A chip quality assessment system and method based on data analysis

Abstract

The present invention relates to the field of chip quality assessment, and in particular to a chip quality assessment system and method based on data analysis, comprising a target chip package marking module, a chip detection result recording module, a package classification module, a result judgment module and a quality assessment processing module; the target chip package marking module is used to mark a package that has passed the printed information detection after the chip is packaged as a target chip package; the chip detection result recording module obtains the output result of the chip detection of the target chip package; the package classification module is used to mark the target chip package with abnormal physical detection as a first package and the target chip package with abnormal electrical performance detection as a second package; the result judgment module judges whether the output results of each of several chip detections recorded for the target chip package with the same printed information are the same for the first package and the second package; the quality assessment processing module outputs the result for quality assessment processing.

Description

Technical Field

[0001] The present invention relates to the technical field of chip quality assessment, and in particular to a chip quality assessment system and method based on data analysis. Background Art

[0002] As an important component of modern electronic devices, the quality and performance stability of chips have a decisive impact on the reliability of the entire system. Therefore, strict quality control and testing are required during the production and use process; the chip testing process mainly includes three links: pre-process testing, post-process testing and pre-shipment testing; and pre-shipment testing refers to the last test after the chip is produced; it is mainly to conduct a comprehensive test of various parameters of the chip to ensure that the quality and performance of the chip meet the requirements; and in this testing link, two types of tests will be conducted, one is physical testing, and the other is electrical performance testing. Usually, the testing process will conduct the above two tests on the same batch of test chips, and after one test is completed, the chip package that shows abnormal results is often directly processed, and the deviation and repairability of the chip package corresponding to the two types of tests are not considered at the same time, which makes the chip detection unintelligent and may cause the chip package to be screened out as abnormal once, and the yield rate decreases in this link. Summary of the invention

[0003] The object of the present invention is to provide a chip quality assessment system and method based on data analysis to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a chip quality assessment method based on data analysis, comprising the following analysis steps:

[0005] Step S100: marking the package body that has passed the printed information test after the chip is packaged as the target chip package body, obtaining the output result of the chip test on the target chip package, the chip test includes physical test and electrical performance test; the output result records the target chip package body with abnormal physical test and the target chip package body with abnormal electrical performance test;

[0006] Step S200: Mark the target chip package with physical test abnormality as the first package, and the target chip package with electrical performance test abnormality as the second package; if the target chip package with the same printed information records the output results of several chip tests each time, and the first package and the second package are the same, then execute step S300; otherwise, execute step S400;

[0007] Step S300: Mark the corresponding printing information as the target printing information. When there is a real-time package body with the same target printing information for chip detection, after any type of detection is performed, the real-time package body with abnormal output result is marked as unqualified. After removing the real-time package body with unqualified quality, the remaining real-time package bodies are subjected to another type of detection. If the abnormal chip packages detected by the two types of detection are the same after several times, it means that the abnormality of one characteristic of the chip of this type will inevitably lead to the abnormality of another characteristic. In this case, it is not necessary to repeat the detection when implementing the detection. It is only necessary to detect the removed chips, and eliminate the influence caused by the corresponding characteristic characteristics of the previous type of detection when the subsequent detection is abnormal.

[0008] Step S400: acquiring the test data of the chip test record of the target chip package corresponding to the printed information, and analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of tests based on the test data;

[0009] Step S500: matching the real-time package with the identifiable features of the target chip package with the same printed information during various types of inspections, and performing chip quality assessment processing on the real-time package based on the identifiable features.

[0010] Further, analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of inspections includes the following specific steps:

[0011] Extracting a target chip package body marked as the first package body but not marked as the second package body as an inspection package body, and removing the inspection package body from the first package body and the remaining target chip packages as control packages;

[0012] The test data refers to the distance L between the pins on both sides of the package recorded during the physical test. 1 , the number of pin pairs M 1 and the height H of the package 1 The height of the package body refers to the vertical distance from the package front surface to the lowest point of all the pins of the package body; the pin pair refers to the number of pins on both sides of the chip package body when there is no pin missing. A pin pair records two pins on the same horizontal line; using the formula:

[0013] T=[M 2 / M 1 ]*[|L 2 -L 1 | / L 2 +|H 2 -H 1 | / H 2 ]

[0014] Calculate the structural characteristic index T, M of each package under investigation 2Indicates the number of pin pairs when the physical test of the target chip package with the same printed information is passed, L 2 It represents the average pin distance when the physical test of the target chip package with the same printed information is passed, H 2 It indicates the average package height of the target chip package with the same printed information when the physical test is passed; the larger the structural characteristic index is, the worse the structural characteristics of the package are when the physical test is conducted;

[0015] Classify the investigated packages whose structural characteristic index difference is less than or equal to the difference threshold value into the same category of investigated packages, and sort them from small to large according to the values ​​of the structural characteristic index corresponding to each category of investigated packages, and generate a first sequence;

[0016] Extract the test results of various packages in the first sequence in several historical chip test records, using the formula:

[0017] E=(e 1 / e 2 )*(s 1 / s 2 ),

[0018] Calculate the abnormal rate E of each package under investigation, where e 1 Indicates the number of packages of the corresponding category in the same chip test, e 2 Indicates the total number of packages inspected during the same chip inspection; s 1 Indicates that the corresponding category examines the package in s 2 The number of detection events recorded in each chip detection, s 2 Indicates the total number of chip detection events in the historical records for the chip package with the same printed information; the higher the abnormality rate, the more frequently the corresponding type of package has only physical detection abnormalities but not electrical performance detection abnormalities in the historical records;

[0019] A variable package is constructed based on the structural characteristic index and abnormality rate of a class of package records, and the correlation coefficient r is calculated based on the variable package, and the correlation coefficient threshold r is set. 0 ;

[0020] When <r 0 When the minimum value tmin of the structural characteristic index recorded by the control package is obtained, the abnormality rate e corresponding to the minimum value tmin of the structural characteristic index recorded by the control package is obtained. 0 , and mark the abnormal rate E recorded by the inspection package to be greater than the abnormal rate e 0 The investigated package body of the category corresponding to the minimum value of is the first valid package body, and the structural feature index T recorded by the first valid package body is extracted as the identifiable feature of the target chip package body during physical inspection;

[0021] When r ≥ r 0 When the structural feature index recorded by each type of package under investigation is output, it is the identifiable feature of the target chip package during physical inspection.

[0022] Furthermore, analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of inspections also includes the following specific steps:

[0023] The test data also includes the electrical performance parameter type and the corresponding parameter value;

[0024] Obtain the difference d between the electrical performance parameter values ​​recorded during the electrical performance test of the package under investigation and the corresponding electrical performance parameter values ​​recorded by the second package, d=q 1 -q 2 ,q 1 Indicates the electrical performance parameter value recorded by the second package, q 2 Indicates the parameter value of the same type of electrical performance parameters of the package under investigation, and calculates the average difference D of all types of electrical performance parameters of each type of package under investigation, D=(1 / g)∑d 0 , d 0 represents the normalized difference of the electrical performance parameter value, and g represents the total number of electrical performance parameter types;

[0025] Extract the same printed information corresponding to the chip package in s 2 In the chip detection event, the average minimum difference Dmin of each type of package under investigation is recorded as the target difference; the minimum value of the target difference corresponding to all types of package under investigation is marked as the identifiable feature of the target chip package during electrical performance detection.

[0026] Further, the chip quality assessment process is performed on the real-time package based on the identifiable features, including the following specific steps:

[0027] When the real-time package is first physically inspected, the distance between the pins on both sides of the package, the number of pin pairs and the height of the package recorded by the real-time package are obtained, and the structural characteristic index T of each real-time package is calculated by substituting them into the structural characteristic index calculation formula. 0 ;

[0028] In r <r 0 When the real-time structural feature index T is extracted 0 The live package that is smaller than the identifiable feature is the first pin quality package; when r ≥ r 0 When , the real-time package body whose difference with the identifiable feature is less than the difference threshold is taken as the first pin quality package body;

[0029] And among the real-time packages with abnormal output detection after the physical inspection, the first pin quality package is eliminated, and the remaining real-time packages are no longer subject to electrical performance inspection; and the first pin quality package is given a quality warning reminder; the purpose of the quality warning reminder is that there is an adjustable deviation in the pin, which has little effect on the actual electrical performance test and can still be evaluated as qualified after adjustment;

[0030] When the real-time package is first tested for electrical performance, the electrical performance parameter type and the corresponding real-time parameter value recorded by the real-time package are obtained, and the real-time average difference D is calculated. 0 , extract the real-time average difference D 0 The real-time package body that is greater than or equal to the target difference Dmin and less than the second difference threshold is the second pin quality package body, and the second difference threshold is the maximum value of the average difference of the electrical performance parameter values ​​recorded for the corresponding chip package body whose output result is qualified quality during electrical performance detection; the second pin quality package body is given a quality warning reminder, and the remaining packages are removed from the packages with abnormal real-time detection results except the second pin quality packages, and no physical detection is performed;

[0031] After the detection is normal and the quality warning reminder is restored, the first pin quality package and the second pin quality package are subjected to electrical performance detection again.

[0032] The present application performs intelligent data screening on chip packages in two different detection processes, so as to avoid judgment bias caused by a one-shot decision during the detection process. That is, although there are defects in physical features such as bent pins during physical testing, they may not necessarily affect the test results in electrical performance testing, and the bent pins can still be used after recovery. Alternatively, in electrical performance testing, chip packages with bent pins that can be determined by differences in electrical performance parameters can be effectively screened out, so that in further physical testing, such packages can be pre-analyzed or no physical testing can be performed, thereby reducing the actual workload of chip package testing and improving efficiency. At the same time, it also avoids the waste of abnormal chip package processing at different detection stages due to judgment errors.

[0033] A chip quality assessment system based on data analysis, comprising a target chip package marking module, a chip detection result recording module, a package classification module, a result judgment module and a quality assessment processing module;

[0034] The target chip package marking module is used to mark the package that has passed the printing information inspection after the chip is packaged as the target chip package;

[0035] The chip detection result recording module is used to obtain the output result of chip detection of the target chip package, and the chip detection includes physical detection and electrical performance detection; the output result records the target chip package with abnormal physical detection and the target chip package with abnormal electrical performance detection;

[0036] The package classification module is used to mark the target chip package with abnormal physical detection as the first package and the target chip package with abnormal electrical performance detection as the second package;

[0037] The result judgment module is used to judge whether each output result of a plurality of chip tests recorded by the target chip package with the same printed information is the same for the first package and the second package;

[0038] The quality assessment processing module is used to perform quality assessment processing based on the output results of the result judgment module.

[0039] Further, the result judgment module includes a target printing information marking unit and a recognizable feature analysis unit;

[0040] The target printed information marking unit is used for recording, on a target chip package of the same printed information, that each output result of a plurality of chip tests is the same as that of the first package and the second package, marking the corresponding printed information as the target printed information;

[0041] The identifiable feature analysis unit is used to obtain detection data of chip detection records of the target chip package corresponding to the printed information, and analyze the identifiable features of the target chip package corresponding to the same printed information when performing various types of detection based on the detection data.

[0042] Further, the identifiable feature analysis unit includes a package marking unit, a detection data classification unit and an identifiable feature output unit;

[0043] The package marking unit is used to extract the target chip package marked as the first package but not marked as the second package as the inspection package, and the remaining target chip packages in the first package excluding the inspection package as the control package;

[0044] The test data classification unit is used to obtain the distance between the pins on both sides of the package, the number of pin pairs and the height of the package recorded in the physical test, as well as the type of electrical performance parameters and the corresponding parameter values;

[0045] The identifiable feature output unit is used to output the identifiable features of physical detection and the identifiable features of electrical performance detection.

[0046] Further, the detection data classification unit includes a structural feature index calculation unit, a first sequence generation unit, an abnormality rate calculation unit and an average difference calculation unit;

[0047] The structural characteristic index calculation unit is used to calculate the structural characteristic index of each package under investigation;

[0048] The first sequence generating unit is used to classify the investigated packages whose structural characteristic index difference in the investigated packages is less than or equal to the difference threshold value as the same type of investigated packages, and sort the investigated packages from small to large according to the values ​​of the structural characteristic index corresponding to each type of investigated packages, and generate a first sequence;

[0049] The abnormality rate calculation unit is used to extract the test results of various types of examined packages in the first sequence in several historical chip test records, and calculate the abnormality rate of various types of examined packages;

[0050] The average difference calculation unit is used to calculate the average difference of all types of electrical performance parameters corresponding to each type of package under investigation.

[0051] Further, the quality assessment processing module includes a quality failure marking unit, a package rejection unit and a detection response unit;

[0052] The unqualified quality marking unit is used for marking the real-time package with abnormal output result as unqualified quality after any type of detection is performed when there is a real-time package with the same printed information as the target for chip detection, and after removing the unqualified real-time package, the remaining real-time package is subjected to another type of detection;

[0053] The package rejection unit is used to reject packages that do not meet the detection requirements;

[0054] The detection response unit is used to provide quality warning reminders for different packages.

[0055] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention conducts digital analysis on the chip packages with historical records, and considers the impact of different degrees of damage to the pins of the chip packages in the chip detection process; it enables intelligent screening of the chip packages when two types of detection are carried out successively, rather than simply directly screening out based on one type of detection result; it enables the output of the results of preliminary analysis or no physical detection of this type of package in further detection, thereby reducing the actual workload of chip package detection and improving efficiency; it also avoids the waste of abnormal chip package processing at different detection stages due to judgment errors, and effectively analyzes the difference between the bent pins of the chip package that can be restored and the missing pins that can be screened out; it greatly improves the yield rate of chip packaging on the basis of intelligent improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0057] Figure 1 It is a structural schematic diagram of a chip quality evaluation system based on data analysis of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] See also Figure 1 The present invention provides a technical solution: a chip quality assessment method based on data analysis, comprising the following analysis steps:

[0060] Step S100: marking the package body that has passed the printed information test after the chip is packaged as the target chip package body, and the printed information includes information such as chip name, specification, model, etc.; obtaining the output result of chip testing on the target chip package body, and the chip testing includes physical testing and electrical performance testing; the output result records the target chip package body with abnormal physical testing and the target chip package body with abnormal electrical performance testing;

[0061] In this application, after the chip is packaged, it indicates that there is no abnormality in the chip itself before packaging, and then the next process can be carried out to achieve packaging;

[0062] Step S200: Mark the target chip package with physical property abnormality as the first package, and the target chip package with electrical property abnormality as the second package; if the target chip package with the same printed information records that the output results of each of several chip tests are the same for the first package and the second package, then execute step S300; otherwise, execute step S400; in this application, the target chip packages recorded in the history extraction for the two tests are the same;

[0063] Step S300: Mark the corresponding printing information as the target printing information. When there is a real-time package body with the same target printing information for chip detection, after any type of detection is performed, the real-time package body with abnormal output result is marked as unqualified. After removing the real-time package body with unqualified quality, the remaining real-time package bodies are subjected to another type of detection. If the abnormal chip packages detected by the two types of detection are the same after several times, it means that the abnormality of one characteristic of the chip of this type will inevitably lead to the abnormality of another characteristic. In this case, it is not necessary to repeat the detection when implementing the detection. It is only necessary to detect the removed chips, and eliminate the influence caused by the corresponding characteristic characteristics of the previous type of detection when the subsequent detection is abnormal.

[0064] Step S400: acquiring the test data of the chip test record of the target chip package corresponding to the printed information, and analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of tests based on the test data;

[0065] Step S500: matching the real-time package with the identifiable features of the target chip package with the same printed information during various types of inspections, and performing chip quality assessment processing on the real-time package based on the identifiable features.

[0066] Analyzing the identifiable features of the target chip package corresponding to the same printed information during various types of inspections includes the following specific steps:

[0067] Extracting a target chip package body marked as the first package body but not marked as the second package body as an inspection package body, and removing the inspection package body from the first package body and the remaining target chip packages as control packages;

[0068] The test data refers to the distance L between the pins on both sides of the package recorded during the physical test. 1 , the number of pin pairs M 1 and the height H of the package 1 The height of the package body refers to the vertical distance from the package front surface to the lowest point of all the pins of the package body; the pin pair refers to the number of pins on both sides of the chip package body when there is no pin missing. A pin pair records two pins on the same horizontal line; using the formula:

[0069] T=[M 2 / M 1 ]*[|L 2 -L 1 | / L 2 +|H 2 -H 1 | / H 2 ]

[0070] Calculate the structural characteristic index T, M of each package under investigation 2Indicates the number of pin pairs when the physical test of the target chip package with the same printed information is passed, L 2 It represents the average pin distance when the physical test of the target chip package with the same printed information is passed, H 2 It indicates the average package height of the target chip package with the same printed information when the physical test is passed; the larger the structural characteristic index is, the worse the structural characteristics of the package are when the physical test is conducted;

[0071] Classify the investigated packages whose structural characteristic index difference is less than or equal to the difference threshold value into the same category of investigated packages, and sort them from small to large according to the values ​​of the structural characteristic index corresponding to each category of investigated packages, and generate a first sequence;

[0072] Extract the test results of various packages in the first sequence in several historical chip test records, using the formula:

[0073] E=(e 1 / e 2 )*(s 1 / s 2 ),

[0074] Calculate the abnormal rate E of each package under investigation, where e 1 Indicates the number of packages of the corresponding category in the same chip test, e 2 Indicates the total number of packages inspected during the same chip inspection; s 1 Indicates that the corresponding category examines the package in s 2 The number of detection events recorded in each chip detection, s 2 Indicates the total number of chip detection events in the historical records for the chip package with the same printed information; the higher the abnormality rate, the more frequently the corresponding type of package has only physical detection abnormalities but not electrical performance detection abnormalities in the historical records;

[0075] A variable package is constructed based on the structural characteristic index and abnormality rate of a class of package records, and the correlation coefficient r is calculated based on the variable package, and the correlation coefficient threshold r is set. 0 In practical applications, the two variables are usually negatively correlated, that is, when the structural characteristic index is smaller, the bending degree of the pins of the chip package is smaller, and the impact on the chip package is smaller, and the corresponding abnormality rate is naturally higher, that is, the frequency of this bending degree corresponding to only physical detection but not electrical performance detection is higher; because the bending degree at this time may not affect the electrical performance detection of the chip application; when r <r 0 hour,

[0076] Obtain the abnormality rate e corresponding to the minimum value tmin of the structural characteristic index recorded by the control package0 , and mark the abnormal rate E recorded by the inspection package to be greater than the abnormal rate e 0 The investigated package body of the category corresponding to the minimum value of is the first valid package body, and the structural feature index T recorded by the first valid package body is extracted as the identifiable feature of the target chip package body during physical inspection;

[0077] When r ≥ r 0 When , it means that the two variables are unrelated, and the output structural feature index recorded for each type of package under investigation is the identifiable feature of the target chip package during physical testing.

[0078] In the present application, when calculating the abnormality rate of the control package, the calculation method is the same as that of the investigation package, but the calculation object is different. When analyzing the abnormality rate of the control package, data extraction with the control package as the category is selected.

[0079] Analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of inspections also includes the following specific steps:

[0080] The test data also includes the electrical performance parameter type and the corresponding parameter value;

[0081] Obtain the difference d between the electrical performance parameter values ​​recorded during the electrical performance test of the package under investigation and the corresponding electrical performance parameter values ​​recorded by the second package, d=q 1 -q 2 ,q 1 Indicates the electrical performance parameter value recorded by the second package, q 2 Indicates the parameter value of the same type of electrical performance parameters of the package under investigation, and calculates the average difference D of all types of electrical performance parameters of each type of package under investigation, D=(1 / g)∑d 0 , d 0 represents the normalized difference of the electrical performance parameter value, and g represents the total number of electrical performance parameter types;

[0082] Extract the same printed information corresponding to the chip package in s 2 In the chip detection event, the average minimum difference Dmin of each type of package under investigation is recorded as the target difference; the minimum value of the target difference corresponding to all types of package under investigation is marked as the identifiable feature of the target chip package during electrical performance detection.

[0083] The chip quality assessment process of the real-time package body based on the identifiable features includes the following specific steps:

[0084] When the real-time package is first physically inspected, the distance between the pins on both sides of the package, the number of pin pairs and the height of the package recorded by the real-time package are obtained, and the structural characteristic index T of each real-time package is calculated by substituting them into the structural characteristic index calculation formula. 0;

[0085] In r <r 0 When the real-time structural feature index T is extracted 0 The live package that is smaller than the identifiable feature is the first pin quality package; when r ≥ r 0 When , the real-time package body whose difference with the identifiable feature is less than the difference threshold is taken as the first pin quality package body;

[0086] And among the real-time packages with abnormal output detection after the physical inspection, the first pin quality package is eliminated, and the remaining real-time packages are no longer subject to electrical performance inspection; and the first pin quality package is given a quality warning reminder; the purpose of the quality warning reminder is that there is an adjustable deviation in the pin, which has little effect on the actual electrical performance test and can still be evaluated as qualified after adjustment;

[0087] When the real-time package is first tested for electrical performance, the electrical performance parameter type and the corresponding real-time parameter value recorded by the real-time package are obtained, and the real-time average difference D is calculated. 0 , extract the real-time average difference D 0 The real-time package body that is greater than or equal to the target difference Dmin and less than the second difference threshold is the second pin quality package body, and the second difference threshold is the maximum value of the average difference of the electrical performance parameter values ​​recorded for the corresponding chip package body whose output result is qualified quality during electrical performance detection; the second pin quality package body is given a quality warning reminder, and the remaining packages are removed from the packages with abnormal real-time detection results except the second pin quality packages, and no physical detection is performed;

[0088] After the detection is normal and the quality warning reminder is restored, the first pin quality package and the second pin quality package are subjected to electrical performance detection again.

[0089] The present application performs intelligent data screening on chip packages in two different detection processes, so as to avoid judgment bias caused by a one-shot decision during the detection process. That is, although there are defects in physical features such as bent pins during physical testing, they may not necessarily affect the test results in electrical performance testing, and the bent pins can still be used after recovery. Alternatively, in electrical performance testing, chip packages with bent pins that can be determined by differences in electrical performance parameters can be effectively screened out, so that in further physical testing, such packages can be pre-analyzed or no physical testing can be performed, thereby reducing the actual workload of chip package testing and improving efficiency. At the same time, it also avoids the waste of abnormal chip package processing at different detection stages due to judgment errors.

[0090] A chip quality assessment system based on data analysis, comprising a target chip package marking module, a chip detection result recording module, a package classification module, a result judgment module and a quality assessment processing module;

[0091] The target chip package marking module is used to mark the package that has passed the printing information inspection after the chip is packaged as the target chip package;

[0092] The chip detection result recording module is used to obtain the output result of chip detection of the target chip package, and the chip detection includes physical detection and electrical performance detection; the output result records the target chip package with abnormal physical detection and the target chip package with abnormal electrical performance detection;

[0093] The package classification module is used to mark the target chip package with abnormal physical detection as the first package and the target chip package with abnormal electrical performance detection as the second package;

[0094] The result judgment module is used to judge whether each output result of a plurality of chip tests recorded by the target chip package with the same printed information is the same for the first package and the second package;

[0095] The quality assessment processing module is used to perform quality assessment processing based on the output results of the result judgment module.

[0096] The result judgment module includes a target printing information marking unit and a recognizable feature analysis unit;

[0097] The target printed information marking unit is used for recording, on a target chip package of the same printed information, that each output result of a plurality of chip tests is the same as that of the first package and the second package, marking the corresponding printed information as the target printed information;

[0098] The identifiable feature analysis unit is used to obtain detection data of chip detection records of the target chip package corresponding to the printed information, and analyze the identifiable features of the target chip package corresponding to the same printed information when performing various types of detection based on the detection data.

[0099] The identifiable feature analysis unit includes a package marking unit, a detection data classification unit and an identifiable feature output unit;

[0100] The package marking unit is used to extract the target chip package marked as the first package but not marked as the second package as the inspection package, and the remaining target chip packages in the first package excluding the inspection package as the control package;

[0101] The test data classification unit is used to obtain the distance between the pins on both sides of the package, the number of pin pairs and the height of the package recorded in the physical test, as well as the type of electrical performance parameters and the corresponding parameter values;

[0102] The identifiable feature output unit is used to output the identifiable features of physical detection and the identifiable features of electrical performance detection.

[0103] The detection data classification unit includes a structural feature index calculation unit, a first sequence generation unit, an abnormality rate calculation unit and an average difference calculation unit;

[0104] The structural characteristic index calculation unit is used to calculate the structural characteristic index of each package under investigation;

[0105] The first sequence generating unit is used to classify the investigated packages whose structural characteristic index difference in the investigated packages is less than or equal to the difference threshold value as the same type of investigated packages, and sort the investigated packages from small to large according to the values ​​of the structural characteristic index corresponding to each type of investigated packages, and generate a first sequence;

[0106] The abnormality rate calculation unit is used to extract the test results of various types of examined packages in the first sequence in several historical chip test records, and calculate the abnormality rate of various types of examined packages;

[0107] The average difference calculation unit is used to calculate the average difference of all types of electrical performance parameters corresponding to each type of package under investigation.

[0108] The quality assessment processing module includes a quality failure marking unit, a package rejection unit and a detection response unit;

[0109] The unqualified quality marking unit is used for marking the real-time package with abnormal output result as unqualified quality after any type of detection is performed when there is a real-time package with the same printed information as the target for chip detection, and after removing the unqualified real-time package, the remaining real-time package is subjected to another type of detection;

[0110] The package rejection unit is used to reject packages that do not meet the detection requirements;

[0111] The detection response unit is used to provide quality warning reminders for different packages.

[0112] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0113] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A chip quality assessment method based on data analysis, characterized in that: The analysis steps include: Step S100: marking the chip after packaging and printing information detection qualified package as the target chip package; Obtaining an output result of chip testing of a target chip package, wherein the chip testing includes physical testing and electrical performance testing; the output result records target chip packages with abnormal physical testing and target chip packages with abnormal electrical performance testing; Step S200: Mark the target chip package with physical test abnormality as the first package, and the target chip package with electrical performance test abnormality as the second package; if the target chip package with the same printed information records the output results of several chip tests each time, and the first package and the second package are the same, then execute step S300; otherwise, execute step S400; Step S300: marking the corresponding printing information as the target printing information, and when there is a real-time package body with the same target printing information for chip detection, after any type of detection is performed, marking the real-time package body with abnormal output result as unqualified quality, and after removing the real-time package body with unqualified quality, performing another type of detection on the remaining real-time package bodies; Step S400: acquiring the test data of the chip test record of the target chip package corresponding to the printed information, and analyzing the identifiable features of the target chip package corresponding to the same printed information when performing various types of tests based on the test data; The analysis of the identifiable features of the target chip package corresponding to the same printed information during various types of inspections includes the following specific steps: Extracting a target chip package body marked as the first package body but not marked as the second package body as an inspection package body, and removing the inspection package body from the first package body and the remaining target chip packages as control packages; The test data refers to the distance L1 between the pins on both sides of the package, the number of pin pairs M1 and the height H1 of the package recorded in the physical test. The height of the package refers to the vertical distance from the package front surface to the lowest point of all the pins of the package; the pin pair refers to the number of pins on both sides of the chip package when there is no pin missing, and one pin pair records two pins on the same horizontal line; using the formula: T=[M2 / M1]*[|L2-L1| / L2+|H2-H1| / H2] Calculate the structural characteristic index T of each package under investigation, M2 represents the number of pin pairs when the target chip package with the same printed information passes the physical test, L2 represents the average pin distance when the target chip package with the same printed information passes the physical test, and H2 represents the average package height when the target chip package with the same printed information passes the physical test; Classify the investigated packages whose structural characteristic index difference is less than or equal to the difference threshold value into the same category of investigated packages, and sort them from small to large according to the values ​​of the structural characteristic index corresponding to each category of investigated packages, and generate a first sequence; Extract the test results of various packages in the first sequence in several historical chip test records, using the formula: E=(e1 / e2)*(s1 / s2), Calculate the abnormality rate E of various types of inspected packages, where e1 represents the number of inspected packages of the corresponding category in the same chip inspection, and e2 represents the total number of inspected packages recorded in the same chip inspection; s1 represents the number of inspection events recorded for the inspected packages of the corresponding category in s2 chip inspections, and s2 represents the total number of chip inspection events in the historical record of the chip packages with the same printing information; Construct a variable package based on the structural feature index and the abnormality rate of a type of inspected package, calculate the correlation coefficient r based on the variable package, and set the correlation coefficient threshold r0. When r < r0, obtain the abnormality rate e0 corresponding to the minimum value tmin of the structural feature index of the control package record, and mark the inspected packages of the category corresponding to the minimum value where the abnormality rate E of the inspected package record is greater than the abnormality rate e0 as the first effective packages, and extract the structural feature index T recorded by the first effective packages as the recognizable feature of the target chip package during physical detection. When r ≥ r0, output the structural feature index recorded for each type of inspected package as the recognizable feature of the target chip package during physical detection. Step S500: Based on the recognizable features of the target chip package that matches the same printing information for the real-time package during various types of inspections, perform chip quality assessment processing on the real-time package based on the recognizable features.

2. The chip quality assessment method based on data analysis according to claim 1, characterized in that: The analysis of the recognizable features of the target chip package corresponding to the same printing information during various types of inspections further includes the following specific steps: The inspection data also includes the type of electrical performance parameters and the corresponding parameter values. Obtain the difference d between the electrical performance parameter values recorded for the inspected package during electrical performance inspection and the electrical performance parameter values corresponding to the second package record, d = q1 - q2, where q1 represents the electrical performance parameter value recorded by the second package, and q2 represents the parameter value of the electrical performance parameter of the same type corresponding to the inspected package. Calculate the average difference D for all types of electrical performance parameters corresponding to each type of inspected package, D = (1 / g)∑d0, where d0 represents the difference in the normalized electrical performance parameter values, and g represents the total number of types of electrical performance parameters. Extract the minimum value Dmin of the average differences of various types of inspected packages recorded in s2 chip inspection events for the chip packages with the same printing information as the target difference; mark the minimum value of the target difference corresponding to all types of inspected packages as the recognizable feature of the target chip package during electrical performance inspection.

3. The chip quality assessment method based on data analysis according to claim 2, characterized in that: The chip quality assessment processing of the real-time package based on the recognizable features includes the following specific steps: When the real-time package first undergoes physical detection, obtain the distance between the pins on both sides of the package, the number of pin pairs, and the height of the package recorded by the real-time package, and substitute them into the structural feature index calculation formula to calculate the structural feature index T0 of each real-time package. When r < r0, extract the real-time packages with the real-time structural feature index T0 less than the recognizable feature as the first pin quality packages; when r ≥ r0, regard the real-time packages with the difference from the recognizable feature less than the difference threshold as the first pin quality packages. And among the real-time packages that have completed the physical inspection and output abnormal detection, the first pin quality package is eliminated, and the remaining real-time packages are no longer subjected to electrical performance inspection; and the first pin quality package is given a quality warning reminder; When the real-time package body is first subjected to electrical performance detection, the electrical performance parameter type and the corresponding real-time parameter value recorded by the real-time package body are obtained, the real-time average difference D0 is calculated, and the real-time package body whose real-time average difference D0 is greater than or equal to the target difference Dmin and less than the second difference threshold is extracted as the second pin quality package body, and the second difference threshold is the maximum value of the average difference of the electrical performance parameter value recorded by the corresponding chip package body when the output result is qualified quality during the electrical performance detection; Issue a quality warning reminder for the second pin quality package, and remove the second pin quality package from the packages with abnormal real-time detection results, and remove the remaining packages without physical inspection; After the detection is normal and the quality warning reminder is restored, the first pin quality package and the second pin quality package are subjected to electrical performance detection again.

4. A chip quality assessment system based on data analysis, such as a chip quality assessment method based on data analysis according to any one of claims 1 to 3, characterized in that: It includes a target chip package marking module, a chip detection result recording module, a package classification module, a result judgment module and a quality assessment processing module; The target chip package marking module is used to mark the package that has passed the printing information detection after the chip is packaged as the target chip package; The chip detection result recording module is used to obtain the output result of chip detection of the target chip package, wherein the chip detection includes physical detection and electrical performance detection; the output result records the target chip package with abnormal physical detection and the target chip package with abnormal electrical performance detection; The package classification module is used to mark the target chip package with abnormal physical performance detection as the first package and the target chip package with abnormal electrical performance detection as the second package; The result judgment module is used to judge whether each output result of several chip tests recorded by the target chip package with the same printed information is the same for the first package and the second package; The result judgment module includes a target printing information marking unit and a recognizable feature analysis unit; The target printing information marking unit is used for marking the corresponding printing information as the target printing information when each output result of several chip tests recorded in the target chip package of the same printing information is the same as that of the first package and the second package; The identifiable feature analysis unit is used to obtain the detection data of the chip detection record of the target chip package corresponding to the printed information, and analyze the identifiable features of the target chip package corresponding to the same printed information when performing various types of detection based on the detection data; The identifiable feature analysis unit includes a package marking unit, a detection data classification unit and an identifiable feature output unit; The package marking unit is used to extract the target chip package marked as the first package but not marked as the second package as the inspection package, and the remaining target chip packages in the first package excluding the inspection package as the control package; The detection data classification unit is used to obtain the distance between the pins on both sides of the package, the number of pin pairs and the height of the package recorded in the physical detection, as well as the type of electrical performance parameters and the corresponding parameter values; The identifiable feature output unit is used to output the identifiable features of physical detection and the identifiable features of electrical performance detection; The quality assessment processing module is used to perform quality assessment processing based on the output result of the result judgment module.

5. A chip quality assessment system based on data analysis according to claim 4, characterized in that: The detection data classification unit includes a structural feature index calculation unit, a first sequence generation unit, an abnormality rate calculation unit and an average difference calculation unit; The structural feature index calculation unit is used to calculate the structural feature index of each package under investigation; The first sequence generating unit is used to classify the investigated packages whose structural feature index difference in the investigated packages is less than or equal to the difference threshold as the same type of investigated packages, and sort the investigated packages from small to large according to the values ​​of the structural feature index corresponding to each type of investigated packages, and generate a first sequence; The abnormality rate calculation unit is used to extract the test results of various types of packages under investigation in the first sequence in several historical chip test records, and calculate the abnormality rate of various types of packages under investigation; The average difference calculation unit is used to calculate the average difference of all types of electrical performance parameters corresponding to each type of package under investigation.

6. A chip quality assessment system based on data analysis according to claim 5, characterized in that: The quality assessment processing module includes a quality failure marking unit, a package rejection unit and a detection response unit; The unqualified quality marking unit is used for marking the real-time package with abnormal output result as unqualified quality after any type of detection is performed when there is a real-time package with the same printed information as the target for chip detection, and after removing the unqualified real-time package, performing another type of detection on the remaining real-time packages; The package rejection unit is used to reject packages that do not meet the detection requirements; The detection response unit is used to provide quality warning reminders for different packages.

Images