transform in step S11 based on FIG. 4 (A). In addition, in FIG. 4B, the comparative example is the one to which the conventional analysis algorithm shown in FIG. 8 is applied, and the numerical differentiation of FIG. 4A is directly subjected to the inverse discrete Fourier transform. As can be seen from FIG. 4B, the graphic shown in the comparative example has a high frequency component corresponding to the noise component remaining, while the graphic shown in the example is a curve in which the residual high frequency component is hardly observed. That is, it can be seen that the noise component is effectively removed. The vertical axis of FIG. 4B is enlarged as compared with that of FIG. 4A.
　次いで、熱時定数スペクトル算出部１３３は、ステップＳ１１で逆離散フーリエ変換された時定数応答に対して、逆畳み込み積分による熱時定数スペクトルの算出を行う（ステップＳ１３）。図５は、熱時定数スペクトルを示す図で、比較例で示す熱時定数スペクトルの図形は全体的になだらかであり、特に初期時間側でのレベル変化も小さい。一方、実施例で示す熱時定数スペクトルの図形は相対的に波打ち波形が見られ、特にピークが相対的に高レベルを示し、ノイズがより除去された熱時定数スペクトルが得られている。さらに初期時間側で波打ち傾向が顕著（レベル変化が大きい）となっていることから全体的にもノイズがより除去されていることが分かる。 Next, the thermal time constant spectrum calculation unit 133 calculates the thermal time constant spectrum by deconvolution integration with respect to the time constant response subjected to the inverse discrete Fourier transform in step S11 (step S13). FIG. 5 is a diagram showing a thermal time constant spectrum, and the figure of the thermal time constant spectrum shown in the comparative example is generally gentle, and the level change is particularly small on the initial time side. On the other hand, the figure of the thermal time constant spectrum shown in the examples shows a relatively wavy waveform, and in particular, the peak shows a relatively high level, and a thermal time constant spectrum in which noise is further removed is obtained. Furthermore, since the wavy tendency is remarkable (the level change is large) on the initial time side, it can be seen that the noise is removed more as a whole.
　次いで、構造関数算出部１４は、算出された熱時定数スペクトルから、熱回路の一次元モデル例としてのFoster回路のパラメータの同定を行い（ステップＳ１５）、続いて、同定されたパラメータを用いて、Foster-Cauer変換を経て構造関数（熱抵抗、熱容量）の算出を行う（ステップＳ１７）。そして、算出された各部材の構造関数は表示部２１に表示される。 Next, the structural function calculation unit 14 identifies the parameters of the Foster circuit as a one-dimensional model example of the thermal circuit from the calculated thermal time constant spectrum (step S15), and subsequently, using the identified parameters. , Foster-Cauer conversion is performed to calculate the structural function (thermal resistance, heat capacity) (step S17). Then, the calculated structural function of each member is displayed on the display unit 21.
　次に、図６、図７により、実施例と比較例によるノイズ低減効果の違いの一例をサンプルノイズを用いて説明する。図７の数値は、各レベルのサンプルノイズをMentor Graphics 社製の解析装置T3Sterに適用した場合の数値であり、比較例の数値は、図８のアルゴリズムの実施に基づき、実施例の数値は、図３のアルゴリズムの実施に基づくものである。 Next, with reference to FIGS. 6 and 7, an example of the difference in noise reduction effect between the examples and the comparative examples will be described using sample noise. The numerical values in FIG. 7 are numerical values when the sample noise of each level is applied to the analysis device T3Ster manufactured by Mentor Graphics, and the numerical values in the comparative example are based on the implementation of the algorithm in FIG. It is based on the implementation of the algorithm of FIG.
　図６では、ジャンクション温度の時系列データ（真値；ideal）のパワースペクトルに、例えば、-60dB（ノイズレベル：0.1%），-46dB（ノイズレベル：0.5%），-40dB（ノイズレベル：1.0%），-26dB（ノイズレベル：5.0%）のレベルのサンプル乱数ノイズのパワースペクトルを重畳している。図６に示す各ノイズを重畳した場合の、真値からのRMS（Root Mean Square）値は、比較例が、-60dB側から、順番に、2.24×10－４，5.93×10－４，1.08×10－3，1.01×10－2，であるのに対して、実施例が、25.02×10－5，2.40×10－４，3.39×10－４，8.02×10－3である。このように、いずれのレベルのノイズに対しても、実施例の方が、略１桁乃至数倍程度、ノイズの除去ができていることが認められる。特に、実際のノイズレベルは、-40dB（ノイズレベル：1.0%）程度と考えられるが、それよりも５倍（-26dB（ノイズレベル：5.0%））程度のノイズについても、より高いノイズ除去効果が得られている。 In FIG. 6, in the power spectrum of the time series data (true value; ideal) of the junction temperature, for example, -60 dB (noise level: 0.1%), -46 dB (noise level: 0.5%), -40 dB (noise level: 1.0). %), -26dB (noise level: 5.0%) level sample random noise power spectrum is superimposed. The RMS (Root Mean Square) value from the true value when each noise shown in Fig. 6 is superimposed is 2.24 × 10 -4 , 5.93 × 10 -4 , 1.08 in order from the -60dB side in the comparative example. × 10 -3, whereas it is 1.01 × 10 -2,, embodiment, 25.02 × 10 -5, 2.40 × 10 -4, 3.39 × 10 -4, a 8.02 × 10 -3. As described above, it is recognized that the example can remove the noise by about one digit to several times as much as the noise of any level. In particular, the actual noise level is considered to be about -40 dB (noise level: 1.0%), but even for noise about 5 times (-26 dB (noise level: 5.0%)), a higher noise removal effect is achieved. Has been obtained.
　なお、図７に示すデータは、以下のパラメータ（Ｒ１＝6.63×10－３，Ｃ１＝1.24×10－２、Ｒ２＝1.35×10－３，Ｃ２＝6.77×10－２、Ｒ３＝5.98×10－３，Ｃ３＝2.22×10－２）を仮定した３段Foster回路から数値計算された、１μ秒サンプリングを仮定した温度応答データをidealとしている。このデータに対し、ノイズレベル（振幅）をパラメータとして、乱数による一様ノイズを重畳したデータに対して解析を対象としている。 The data shown in FIG. 7 includes the following parameters (R 1 = 6.63 × 10 -3 , C 1 = 1.24 × 10 -2 , R 2 = 1.35 × 10 -3 , C 2 = 6.77 × 10 -2 , R. The temperature response data assuming 1 μsampling, which is numerically calculated from the 3-stage Foster circuit assuming 3 = 5.98 × 10 -3 , C 3 = 2.22 × 10 -2 ), is used as ideal. For this data, the noise level (amplitude) is used as a parameter, and the data in which uniform noise due to random numbers is superimposed is targeted for analysis.
　本発明は、以下の態様を含むことができる。 The present invention can include the following aspects.
（１）解析対象のパワーモジュールには、電力変換用の他、発熱が問題となるような半導体デバイスの全般、さらに発光用ＬＥＤを含めてもよい。 (1) The power module to be analyzed may include not only power modules but also general semiconductor devices in which heat generation is a problem, and LEDs for light emission.
（２）過渡熱特性解析装置１は、構造関数算出部１４までを含める態様の他、熱時定数スペクトル算出までの構成と、構造関数算出を行う構成とに分けた態様としてもよい。 (2) The transient thermal characteristic analysis device 1 may be divided into a configuration including the structural function calculation unit 14 and a configuration up to the thermal time constant spectrum calculation and a configuration for performing the structural function calculation.
（３）過渡熱特性解析装置１は、前記各部を全て備えた態様である必要はなく、演算部１３までを備えた態様、あるいは演算部１３のノイズ除去部１３２までの構成を少なくとも備えた態様であってもよく、これらの場合、構造関数算出部１４までの残りの各部は別個の装置で実行可能にすればよい。 (3) The transient thermal characteristic analysis device 1 does not have to have all of the above-mentioned parts, but has a mode including up to the calculation unit 13 or a mode having at least a configuration up to the noise removing unit 132 of the calculation unit 13. In these cases, the remaining parts up to the structural function calculation unit 14 may be made executable by a separate device.
（４）さらに、過渡熱特性解析装置は、以下の態様を採用してもよい。すなわち、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析装置において、前記時系列データを対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段と、前記数値微分が施された時系列データにフーリエ変換を施すフーリエ変換手段と、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段とを備えたものでもよい。このように、時系列データに対して周波数領域でのノイズ除去処理を行うことで、従来装置に比してより精度の高い熱時定数スペクトルを算出することが可能となる。 (4) Further, the transient thermal characteristic analysis device may adopt the following aspects. That is, in the transient thermal characteristic analyzer that analyzes the transient thermal characteristics of the power module from the time series data that are the junction temperatures sampled at regular intervals in linear time in the heat dissipation process of the power module, the time series data is converted into logarithmic hours. , A conversion processing means that applies numerical differentiation to the converted time-series data, a Fourier transform means that performs Fourier transform to the time-series data that has undergone numerical differentiation, and a frequency region for the Fourier-transformed time-series data. It may be provided with a noise removing means for performing the noise removing processing in the above. In this way, by performing the noise removal processing in the frequency domain on the time series data, it is possible to calculate the thermal time constant spectrum with higher accuracy than that of the conventional apparatus.
　そして、前記の構成において、前記変換処理手段を、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施すものとし、前記フーリエ変換手段を、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すものとしてもよい。この場合、対数及びフーリエ変換された時系列データを不等間隔とすれば、等間隔にリサンプリングする場合に比して、より精度の高い過渡熱特性の解析が期待できる。 Then, in the above configuration, the conversion processing means resamples the time series data at unequal intervals, converts the time series data into logarithmic hours, and applies numerical differentiation to the converted time series data, and the Fourier transform means. May be subjected to an unequal interval Fourier transform on the time series data subjected to the numerical differentiation. In this case, if the logarithmic and Fourier transformed time series data are set at unequal intervals, more accurate analysis of transient thermal characteristics can be expected as compared with the case of resampling at equal intervals.
　以上説明したように、本発明は、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析装置において、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段と、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段と、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段とを備えることが好ましい。 As described above, the present invention relates to a transient thermal characteristic analyzer that analyzes the transient thermal characteristics of the power module from time series data which are junction temperatures sampled at linear time equal intervals in the heat dissipation process of the power module. A conversion processing means that resamples the series data at unequal intervals, converts it to logarithmic time, and applies numerical differentiation to the converted time series data, and an unequal interval Fourier transform to the time series data that has undergone numerical differentiation. It is preferable to provide the Fourier transform means for performing the Fourier transform and the noise removing means for performing the noise removing processing in the frequency region on the Fourier transformed time series data.
　また、本発明は、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析方法において、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換工程と、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換工程と、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去工程とを備えることが好ましい。 Further, the present invention does not use the time-series data in the transient thermal characteristic analysis method for analyzing the transient thermal characteristics of the power module from the time-series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module. A conversion step of resampling at equal intervals, converting to logarithmic time, and applying numerical differentiation to the converted time series data, and a Fourier conversion step of applying unequal interval Fourier transformation to the time series data subjected to the numerical differentiation. It is preferable to include a noise removing step of performing noise removing processing in the frequency region on the Fourier-transformed time series data.
　また、本発明は、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性をコンピュータにより解析するプログラムにおいて、前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段、前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段、及び前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段として、前記コンピュータを機能させることが好ましい。 Further, according to the present invention, in a program for analyzing the transient thermal characteristics of the power module by a computer from the time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module, the time series data are unequally spaced. A conversion processing means for resampling and converting to logarithmic time and performing numerical differentiation on the converted time series data, a Fourier conversion means for performing unequal interval Fourier conversion on the time series data subjected to the numerical differentiation, and the above. It is preferable to make the computer function as a noise removing means for performing noise removing processing in the frequency region on the Fourier-transformed time series data.
　これらの発明によれば、パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データが得られ、ジャンクション温度の時系列データからパワーモジュールの過渡熱特性の解析が行われる。すなわち、前記時系列データは、変換処理手段により、不等間隔でリサンプリングされて対数時間に変換され、さらに変換された時系列データに数値微分が施される。また、前記数値微分が施された時系列データは、フーリエ変換手段により、不等間隔フーリエ変換が施される。そして、ノイズ除去手段により、前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理が行われる。 According to these inventions, time-series data of junction temperatures sampled at linear time equal intervals is obtained in the heat dissipation process of the power module, and the transient thermal characteristics of the power module are analyzed from the time-series data of the junction temperature. That is, the time-series data is resampled at unequal intervals by the conversion processing means, converted into logarithmic time, and the converted time-series data is numerically differentiated. Further, the time series data subjected to the numerical differentiation is subjected to an unequal interval Fourier transform by the Fourier transform means. Then, the noise removing means performs noise removing processing in the frequency domain on the Fourier-transformed time-series data.
　例えば、パワーモジュールの過渡熱特性を表す熱等価回路パラメータを求めるために、パワー半導体デバイスの電気特性の温度依存性を利用してパワーモジュールのジャンクション温度の時系列データは、測定時の電圧、電流値から換算される。このとき、電圧、電流値にノイズが重畳すると、このノイズが過渡熱特性の解析において誤差要因となる。そこで、従来用いられていた時系列データに対する時間領域でのノイズフィルタリングの代わりに、すなわち、得られた時系列データを時間領域でフィルタリングせずに、一旦不等間隔で対数時間に変換し、数値微分した後に不等間隔でフーリエ変換を行って、周波数領域においてノイズを除去するようにしてノイズの影響を一層低減可能にする。測定された時系列データの線形時間領域から対数時間への変換を不等間隔で行うことで、従来のようなデータの間引きや補間による類似データを適用する方法に代えて、元々測定されたデータ自身を全て使用可能としている。また、測定された時系列データのうちの全域の他、一部期間を抽出し、当該期間に特有な過渡熱特性の解析に対して不要となる周波数特性のノイズを適宜除去するフィルタを適用する態様もある。以上によれば、ノイズ成分の周波数特性に対応させて、例えば高周波成分の除去とか中間周波数成分の除去とかの効果的なノイズ除去が可能となる。しかも、解析ソフトウェア側でノイズ低減が可能となるため、再現性の高い過渡熱特性解析が可能となる。 For example, in order to obtain the thermal equivalent circuit parameters that represent the transient thermal characteristics of a power module, the time-series data of the junction temperature of the power module is obtained by using the temperature dependence of the electrical characteristics of the power semiconductor device to obtain the voltage and current at the time of measurement. Converted from the value. At this time, if noise is superimposed on the voltage and current values, this noise becomes an error factor in the analysis of the transient thermal characteristics. Therefore, instead of the noise filtering in the time domain for the time series data that has been conventionally used, that is, the obtained time series data is once converted into logarithmic hours at unequal intervals without filtering in the time domain, and numerical values are obtained. After the differentiation, the Fourier transform is performed at unequal intervals to remove the noise in the frequency domain so that the influence of the noise can be further reduced. By converting the measured time series data from the linear time domain to logarithmic time at unequal intervals, the originally measured data can be used instead of the conventional method of applying similar data by thinning out or interpolating the data. You can use all of yourself. In addition to the entire area of the measured time-series data, a part of the period is extracted, and a filter is applied to appropriately remove noise of frequency characteristics that is unnecessary for analysis of transient thermal characteristics peculiar to the period. There is also an aspect. According to the above, it is possible to effectively remove noise such as removal of high frequency components and removal of intermediate frequency components in correspondence with the frequency characteristics of noise components. Moreover, since noise can be reduced on the analysis software side, transient thermal characteristic analysis with high reproducibility becomes possible.
　また、前記ノイズ除去手段は、周波数領域における高周波数成分を除去するものであることが好ましい。また、特に前記ノイズ除去手段は、ローパスフィルタであることが好ましい。かかる構成によれば、時系列データの測定時に重畳するノイズの特性に応じた周波数成分、例えば高周波成分を選択的に除去することが可能となり、その場合、簡易なローパスフィルタが適用されることが好ましい。 Further, it is preferable that the noise removing means removes high frequency components in the frequency domain. Further, it is particularly preferable that the noise removing means is a low-pass filter. According to such a configuration, it is possible to selectively remove a frequency component, for example, a high frequency component according to the characteristics of noise superimposed when measuring time series data, and in that case, a simple low-pass filter can be applied. preferable.
　また、本発明は、前記ノイズ除去手段で処理された時系列データを対数時間等間隔でリサンプリングして逆離散フーリエ変換する逆フーリエ変換手段と、前記逆フーリエ変換手段で処理された時系列データに、逆畳み込み積分を施して熱時定数スペクトルを算出する算出手段とを備えることが好ましい。この構成によれば、周波数領域においてノイズを除去することで、熱等価回路パラメータの算出に必要な熱時定数スペクトルに対するノイズの影響を低減可能にしている。また、測定した時系列データは、線形時間領域において等間隔でサンプリングされたデータであるため、対数時間領域ではサンプリング間隔が不等間隔となる。そこで、線形時間領域において不等間隔フーリエ変換を適用することで、線形時間でのリサンプリングによる元データの欠落を抑制した。さらに、逆フーリエ変換で、対数時間等間隔でのリサンプリングを行った。このように、ノイズがより抑制された熱時定数スペクトルを得ることで、構造関数を高い精度で算出することが可能となる。 Further, the present invention includes an inverse Fourier transform means that resamples the time series data processed by the noise removing means at equal intervals in logarithmic time and performs an inverse discrete Fourier transform, and a time series data processed by the inverse Fourier transform means. It is preferable to provide a calculation means for calculating the thermal time constant spectrum by performing deconvolution integration. According to this configuration, by removing noise in the frequency domain, it is possible to reduce the influence of noise on the thermal time constant spectrum required for calculating the thermal equivalent circuit parameters. Further, since the measured time series data is data sampled at equal intervals in the linear time domain, the sampling intervals are unequal in the logarithmic time domain. Therefore, by applying the unequal interval Fourier transform in the linear time domain, the loss of the original data due to resampling in the linear time was suppressed. Furthermore, the inverse Fourier transform was used to perform resampling at regular intervals for logarithmic hours. In this way, by obtaining the thermal time constant spectrum in which noise is further suppressed, the structural function can be calculated with high accuracy.
　１　過渡熱特性解析装置
　１０　制御部
　１２　測定処理部
　１３　演算部
　１３１　変換処理部（変換処理手段、フーリエ変換手段、逆フーリエ変換手段）
　１３２　ノイズ除去部（ノイズ除去手段）
　１３３　熱時定数スペクトル算出部（算出手段）
　１４　構造関数算出部
　４０　パワーモジュール 1 Transient thermal characteristic analyzer 10 Control unit 12 Measurement processing unit 13 Calculation unit 131 Conversion processing unit (conversion processing means, Fourier transform means, inverse Fourier transform means)
132 Noise removal unit (noise removal means)
133 Thermal time constant spectrum calculation unit (calculation means)
14 Structural function calculation unit 40 Power module

Claims (6)

  1. 　パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析装置において、
    　前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段と、
    　前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段と、
    　前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段とを備えた過渡熱特性解析装置。     In a transient thermal characteristic analyzer that analyzes the transient thermal characteristics of the power module from time-series data that is the junction temperature sampled at regular intervals in linear time in the heat dissipation process of the power module.
    A conversion processing means that resamples the time series data at unequal intervals, converts it into logarithmic time, and performs numerical differentiation on the converted time series data.
    A Fourier transform means for performing an unequal interval Fourier transform on the time series data subjected to the numerical differentiation,
    A transient thermal characteristic analysis device including a noise removing means for performing noise removing processing in a frequency domain on the Fourier-transformed time series data.
  2. 　前記ノイズ除去手段は、周波数領域における高周波数成分を除去する請求項１に記載の過渡熱特性解析装置。 The transient thermal characteristic analysis device according to claim 1, wherein the noise removing means removes high frequency components in the frequency domain.
  3. 　前記ノイズ除去手段は、ローパスフィルタである請求項２に記載の過渡熱特性解析装置。 The transient thermal characteristic analysis device according to claim 2, wherein the noise removing means is a low-pass filter.
  4. 　前記ノイズ除去手段で処理された時系列データを対数時間等間隔でリサンプリングして逆離散フーリエ変換する逆フーリエ変換手段と、
    　前記逆フーリエ変換手段で処理された時系列データに、逆畳み込み積分を施して熱時定数スペクトルを算出する算出手段とを備えた請求項１～３のいずれかに記載の過渡熱特性解析装置。     An inverse Fourier transform means that resamples the time series data processed by the noise removing means at logarithmic time equal intervals and performs an inverse discrete Fourier transform.
    The transient thermal characteristic analysis device according to any one of claims 1 to 3, further comprising a calculation means for calculating a thermal time constant spectrum by performing deconvolution integration on the time series data processed by the inverse Fourier transform means.
  5. 　パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性を解析する過渡熱特性解析方法において、
    　前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換工程と、
    　前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換工程と、
    　前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去工程とを備えた過渡熱特性解析方法。     In the transient thermal characteristic analysis method for analyzing the transient thermal characteristic of the power module from the time series data which is the junction temperature sampled at linear time equal intervals in the heat dissipation process of the power module.
    A conversion process in which the time series data is resampled at unequal intervals, converted to logarithmic time, and the converted time series data is numerically differentiated.
    A Fourier transform step of performing an unequal interval Fourier transform on the time series data subjected to the numerical differentiation,
    A transient thermal characteristic analysis method including a noise removal step of performing noise removal processing in the frequency domain on the Fourier-transformed time series data.
  6. 　パワーモジュールの放熱過程において線形時間等間隔でサンプリングしたジャンクション温度である時系列データから前記パワーモジュールの過渡熱特性をコンピュータにより解析するプログラムにおいて、
    　前記時系列データを不等間隔でリサンプリングして対数時間に変換し、変換された時系列データに数値微分を施す変換処理手段、
    　前記数値微分が施された時系列データに不等間隔フーリエ変換を施すフーリエ変換手段、及び
    　前記フーリエ変換された時系列データに対して周波数領域でのノイズ除去処理を行うノイズ除去手段として、前記コンピュータを機能させるプログラム。     In a program that uses a computer to analyze the transient thermal characteristics of the power module from time-series data that is the junction temperature sampled at regular intervals in linear time during the heat dissipation process of the power module.
    A conversion processing means that resamples the time series data at unequal intervals, converts it into logarithmic time, and performs numerical differentiation on the converted time series data.
    The computer as a Fourier transform means for performing an unequal interval Fourier transform on the time series data subjected to the numerical differentiation, and a noise removing means for performing a noise removing process on the Fourier transformed time series data in the frequency domain. A program that makes the function work.
PCT/JP2020/026665 2019-07-09 2020-07-08 Transient thermal characteristic analysis device, analysis method, and program WO2021006288A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2019127344A JP7241349B2 (en) 2019-07-09 2019-07-09 Transient thermal characteristic analysis device, analysis method and program
JP2019-127344 2019-07-09

Publications (1)

Publication Number Publication Date
WO2021006288A1 true WO2021006288A1 (en) 2021-01-14

Family

ID=74114838

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2020/026665 WO2021006288A1 (en) 2019-07-09 2020-07-08 Transient thermal characteristic analysis device, analysis method, and program

Country Status (2)

Country Link
JP (1) JP7241349B2 (en)
WO (1) WO2021006288A1 (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003194755A (en) * 2001-12-27 2003-07-09 Mitsumi Electric Co Ltd Method of testing semiconductor package
US20140079092A1 (en) * 2012-09-19 2014-03-20 Abb Oy Method and apparatus for pre-emptive power semiconductor module fault indication
CN104569065A (en) * 2015-02-13 2015-04-29 重庆大学 Rapid evaluation method for cooling property of solid crystal layer of high-power LED apparatus
JP2019015564A (en) * 2017-07-05 2019-01-31 新日本無線株式会社 Thermal resistance measuring device and thermal resistance measuring method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003194755A (en) * 2001-12-27 2003-07-09 Mitsumi Electric Co Ltd Method of testing semiconductor package
US20140079092A1 (en) * 2012-09-19 2014-03-20 Abb Oy Method and apparatus for pre-emptive power semiconductor module fault indication
CN104569065A (en) * 2015-02-13 2015-04-29 重庆大学 Rapid evaluation method for cooling property of solid crystal layer of high-power LED apparatus
JP2019015564A (en) * 2017-07-05 2019-01-31 新日本無線株式会社 Thermal resistance measuring device and thermal resistance measuring method

Also Published As

Publication number Publication date
JP2021012144A (en) 2021-02-04
JP7241349B2 (en) 2023-03-17

Similar Documents

Publication Publication Date Title
Dusmez et al. Remaining useful lifetime estimation for power MOSFETs under thermal stress with RANSAC outlier removal
Musallam et al. An efficient implementation of the rainflow counting algorithm for life consumption estimation
CN106443401B (en) Device and method for testing temperature rise and thermal resistance composition of power MOS device
JP2005507557A (en) Method and apparatus for semiconductor testing
EP3220537B1 (en) Motor control apparatus, motor control method, control system, information processing program and recording medium
CN104321659B (en) Method and device for testing electric energy accumulation systems for driving vehicles
KR102467129B1 (en) State analysis device, state analysis method, and program
CN111505475B (en) Method and device for calibrating parameters of electric heating model of power semiconductor module
JP2017135673A (en) Signal processor, method of controlling signal processor, control program, and record medium
WO2021006288A1 (en) Transient thermal characteristic analysis device, analysis method, and program
CN114829956A (en) Method and power unit for estimating parameters of a junction of a power semiconductor element
CN107725450B (en) Fan control method and system and electrical equipment with cooling fan
CN116345877A (en) Dead zone compensation method of switching tube and related equipment
CN110446910A (en) The method of multi-wafer temperature control equipment and the temperature for controlling multi-wafer power module
JP2019060792A (en) Diagnostic circuit of power conversion unit and diagnostic system
CN112287510A (en) System and method for modeling thermal circuits
JP5742642B2 (en) Semiconductor element junction temperature estimation method, estimation system, and estimation program
JP4703850B2 (en) Temperature control of electronic devices using power following feedback action
CN115575795A (en) Aging test device and method for circuit board card
JP7379820B2 (en) Temperature calculation program for power conversion equipment
CN101174149A (en) Establishment method of control specification limit
JP2582906B2 (en) Method for measuring DC current / voltage characteristics of semiconductor device
CN114172438A (en) Permanent magnet synchronous motor control method and related equipment
JP2020048276A (en) Diagnosis circuit and diagnosis system for power conversion device
CN117148092B (en) Test method and device for accelerating bipolar degradation of SiC MOSFET

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20837375

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20837375

Country of ref document: EP

Kind code of ref document: A1