A method for processing the performance stability of power semiconductor device IGBT

Abstract

The present invention relates to a method for processing the performance stability of a power semiconductor device IGBT, and belongs to the technical field of semiconductor device manufacturing and processing. The present invention ensures the performance stability of the power semiconductor device IGBT, and the processing process is simple and easy to operate. When the method of the present invention processes the performance stability of the power semiconductor device IGBT, the distance percentage data set uses the distance between the performance parameter detection value and the nominal value or the mean, so that when obtaining the stability state of the power semiconductor device IGBT, there will be no situation where the state value cannot be obtained. When each detection value is equal and equal to the nominal value or the mean, the stability state can still be obtained, and it is used to process the performance stability of the power semiconductor device IGBT. The accuracy, operability and applicability of the method are increased, and the overall stability of processing the power semiconductor device IGBT is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of semiconductor device manufacturing and processing, and in particular relates to a method for processing the performance stability of a power semiconductor device IGBT. Background Art

[0002] Semiconductor devices are widely used in today's society, especially power semiconductor devices IGBT, which are widely used in automobiles, energy, and rail transportation. As a result, the performance requirements and demand for power semiconductor devices IGBT are becoming increasingly higher. The production of semiconductor devices IGBT with high performance and high quantity requirements makes the stability of power semiconductor devices IGBT during the manufacturing and production process particularly important.

[0003] In reality, power semiconductor devices IGBTs that achieve the same function often encounter problems with the performance stability of power semiconductor devices of the same model manufactured by multiple manufacturers, or even by the same manufacturer. There are a lot of studies on the manufacturing and processing stability of semiconductor devices, such as the patent applications with announcement numbers CN115708190A and CN1716520A. Summary of the invention

[0004] In view of the above problems, the present invention provides a method for processing the performance stability of a power semiconductor device IGBT, which solves the problem of ensuring the performance stability of the power semiconductor device IGBT, and the processing process is simple and easy to operate.

[0005] The present invention provides a method for processing the performance stability of a power semiconductor device IGBT, and the specific steps are:

[0006] Step 1, sampling a plurality of power semiconductor devices IGBT; obtaining a parameter detection value of each power semiconductor device IGBT; obtaining a parameter detection absolute distance data set of the power semiconductor device IGBT based on the parameter detection value; and obtaining a parameter constraint range of the power semiconductor device IGBT;

[0007] Step 2, processing the parameter detection difference data set and parameter constraint range of the power semiconductor device IGBT in step 1 to obtain a distance percentage data set;

[0008] Step 3: Process the distance percentage data set to obtain the mean of the distance percentage data set;

[0009] Step 4: Process the mean of the distance percentage data set to obtain the distance percentage standard deviation;

[0010] Step 5, processing the mean and standard deviation of the distance percentage data set to obtain the stability state of a single performance parameter of the power semiconductor device IGBT;

[0011] Step 6: Process the stability state of the performance parameters of the power semiconductor device IGBT to determine the stability of the power semiconductor device IGBT.

[0012] Optionally, the expression of the absolute distance of parameter detection of the power semiconductor device IGBT is:

[0013] ;

[0014] in, Indicates i The absolute distance of parameter detection of a power semiconductor device IGBT is the absolute value of the difference between the parameter detection value and the nominal value or the detection mean value; Indicates i Parameter detection value of a power semiconductor device IGBT, i =1,2,…, n , n Indicates the total number of power semiconductor devices IGBT; Indicates nominal value; Represents the detection mean.

[0015] Optionally, the parameter constraint range of the power semiconductor device IGBT is expressed as:

[0016]

[0017] in, L Indicates the parameter constraint range of the power semiconductor device IGBT; P max Indicates the nominal maximum value; P min Indicates the nominal minimum value; x max Indicates the maximum value of parameter detection; x min Indicates the minimum value of parameter detection.

[0018] Optionally, the percentage of the parameter detection difference of the power semiconductor device IGBT in the parameter constraint range is expressed as:

[0019]

[0020] in, Indicates i The percentage of the absolute distance of parameter detection of a power semiconductor device IGBT in the parameter constraint range;

[0021] A distance percentage data set is obtained based on the percentage of parameter detection differences of the power semiconductor device IGBT in the parameter constraint range.

[0022] Optionally, process the distance percentage dataset to obtain the mean of the distance percentage dataset The expression is:

[0023] .

[0024] Optionally, the mean and standard deviation of the distance percentage data set are processed to obtain the stability state of a single performance parameter of the power semiconductor device IGBT. ρ The expression is:

[0025] ;

[0026] in, is the standard deviation of the distance percentage.

[0027] Optionally, the method further comprises the following steps:

[0028] Obtaining stability status of multiple performance parameters of a power semiconductor device IGBT;

[0029] Obtaining the stability state of the power semiconductor device IGBT under multiple performance parameters based on the performance parameter weights;

[0030] The stability of the power semiconductor device IGBT is determined based on a plurality of energy parameter stability states of the power semiconductor device IGBT.

[0031] Compared with the prior art, the present invention has at least the following beneficial effects:

[0032] (1) The distance percentage data set of the present invention uses the distance between the performance parameter detection value and the nominal value or mean, so that when obtaining the stability state of the power semiconductor device IGBT, there will be no situation where the state value cannot be obtained, that is, the mean value is 0. Even in extreme cases, the mean value is 0, which means that each detection value is equal and equal to the nominal value or mean, and the stability state can still be obtained and processed. The accuracy, stability and applicability of the method are increased.

[0033] (2) The distance percentage data set of the present invention includes relevant information such as the detection mean, nominal value, and nominal extreme value, and associates the processing results used to process the performance stability of the power semiconductor device IGBT with the specified indicator values ​​in its specification manual. It can put the performance stability processing of power semiconductor devices IGBT of the same type, function, and model from different manufacturers and different batches under the same relatively fair and objective evaluation standard, and provide more objective evaluation results.

[0034] (3) The method of the present invention associates the mean of the distance percentage data set with the distance percentage standard deviation, so that the influence of the two parameters on the performance stability of the power semiconductor device IGBT is equalized, thereby avoiding the situation where the performance stability of the power semiconductor device IGBT is more dependent on the mean of the distance percentage data set due to the order of magnitude difference.

[0035] (4) The method of the present invention can make a horizontal comparison of the nominal performance indicators of the same type, same function and same model produced by multiple different manufacturers, thereby providing a basis for enterprises using power semiconductor devices IGBT to quickly process the performance stability of power semiconductor devices IGBT. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings are only for the purpose of illustrating particular embodiments and are not to be construed as limiting the invention.

[0037] Figure 1 A flow chart of a method for processing performance stability of a power semiconductor device IGBT according to the present invention;

[0038] Figure 2 It is a comparison chart of normal distribution simulation data of the true value 1 of Example 1 of the present invention and its evaluation index ρ value;

[0039] Figure 3 It is a comparison chart of normal distribution simulation data of the true value 2 of Example 1 of the present invention and its evaluation index ρ value;

[0040] Figure 4 It is a comparison chart of uniform distribution simulation data of the true value 1 and its evaluation index ρ value of Example 1 of the present invention;

[0041] Figure 5 It is a comparison chart of uniform distribution simulation data of the true value 2 of Example 1 of the present invention and its evaluation index ρ value;

[0042] Figure 6 It is a comparison chart of the sensitivity of the analysis results of Example 1 of the present invention. DETAILED DESCRIPTION

[0043] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. In addition, the present invention can also be implemented in other ways different from those described herein, and therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0044] A specific embodiment of the present invention, as Figure 1-Figure 6, discloses a method for processing the performance stability of a power semiconductor device IGBT, the specific steps of which are:

[0045] Step 1, sampling a plurality of power semiconductor devices IGBT; obtaining a parameter detection value of each power semiconductor device IGBT; obtaining a parameter detection absolute distance data set of the power semiconductor device IGBT based on the parameter detection value; and obtaining a parameter constraint range of the power semiconductor device IGBT;

[0046] The absolute distance of parameter detection of power semiconductor device IGBT is expressed as:

[0047] ;

[0048] in, Indicates i The absolute distance of parameter detection of a power semiconductor device IGBT is the absolute value of the difference between the parameter detection value and the nominal value or the detection mean value; Indicates i Parameter detection value of a power semiconductor device IGBT, i =1,2,…, n , n Indicates the total number of power semiconductor devices IGBT; Indicates nominal value; Represents the detection mean.

[0049] A parameter detection absolute distance data set of the power semiconductor device IGBT is obtained based on the parameter detection value of the power semiconductor device IGBT.

[0050] It can be understood that the nominal value is the rated value of the power semiconductor device IGBT parameter specified in the specification of the power semiconductor device IGBT; the detection mean value is the mean or median value of the detection values ​​of the parameters of all power semiconductor devices IGBT.

[0051] Get the parameter constraint range of the power semiconductor device IGBT, the expression is:

[0052]

[0053] in, L Indicates the parameter constraint range of the power semiconductor device IGBT; P max Indicates the nominal maximum value; P min Indicates the nominal minimum value; x max Indicates the maximum value of parameter detection; x min Indicates the minimum value of parameter detection.

[0054] Furthermore, the parameters of the power semiconductor device IGBT detected are threshold voltage, leakage current and / or negative temperature coefficient thermistor. R NT .

[0055] Step 2, processing the parameter detection difference data set and parameter constraint range of the power semiconductor device IGBT in step 1 to obtain a distance percentage data set;

[0056] The parameter detection difference data set and parameter constraint range of the power semiconductor device IGBT in step 1 are processed to obtain the percentage of the parameter detection difference of the power semiconductor device IGBT in the parameter constraint range, which is expressed as:

[0057]

[0058] in, Indicates i The absolute distance of parameter detection of a power semiconductor device IGBT is the percentage of the parameter constraint range.

[0059] Obtain the distance percentage data set based on the percentage of the parameter detection difference of the power semiconductor device IGBT in the parameter constraint range Y .

[0060] The distance percentage data set constructed by the present invention can remove the influence of the detection data dimension on the evaluation effect, and facilitate parameter normalization when multiple different types of parameters are monitored together.

[0061] Step 3: Based on the distance percentage dataset Y Get the mean of the distance percentage data set. The expression is:

[0062]

[0063] Step 4: Process the mean of the distance percentage data set to obtain the distance percentage standard deviation , the expression is:

[0064] .

[0065] Step 5: Process the mean and standard deviation of the distance percentage data set to obtain the stability status of a single performance parameter of the power semiconductor device IGBT. ρ , the expression is:

[0066] .

[0067] Step 6, processing the single performance parameter stability state of the power semiconductor device IGBT, and determining the stability of the power semiconductor device IGBT;

[0068] If the stability state is close to 0, it is determined that the power semiconductor device IGBT is stable;

[0069] If the stability state is close to 1, it is determined that the power semiconductor device IGBT is unstable.

[0070] It can be understood that for a single performance parameter stability state ρ The smaller the value (the closer to 0), the better the stability of the power semiconductor device IGBT; the larger the value (the closer to 1), the worse the stability of the power semiconductor device IGBT.

[0071] Another embodiment of the present invention discloses a method for processing the performance stability of a power semiconductor device IGBT based on multiple performance parameters, further comprising the following steps:

[0072] Using the aforementioned method to obtain the stability state of multiple performance parameters of the power semiconductor device IGBT;

[0073] Based on the performance parameter weights, the stability state of the power semiconductor device IGBT under multiple performance parameters is obtained, and the expression is:

[0074]

[0075]

[0076] in, Indicates the stability status of power semiconductor devices IGBT under multiple performance parameters; Indicates the power semiconductor device IGBT m The weight of each performance parameter, Indicates the power semiconductor device IGBT m The stability state of a performance parameter.

[0077] Determining the stability of the power semiconductor device IGBT based on the stability states of multiple performance parameters of the power semiconductor device IGBT;

[0078] If the stability state of the multiple performance parameters is close to 0, it is determined that the power semiconductor device IGBT is stable;

[0079] If the stability state of the performance parameter is close to 1, it is determined that the power semiconductor device IGBT is unstable.

[0080] Furthermore, in step S1, when sampling the power semiconductor device IGBT, the power semiconductor device IGBT is sampled using a detection method for accepting the power semiconductor device IGBT, and the specific steps are as follows:

[0081] Step S11. Construct the agreed risks, acceptance indicators and acceptance criteria for both parties;

[0082] Step S12. Construct a theoretical calculation formula for the risk of both parties;

[0083] Step S13. Construct a reliability acceptance sampling scheme design optimization model based on the agreed bilateral risks, acceptance indicators, acceptance criteria and theoretical bilateral risk calculation formula;

[0084] Step S14: constructing a batch product degradation model for simulating and obtaining degradation test data;

[0085] Step S15. Construct a reliability acceptance sampling degradation model, and obtain a reliability function of the batch product considering margin degradation according to the reliability acceptance sampling degradation model; then estimate the parameters of the reliability acceptance sampling degradation model according to the degradation test data, substitute them into the reliability function of the batch product considering margin degradation, and then calculate the reliability of the batch product considering margin degradation using the reliability function of the batch product considering margin degradation determined by the parameters;

[0086] Step S16. According to the reliability of the batch product considering the margin degradation, the theoretical bilateral risk is calculated using the calculation formula of the theoretical bilateral risk;

[0087] Step S17. Based on the theoretical bilateral risk of step S6 and the agreed bilateral risk of step S1, the reliability acceptance sampling scheme of step S3 is used to design an optimization model, and a reliability acceptance sampling scheme that takes margin degradation into account is output.

[0088] Furthermore, the agreed risks of both parties in step S11 include the agreed production risks and agreed user risks ; Among them, the agreed production party risk It is the maximum risk that the producer can bear, agreed upon by both parties in advance, and the risk of the user is agreed upon. It is the maximum risk that the user can bear, as agreed in advance by both parties;

[0089] The acceptance index in step S11 includes the production acceptance index and user acceptance indicators ; Among them, the production acceptance index is stricter than the user acceptance index, expressed as .

[0090] Furthermore, the receiving criteria in step S11 include and ;in, and are the preset initial value of performance degradation and the preset performance degradation rate, respectively, and For fixed values ​​and is a fixed value; is the initial value of product performance degradation; It is the performance degradation rate of the product at any time.

[0091] Furthermore, the calculation formula of the theoretical risk of both parties in step S12 is expressed as:

[0092] ;

[0093] in, Indicates reliability; and They are respectively the theoretical production risk and the theoretical user risk to be obtained through theoretical calculation; represents the reliability function of the batch product considering margin degradation, which is calculated by the reliability acceptance sampling degradation model. s is the stress level, such as: preset temperature; Acceptance sampling plan for reliability; Sample size for reliability acceptance sampling; Time, specifically refers to the test time in reliability acceptance sampling; The detection time interval is equally spaced.

[0094] Furthermore, the step S13 specifically includes the following steps:

[0095] Reliability Acceptance Sampling Plan As the decision variable of the reliability acceptance sampling scheme design optimization model; the optimization goal of the reliability acceptance sampling scheme design optimization model is constructed as the minimum error between the agreed producer risk and the theoretical producer risk, and the minimum error between the agreed user risk and the theoretical user risk, and the decision variable design space is established, which is set as the reliability acceptance sampling sample size , Test time in reliability acceptance sampling and equally spaced detection time intervals The minimum and maximum values ​​of , so the reliability acceptance sampling scheme design optimization model is expressed as:

[0096] .

[0097] in, and They represent the minimum and maximum values ​​of the reliability acceptance sampling sample size respectively; and They represent the minimum and maximum test time in reliability acceptance sampling respectively; and Respectively represent the minimum and maximum values ​​of equally spaced detection time intervals.

[0098] Furthermore, the reliability acceptance sampling sample degradation model in step S15 is expressed as:

[0099] ;

[0100] Among them, the reliability acceptance sampling sample degradation model represents a normal uncertain distribution; , , , are the parameters of the reliability acceptance sampling degradation model that conforms to the characteristics of the degradation test data; among them, is the initial value of the performance degradation of the product in the sampled product, and its mean is , the standard deviation is Normal uncertainty distribution of ; is the indicator function; The mean is , the standard deviation is C(s) represents a normal uncertain distribution with a mean of 0 and a standard deviation of s. is the diffusion coefficient; The mean is 0 and the standard deviation is The normal uncertainty distribution of To be determined based on degradation test data The index of .

[0101] Furthermore, the reliability function of the batch product considering margin degradation in step S15 is expressed as:

[0102] ;

[0103] in, is a predetermined product performance threshold; is the product performance margin; reliability function It is the confidence that the product performance margin is greater than 0.

[0104] Furthermore, the step S15 estimates the parameters of the reliability acceptance sampling degradation model based on the degradation test data, specifically, estimates several groups of unknown parameters in the reliability acceptance sampling degradation model using an uncertain weighted least squares algorithm based on the degradation test data; each group of unknown parameters includes , , , , , , and .

[0105] Further, the step S15 in which the reliability function of the batch product considering margin degradation is substituted, and then the reliability function of the batch product considering margin degradation is calculated using the reliability function of the batch product considering margin degradation determined by the parameters to obtain the reliability of the batch product considering margin degradation, specifically refers to substituting several groups of unknown parameters into the reliability function of the batch product considering margin degradation, respectively obtaining several groups of reliability functions of the batch product considering margin degradation determined by the parameters, and then calculating the reliability of the batch product considering margin degradation using the reliability function of the batch product considering margin degradation determined by the parameters, and then respectively substituting them into the calculation formula of the theoretical two-party risk to calculate the theoretical production risk and theoretical user risk .

[0106] Furthermore, step S17 specifically includes:

[0107] According to the calculated theoretical producer risk and theoretical user risk , combined with the agreed production party risk and agreed user risks , set the theoretical producer risk and agreed producer risks The maximum permissible error between the production ε , Theoretical User Risk and agreed user risks The maximum allowable error between the users ;

[0108] calculate and Are both true? If so, use the reliability acceptance sampling scheme to design an optimization model to output the optimal reliability acceptance sampling scheme. If not, then in the decision variable design space, add 1 to the number of iterations of the reliability acceptance sampling scheme, return to step S14 to re-simulate and obtain the degradation test data for reliability acceptance sampling, and execute the subsequent steps in sequence until the reliability acceptance sampling scheme design optimization model is used to output the optimal reliability acceptance sampling scheme.

[0109] It can be understood that if the initial performance value and the performance degradation rate meet the requirements, the accepted power semiconductor device IGBT is qualified; if either the initial performance value or the performance degradation rate does not meet the requirements, the accepted power semiconductor device IGBT is unqualified.

[0110] In order to illustrate the effectiveness of the method proposed in the present invention, the above-mentioned technical scheme of the present invention is described in detail below through a specific embodiment. In the following description, for the purpose of illustration rather than limitation, specific details such as specific model parameters are proposed so as to thoroughly understand the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention can also be implemented in other embodiments without these specific details. In other cases, the detailed description of the well-known method is omitted to prevent unnecessary details from hindering the description of the present invention.

[0111] This example gives a negative temperature coefficient thermistor resistance R NTC A reliability acceptance sampling scheme design method considering margin degradation is disclosed. The margin degradation of a power semiconductor device IGBT refers to the margin degradation of the resistance value of the power semiconductor device IGBT.

[0112] Step A1. Establish agreed mutual risks, acceptance indicators and acceptance criteria;

[0113] Step A11. Establish agreed risks for both parties: =10% and =8%.

[0114] Step A12. Establish acceptance indicators: Production acceptance indicators =0.99, user acceptance index = 0.01, where s =0.3242, which is the normalized stress level corresponding to 30℃, t = Hour,

[0115] Step A13. Establish acceptance criteria: and .

[0116] Step A2. Construct a theoretical formula for calculating the risk of both parties:

[0117] .

[0118] Step A3. According to the agreed bilateral risk, acceptance index, acceptance criteria and theoretical bilateral risk calculation formula, a reliability acceptance sampling scheme design optimization model for power semiconductor device IGBT is constructed:

[0119] .

[0120] Step A4: construct a batch power semiconductor device IGBT degradation model for simulation to obtain degradation test data.

[0121] Step A41. Constructing a batch power semiconductor device IGBT degradation model as follows:

[0122] ;

[0123] Among them, the IGBT degradation model parameters of this batch of power semiconductor devices are shown in Table 1, and the upper and lower limits of the stress level are 120°C and 0°C respectively.

[0124] Table 1 Designed batch power semiconductor device IGBT degradation model parameters

[0125]

[0126] Step A42: Based on the batch power semiconductor device IGBT degradation model (12), 500 sets of degradation test data for reliability acceptance sampling are obtained through simulation.

[0127] Step A5. Construct a reliability acceptance sampling degradation model for power semiconductor devices IGBT, and obtain a reliability function of batch power semiconductor devices IGBT considering margin degradation based on the reliability acceptance sampling degradation model for power semiconductor devices IGBT; then, based on degradation test data, estimate the parameters of the reliability acceptance sampling degradation model for power semiconductor devices IGBT, substitute them into the reliability function of batch power semiconductor devices IGBT considering margin degradation, and then use the reliability function of batch power semiconductor devices IGBT considering margin degradation determined by the parameters to calculate the reliability of batch power semiconductor devices IGBT considering margin degradation.

[0128] Step A51. Construct a reliability acceptance sampling degradation model for power semiconductor device IGBT:

[0129] .

[0130] Furthermore, the reliability function of batch power semiconductor devices IGBT with margin degradation is:

[0131] .

[0132] Step A52. Based on the degradation test data of step A4, use the uncertain weighted least squares algorithm to estimate 50 sets of unknown parameters in the reliability acceptance sampling degradation model. , k =1,2,…,50.

[0133] Step A53. Estimated 50 sets of unknown parameters Substitute them into the reliability function of the batch power semiconductor device IGBT considering margin degradation respectively, and use the reliability function of the batch power semiconductor device IGBT considering margin degradation determined by parameters to calculate the reliability of 50 groups of batch power semiconductor devices IGBT considering margin degradation. , that is, each Corresponding to one .

[0134] Step A6. Based on the reliability of the batch power semiconductor device IGBT considering margin degradation, the theoretical double risk is calculated using the calculation formula of the theoretical double risk. Specifically, based on the reliability of 50 groups of batch power semiconductor devices IGBT considering margin degradation , and the theoretical double-party risk calculation formula, according to the uncertainty theory and uncertainty simulation operation rules, the theoretical production risk is calculated and theoretical user risk :

[0135] Step A7. Based on the theoretical bilateral risk of step A6 and the agreed bilateral risk of step A1, the reliability acceptance sampling scheme of the power semiconductor device IGBT of step A3 is used to design an optimization model, and the reliability acceptance sampling scheme of the power semiconductor device IGBT considering margin degradation is output. The maximum allowable error between the theoretical risk and the agreed risk is set. ε =10%.

[0136] Finally, after calculation, we can get the risk of both parties meeting the agreement under the condition of meeting the error condition. and The reliability acceptance sampling plan for power semiconductor devices IGBT is = (8, 1000, 100), at this time, the production risk and user risk calculated by the reliability acceptance sampling plan of the power semiconductor device IGBT are (10.81%, 7.69%) respectively.

[0137] In order to illustrate the effectiveness of the method proposed in the present invention, the above technical solution of the present invention is described in detail below through a specific embodiment. In the following simulation analysis case, the negative temperature coefficient thermistor of the power semiconductor device IGBT is mainly simulated. R NTC The test data of the same power semiconductor device IGBT from two manufacturers was used in the actual test case. V GE(th) , leakage current I GES and negative temperature coefficient thermistor R NTC The measured data.

[0138] In Example 1, a simulation method is first used to compare two groups of data with obvious differences in various situations to see whether the method of the present invention can obtain expected evaluation results, and a sensitivity analysis is performed; then a weighted comprehensive evaluation is calculated based on the measured data of three parameters of power semiconductor devices IGBT from two different manufacturers to illustrate the entire application process and evaluation effect of the algorithm.

[0139] 1) Parameter stability evaluation and analysis results based on simulation algorithm

[0140] Two random distribution models are used for simulation sampling: the normal distribution model, the main parameters are the mean μ and variance σ of the performance parameter detection value; the uniform distribution model, the main parameters are the maximum possible value a and the minimum possible value b of the performance parameter detection value. In the analysis process, the assumed parameter nominal value and its maximum and minimum nominal values ​​are first given as the evaluation data. The closer the distance between the mean and the nominal value in the expected simulation data, the smaller the fluctuation, and the better the stability.

[0141] The assumed parameter specifications are shown in Table 2:

[0142] Table 2 Assumed parameter specifications

[0143]

[0144] a) Sampling simulation data of parameter detection values ​​based on normal distribution;

[0145] ① Case 1: Data with the same fluctuation but significantly different mean and nominal value:

[0146] In this case, it is assumed that the device performance stability represented by the data whose test results are close to the nominal value is better. According to the algorithm, the smaller the ρ value, the better the stability.

[0147] The true value 1 of the simulation data based on normal distribution is shown in Table 3:

[0148] Table 3 True value of simulation data of normal distribution 1

[0149]

[0150] From the above table, we can see that it is currently assumed that data 2 is closer to the nominal value, so the stability of data 2 should be better.

[0151] Based on the above true value 1, the normal distribution simulation data and its evaluation index ρ value are as follows Figure 2 As shown, in order to intuitively show the distance between the two sets of data and the nominal value, the nominal value data is also given in the figure as a reference.

[0152] Figure 2The specific values ​​of the evaluation values ​​ρ (the first pair of data) of the two sets of simulation data are shown in Table 4:

[0153] Table 4 Evaluation value ρ of two sets of simulation data (first pair of data)

[0154]

[0155] from Figure 2 From the results in Table 4, it can be seen that the results of the stability evaluation are consistent with expectations.

[0156] ② Case 2: Data with significantly different fluctuations but the same difference between the mean and the nominal value

[0157] In this case, we assume that the smaller the fluctuation of the test results, the better the stability of the device performance. According to the algorithm, the smaller the ρ value, the better the stability.

[0158] The true value 2 of the simulation data based on normal distribution is shown in Table 5:

[0159] Table 5 True value of simulation data of normal distribution 2

[0160]

[0161] From the above table, we can see that it is currently assumed that data set 2 has smaller fluctuations, so the stability of data set 2 should be better.

[0162] Based on the above true value 2, the normal distribution simulation data and its evaluation index ρ value are as follows Figure 3 As shown, in order to intuitively show the distance between the two sets of data and the nominal value, the nominal value data is also given in the figure as a reference.

[0163] Figure 3 The specific values ​​of the evaluation values ​​ρ (the second pair of data) of the two sets of simulation data are shown in Table 6:

[0164] Table 6 Evaluation values ​​ρ of two sets of simulation data (second pair of data)

[0165]

[0166] From the results in Figure 3 and Table 6, it can be seen that the results of the stability evaluation are consistent with expectations.

[0167] b) Sampling simulation data based on uniformly distributed parameter detection values

[0168] ① Case 1: Data with the same range but significantly different median and nominal values

[0169] In this case, we assume that the data in the test results that are close to the nominal value represent better device performance stability. According to the algorithm, the smaller the ρ value, the better the stability.

[0170] The true value of simulation data based on uniform distribution 1 is shown in Table 7:

[0171] Table 7 Uniformly distributed simulation data true value 1

[0172]

[0173] From the above table, we can see that it is currently assumed that the median value of data 2 is closer to the nominal value, so the stability of data 2 should be better.

[0174] Based on the uniform distribution simulation data of the above true value 1 and its evaluation index ρ value, Figure 4 As shown, in order to intuitively show the distance between the median of the two sets of data and the nominal value, the nominal value data is also given in the figure as a reference.

[0175] Figure 4 The specific values ​​of the evaluation value ρ (the third pair of data) of the two sets of simulation data are shown in Table 8:

[0176] Table 8 Evaluation value ρ of two sets of simulation data (third pair of data)

[0177]

[0178] From the results in Figure 4 and Table 8, it can be seen that the results of the stability evaluation are consistent with expectations.

[0179] ② Case 2: Data with significantly different ranges but consistent median and nominal values

[0180] In this case, we assume that the smaller the range of the test results, the better the stability of the device performance. According to the algorithm, the smaller the ρ value, the better the stability.

[0181] The true value 2 of the simulation data based on uniform distribution is shown in Table 9:

[0182] Table 9 True value of uniformly distributed simulation data 2

[0183]

[0184] From the above table, we can see that it is currently assumed that the range of data 2 is smaller, so the stability of data 2 should be better.

[0185] The uniformly distributed simulation data based on the above true value and its evaluation index ρ value are as follows Figure 5 As shown, in order to intuitively show the distance between the median of the two sets of data and the nominal value, the nominal value data is also given in the figure as a reference.

[0186] Figure 5 The specific values ​​of the evaluation value ρ (the fourth pair of data) of the two sets of simulation data are shown in Table 10:

[0187] Table 10 Evaluation value ρ of two sets of simulation data (the fourth pair of data)

[0188]

[0189] From the results in Figure 5 and Table 10, it can be seen that the results of the stability evaluation are consistent with expectations.

[0190] c) Sensitivity analysis

[0191] In order to illustrate the sensitivity of this evaluation method, the influence of ρ value on the mean and standard deviation of the detection value is analyzed based on a normal distribution. The simulation true value used for sensitivity analysis is shown in Table 11.

[0192] Table 11 Sensitivity analysis simulation true value

[0193]

[0194] ① Fix the σ value, and simulate the μ value by ±2%, ±4%, ±6%, ±8%, and ±10% for 20 times respectively. Then take the mean value of ρ as the corresponding changed ρ value and draw the graph.

[0195] ② Fix the μ value, and simulate σ values ​​of ±2%, ±4%, ±6%, ±8%, and ±10% for 20 times respectively. Then take the mean value of ρ as the corresponding changed ρ value and draw the graph.

[0196] Depend on Figure 6 It can be seen that the parameter stability evaluation value obtained by this algorithm is more sensitive to the mean of the samples.

[0197] 2) Implementation process of comprehensive parameter stability evaluation based on measured data of power semiconductor device IGBT

[0198] The following is an implementation description of the comprehensive stability evaluation of this algorithm using the measured data of power semiconductor devices IGBT from two manufacturers. Since there are many performance parameters of power semiconductor devices IGBT in the specification, we mainly use three of them in this example. GES 、V GE(th) and R NTC To evaluate, after communicating with hardware engineers, it can be determined that in actual applications I GES It is the most important parameter, so it is based on I in the comprehensive evaluation. GES The higher the value of ρ, the higher the weight.

[0199] The nominal values ​​in the IGBT specifications of two power semiconductor devices are shown in Table 12 and Table 13:

[0200] Table 12 Nominal values ​​in manufacturer I specification manual

[0201]

[0202] Table 13 Nominal values ​​in manufacturer's S specification sheet

[0203]

[0204] In this test, there are 20 power semiconductor devices IGBT from manufacturer I and 24 power semiconductor devices IGBT from manufacturer S. The measured values ​​of the three parameters are shown in Table 14-Table 19:

[0205] Table 14 I of device from manufacturer I GES Measured value

[0206]

[0207] Table 15 V of device from manufacturer I GE(th) Measured value

[0208]

[0209] Table 16 R of device from manufacturer I NTC Measured value

[0210]

[0211] Table 17 I of the device of manufacturer S GES Measured value

[0212]

[0213] Table 18 V of manufacturer S devices GE(th) Measured value

[0214]

[0215] Table 19 Manufacturer S Device R NTC Measured value

[0216]

[0217] The calculation results of the stability evaluation values ​​of the three parameters are shown in Table 20-Table 22:

[0218] Table 20 I GES Stability comparison

[0219]

[0220] Table 21 V GE(th) Stability comparison

[0221]

[0222] Table 22 R NTC Stability comparison

[0223]

[0224] Because engineers pay more attention to I GES Therefore, its weight is set to 0.8, and the stability weights of the other two parameters are set to 0.1, namely:

[0225]

[0226]

[0227]

[0228] The weighted comprehensive stability evaluation results are shown in Table 23:

[0229] Table 23 Comprehensive stability evaluation results

[0230]

[0231] From the results in the above table, we can see that the IGBT performance parameters of manufacturer S are more stable.

[0232] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for processing the performance stability of a power semiconductor device IGBT, characterized in that: The specific steps are: Step 1, sampling a plurality of power semiconductor devices IGBT; obtaining a parameter detection value of each power semiconductor device IGBT; obtaining a parameter detection absolute distance data set of the power semiconductor device IGBT based on the parameter detection value; and obtaining a parameter constraint range of the power semiconductor device IGBT; The expression of absolute distance for parameter detection of power semiconductor device IGBT is: ； in, Indicates i The absolute distance of parameter detection of a power semiconductor device IGBT is the absolute value of the difference between the parameter detection value and the nominal value or the detection mean value; Indicates i Parameter detection value of a power semiconductor device IGBT, i =1,2,…, n , n Indicates the total number of power semiconductor devices IGBT; Indicates nominal value; represents the detection mean; Step 2, processing the parameter detection absolute distance data set and parameter constraint range of the power semiconductor device IGBT in step 1 to obtain a distance percentage data set; The expression of the absolute distance of parameter detection of power semiconductor device IGBT as a percentage of the parameter constraint range is: in, Based on the percentage of absolute distance of parameter detection of power semiconductor device IGBT in parameter constraint range, a distance percentage data set is obtained; Step 3: Process the distance percentage data set to obtain the mean of the distance percentage data set. The expression is: in, Represents the mean of the distance percentage data set; Indicates i The percentage of the absolute distance of parameter detection of a power semiconductor device IGBT in the parameter constraint range; Step 4: Process the mean of the distance percentage data set to obtain the distance percentage standard deviation, expressed as: ； Step 5: Process the mean and standard deviation of the distance percentage data set to obtain the stability status of a single performance parameter of the power semiconductor device IGBT. ρ , the expression is: in, represents the standard deviation of the distance percentage; Step 6: Process the stability state of a single performance parameter of the power semiconductor device IGBT to determine the stability of the power semiconductor device IGBT.

2. The method for processing the performance stability of the power semiconductor device IGBT according to claim 1, characterized in that: The parameter constraint range of the power semiconductor device IGBT is expressed as: in, P max Indicates the nominal maximum value; P min Indicates the nominal minimum value; x max Indicates the maximum value of parameter detection; x min Indicates the minimum value of parameter detection.

3. The method for processing the performance stability of a power semiconductor device IGBT according to any one of claims 1 to 2, characterized in that: The following steps are also included: Obtaining stability states of multiple single performance parameters of a power semiconductor device IGBT; Obtaining the stability state of the power semiconductor device IGBT under multiple performance parameters based on the performance parameter weights and the stability states of multiple single performance parameters; The stability of the power semiconductor device IGBT is determined based on the stability state of the power semiconductor device IGBT under multiple performance parameters.

Images