Quality evaluation method for storage chip

Abstract

The invention relates to the technical field of data analysis, and particularly discloses a storage chip quality evaluation method, which comprises the following steps of S1, determining quality parameters of a target chip, and obtaining a quality score of the target chip; s2, generating coordinate points, and fitting the coordinate points to obtain a fitted curve; determining an abnormal proportion according to the fitting curve, and determining a judgment value according to the total abnormal proportion and the fitting curve; calculating an average judgment value, and determining an abnormal production line; and S3, dividing the production serial number interval to obtain production serial number sub-intervals, and carrying out sampling inspection on the storage chips in the production serial number sub-intervals. According to the invention, the detection efficiency can be improved, and omission of quality problem chips is reduced.

Description

Technical Field

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

[0002] A memory chip is a semiconductor device used to store data. Its main function is to save programs and data, and to access programs or data at high speed and automatically during computer operation. As a core component of modern electronic devices, the quality of memory chips is directly related to the performance of the device and user experience.

[0003] The quality assessment of memory chips requires professional technicians and testing equipment. In the actual production process, due to the large number of memory chips, it will consume a lot of manpower and time to conduct quality assessment on each chip one by one. In order to reduce costs and improve overall testing efficiency, random sampling is usually used to determine the quality of batch chips. However, conventional random sampling may miss a large number of memory chips with quality problems, affecting the reliability of the entire product batch. Summary of the invention

[0004] The purpose of the present invention is to provide a method for evaluating the quality of a memory chip to solve the following technical problems:

[0005] Conventional sampling methods may miss a large number of memory chips with quality problems, affecting the reliability of the entire product batch.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for evaluating the quality of a memory chip comprises the following steps:

[0008] S1: Obtain a sampling record of a memory chip, mark the memory chip in the sampling record as a target chip, determine the quality parameters of the target chip, the quality parameters include a read delay H1, a write delay H2, and an erase time H3, and determine a quality score D=H1+H2+H3 of the target chip;

[0009] S2: Generate coordinate points (Bi, Di), where Bi represents the production serial number of the i-th target chip, and Di represents the quality score of the i-th target chip, and fit the coordinate points to obtain a fitting curve f(x), where x represents the production serial number;

[0010] The curve part of the a-th fitting curve that is above the b-th fitting curve is marked as an abnormal curve. The proportion of the abnormal curve to the a-th fitting curve is counted and used as the abnormal proportion. The total abnormal proportion A is calculated and the judgment value C is calculated. The calculation formula is:

[0011]

[0012] Among them, Asta and Aend represent the minimum production serial number and the maximum production serial number of the target chip respectively;

[0013] Calculate the average judgment value CPJ, and when the screening value ΔC=C-CPJ≥ΔCys, take the production line corresponding to the screening value C as an abnormal production line;

[0014] S3: Calculate the sampling length BCS represents the preset initial sampling length, β represents the preset first correction coefficient and β>1, It means that C*β is rounded down, and the production serial number interval [BDQ, BYC] is divided into n production serial number sub-intervals of the same length, BDQ represents the current production serial number, the pending production serial number BYC = BDQ ​​+ BCJ, n is a preset number, and the storage chips corresponding to the production serial number sub-interval are sampled according to the preset sampling ratio.

[0015] As a further solution of the present invention: in step S3, the process of sampling the memory chips corresponding to the production serial number sub-intervals specifically includes:

[0016] The production serial number sub-intervals are sorted according to the time axis order to obtain a sub-interval sorting, wherein the sampling ratio CJk=1-μ*k / n corresponding to the k-th production serial number sub-interval in the sorting, k<n, μ is a preset coefficient and μ<1, and CJ1*(BYC-BDQ) / n memory chips are randomly selected in the first production serial number sub-interval for testing;

[0017] Determine the ratio F of the number of chips with a quality score higher than the quality score threshold to the number of random inspections CJ1*(BYC-BDQ) / n corresponding to the first production serial number sub-interval, and update the random inspection ratio CJ2 corresponding to the second production serial number sub-interval. The updated value is CJ2*F*η, where η represents a preset update coefficient;

[0018] Repeat the above steps until the sampling inspection of the nth production serial number sub-interval is completed.

[0019] As a further solution of the present invention: in the step S2, when the sampling length BCJ is greater than or equal to the preset production serial number difference, an early warning message is sent for reporting.

[0020] As a further solution of the present invention: the step S2 further includes the following steps:

[0021] The ratio of the number of abnormal proportions of 0 to the total number of abnormal proportions is counted. When the ratio is less than the preset ratio threshold, the corresponding production line is not regarded as an abnormal production line.

[0022] As a further solution of the present invention: in the step S2, when the judgment value C≥Cys, the corresponding production line is marked as an abnormal production line, and Cys represents a preset judgment value threshold.

[0023] As a further solution of the present invention: in the step S2, the process of calculating the average judgment value further includes the following steps:

[0024] The variance of the judgment value is calculated, and when the variance is greater than a preset variance threshold, the minimum judgment value is removed and the variance is calculated again until the variance is less than or equal to the variance threshold.

[0025] Beneficial effects of the present invention: In this solution, by determining the quality parameters of each chip and scoring the quality of each chip, the quality of each chip can be evaluated more objectively and accurately, reducing the interference of subjective factors. After establishing a scoring system, analysis and decision-making can be made based on data to improve the scientificity and accuracy of detection; thereafter, a curve is fitted through the relationship between the production serial number and the quality score, and a judgment value is determined according to the fitting curve to make a horizontal comparison of different production lines, so as to identify production lines with poor overall quality performance. There is no need to check all production lines one by one, and abnormal production lines that need to be focused on can be quickly screened out through A and C values, saving time and resources; it is worth noting that one production serial number corresponds to one time, and one production serial number corresponds to one produced The chips produced by the production line, therefore, the fitting curve can be regarded as the curve of the quality score of a single production line changing over time. It is worth noting that the production serial number in this scheme is a virtual production serial number. In actual situations, the production serial numbers of different memory chips are different, while in this scheme, the production serial numbers of the m-th memory chips produced on different production lines are the same (i.e., virtual production serial numbers), and m is an arbitrary real number; afterwards, the potential problem serial number interval (i.e., the production serial number interval [BDQ, BYC]) is predicted, and by dividing the sub-intervals and dynamically adjusting the sampling ratio, it is ensured that the high-risk sub-intervals are detected more frequently, reducing the probability of missing defective chips, refining the sampling process, making the detection more targeted, saving time and resources, and improving the detection coverage. The present invention not only effectively reduces labor and time costs, but also significantly improves detection efficiency and accuracy, reduces the omission of chips with quality problems, thereby improving the reliability and market competitiveness of the entire product batch. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention will be further described below in conjunction with the accompanying drawings.

[0027] Figure 1 It is a flow chart of a method for evaluating the quality of a memory chip according to the present invention. DETAILED DESCRIPTION

[0028] 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.

[0029] See also Figure 1 As shown, the present invention is a method for evaluating the quality of a memory chip, comprising the following steps:

[0030] S1: Obtain a sampling record of a memory chip, mark the memory chip in the sampling record as a target chip, determine the quality parameters of the target chip, the quality parameters include a read delay H1, a write delay H2, and an erase time H3, and determine a quality score D=H1+H2+H3 of the target chip;

[0031] S2: Generate coordinate points (Bi, Di), where Bi represents the production serial number of the i-th target chip, and Di represents the quality score of the i-th target chip, and fit the coordinate points to obtain a fitting curve f(x), where x represents the production serial number;

[0032] The curve part of the a-th fitting curve that is above the b-th fitting curve is marked as an abnormal curve. The proportion of the abnormal curve to the a-th fitting curve is counted and used as the abnormal proportion. The total abnormal proportion A is calculated and the judgment value C is calculated. The calculation formula is:

[0033]

[0034] Among them, Asta and Aend represent the minimum production serial number and the maximum production serial number of the target chip respectively;

[0035] Calculate the average judgment value CPJ, and when the screening value ΔC=C-CPJ≥ΔCys, take the production line corresponding to the screening value C as an abnormal production line;

[0036] S3: Calculate the sampling length BCS represents the preset initial sampling length, β represents the preset first correction coefficient and β>1, It means that C*β is rounded down, and the production serial number interval [BDQ, BYC] is divided into n production serial number sub-intervals of the same length, BDQ represents the current production serial number, the pending production serial number BYC = BDQ ​​+ BCJ, n is a preset number, and the storage chips corresponding to the production serial number sub-interval are sampled according to the preset sampling ratio.

[0037] It should be noted that by determining the quality parameters of each chip and scoring the quality of each chip, the quality of each chip can be evaluated more objectively and accurately, reducing the interference of subjective factors. After establishing a scoring system, analysis and decision-making can be made based on the data to improve the scientificity and accuracy of the detection. Afterwards, a curve is fitted based on the relationship between the production serial number and the quality score, and the judgment value is determined based on the fitting curve to make a horizontal comparison of different production lines, so as to identify the production lines with poor overall quality performance. There is no need to check all production lines one by one, and the A and C values ​​can be used to quickly screen out abnormal production lines that need to be focused on, saving time and resources. It is worth noting that a production serial number corresponds to a time, and a production serial number corresponds to a chip produced by the production line. Chip, therefore, the fitting curve can be regarded as the curve of the quality score of a single production line changing over time. It is worth noting that the production serial number in this scheme is a virtual production serial number. In actual situations, the production serial numbers of different memory chips are different. In this scheme, the production serial numbers of the m-th memory chips produced on different production lines are the same (that is, the virtual production serial number), and m is an arbitrary real number; then, the potential problem serial number interval (that is, the production serial number interval [BDQ, BYC]) is predicted, and by dividing the sub-intervals and dynamically adjusting the sampling ratio, it is ensured that the high-risk sub-intervals are tested more frequently, reducing the probability of missing defective chips, refining the sampling process, making the detection more targeted, saving time and resources, and improving the detection coverage.

[0038] In another preferred embodiment of the present invention, in step S3, the process of sampling the memory chips corresponding to the production serial number sub-intervals specifically includes:

[0039] The production serial number sub-intervals are sorted according to the time axis order to obtain a sub-interval sorting, wherein the sampling ratio CJk=1-μ*k / n corresponding to the k-th production serial number sub-interval in the sorting, k<n, μ is a preset coefficient and μ<1, and CJ1*(BYC-BDQ) / n memory chips are randomly selected in the first production serial number sub-interval for testing;

[0040] Determine the ratio F of the number of chips with a quality score higher than the quality score threshold to the number of random inspections CJ1*(BYC-BDQ) / n corresponding to the first production serial number sub-interval, and update the random inspection ratio CJ2 corresponding to the second production serial number sub-interval. The updated value is CJ2*F*η, where η represents a preset update coefficient;

[0041] Repeat the above steps until the sampling inspection of the nth production serial number sub-interval is completed.

[0042] It is worth noting that by updating the sampling ratio of the next sub-interval after sampling each sub-interval, the sampling strategy is not rigid, and a set of automatic adjustment mechanism is formed; compared with the fixed-ratio sampling scheme, this method can allocate more sampling resources to intervals that need more attention, so as to improve the overall detection efficiency and avoid excessive or insufficient sampling.

[0043] In another preferred embodiment of the present invention, in the step S2, when the sampling length BCJ is greater than or equal to the preset production serial number difference, an early warning message is sent for reporting.

[0044] In another preferred embodiment of the present invention, the step S2 further includes the following steps:

[0045] The ratio of the number of abnormal proportions of 0 to the total number of abnormal proportions is counted. When the ratio is less than the preset ratio threshold, the corresponding production line is not regarded as an abnormal production line.

[0046] In another preferred embodiment of the present invention, in the step S2, when the judgment value C≥Cys, the corresponding production line is marked as an abnormal production line, and Cys represents a preset judgment value threshold.

[0047] In another preferred embodiment of the present invention, in the step S2, the process of calculating the average judgment value further includes the following steps:

[0048] The variance of the judgment value is calculated, and when the variance is greater than a preset variance threshold, the minimum judgment value is removed and the variance is calculated again until the variance is less than or equal to the variance threshold.

[0049] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for evaluating the quality of a memory chip, characterized in that: The following steps are involved: S1: Obtain a sampling record of a memory chip, mark the memory chip in the sampling record as a target chip, determine the quality parameters of the target chip, the quality parameters include a read delay H1, a write delay H2, and an erase time H3, and determine a quality score D=H1+H2+H3 of the target chip; S2: Generate coordinate points (Bi, Di), where Bi represents the production serial number of the i-th target chip, and Di represents the quality score of the i-th target chip, and fit the coordinate points to obtain a fitting curve f(x), where x represents the production serial number; The curve part of the a-th fitting curve that is above the b-th fitting curve is marked as an abnormal curve. The proportion of the abnormal curve to the a-th fitting curve is counted and used as the abnormal proportion. The total abnormal proportion A is calculated and the judgment value C is calculated. The calculation formula is: Among them, Asta and Aend represent the minimum production serial number and the maximum production serial number of the target chip respectively; Calculate the average judgment value CPJ, and when the screening value ΔC=C-CPJ≥ΔCys, take the production line corresponding to the screening value C as an abnormal production line; S3: Calculate the sampling length BCS represents the preset initial sampling length, β represents the preset first correction coefficient and β>1, It means that C*β is rounded down, and the production serial number interval [BDQ, BYC] is divided into n production serial number sub-intervals of the same length, BDQ represents the current production serial number, the pending production serial number BYC = BDQ ​​+ BCJ, n is a preset number, and the storage chips corresponding to the production serial number sub-interval are sampled according to the preset sampling ratio.

2. A method for evaluating the quality of a memory chip according to claim 1, characterized in that: In step S3, the process of sampling the memory chips corresponding to the production serial number sub-intervals specifically includes: The production serial number sub-intervals are sorted according to the time axis order to obtain a sub-interval sorting, wherein the sampling ratio CJk=1-μ*k / n corresponding to the k-th production serial number sub-interval in the sorting, k<n, μ is a preset coefficient and μ<1, and CJ1*(BYC-BDQ) / n memory chips are randomly selected in the first production serial number sub-interval for testing; Determine the ratio F of the number of chips with a quality score higher than the quality score threshold to the number of random inspections CJ1*(BYC-BDQ) / n corresponding to the first production serial number sub-interval, and update the random inspection ratio CJ2 corresponding to the second production serial number sub-interval. The updated value is CJ2*F*η, where η represents a preset update coefficient; Repeat the above steps until the sampling inspection of the nth production serial number sub-interval is completed.

3. The method for evaluating the quality of a memory chip according to claim 1, characterized in that: In the step S2, when the sampling length BCJ is greater than or equal to the preset production serial number difference, an early warning message is sent for reporting.

4. The method for evaluating the quality of a memory chip according to claim 1, characterized in that: The step S2 further includes the following steps: The ratio of the number of abnormal proportions of 0 to the total number of abnormal proportions is counted. When the ratio is less than the preset ratio threshold, the corresponding production line is not regarded as an abnormal production line.

5. The method for evaluating the quality of a memory chip according to claim 1, characterized in that: In the step S2, when the judgment value C≥Cys, the corresponding production line is marked as an abnormal production line, and Cys represents a preset judgment value threshold.

6. The method for evaluating the quality of a memory chip according to claim 1, characterized in that: In the step S2, the process of calculating the average judgment value further includes the following steps: The variance of the judgment value is calculated, and when the variance is greater than a preset variance threshold, the minimum judgment value is removed and the variance is calculated again until the variance is less than or equal to the variance threshold.

Images