A method and system for extracting test data
By extracting and processing associated test items from wafer test data, the problem of users struggling to obtain effective information from large amounts of data is solved, achieving intuitive presentation and readability of test results.
Patent Information
- Application Number
- CN202211666906.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-21
- Publication Date
- 2026-02-17
- Estimated Expiration
- 2042-12-21
AI Technical Summary
During wafer testing, users often struggle to extract useful information directly from large amounts of test data, and existing technologies are inadequate for effectively extracting and processing this data.
By extracting associated test items, preprocessing and filtering noise multiple times, calculating the mean and standard deviation, and generating a mapping table between associated test items and test value distribution indicators, the effective data of the test results are presented.
It enables users to intuitively grasp the effective data of test results, improves the readability and understanding of test results, and simplifies the data analysis process.
Smart Images
Figure CN115982269B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of test data processing technology, and in particular to a method and system for extracting test data. Background Technology
[0002] Wafer testing refers to probing each die on a wafer to test its functionality and electrical characteristics. Each die on a wafer is a chip, and chip testing generates a large amount of data. Specifically, a wafer consists of multiple dies, each with multiple test parameters, and each test parameter corresponds to multiple functional or electrical characteristic tests. During testing, each parameter is tested multiple times, resulting in a vast amount of test data. Faced with this massive amount of test data, users often find it difficult to directly extract useful information from the test results. Summary of the Invention
[0003] To address the aforementioned technical problems, the present invention adopts the following technical solution: a method for extracting test data, the extraction method comprising: S100, acquiring test data Data, wherein Data includes N test items E = {E1, E2, ..., E...} i ,…,E N}, E i For the i-th test item, the value of i ranges from 1 to N; E i Including the qth test metric TR q The name of the f-th test object is TA. f and TR at M test points q The test value.
[0004] S300, extract the test object name and test index from E to obtain P kinds of related test items AT = {AT1,AT2,…,AT...} p ,…,AT P}, AT p ={TA f TR q},AT p Let p be the p-th association test item, where p ranges from 1 to p.
[0005] S500 preprocesses all test values associated with each associated test item in AT, filtering out invalid data to obtain a preprocessed set of P valid test values.
[0006] S700, filter the noise in the P types of valid test value sets respectively to obtain the P types of target test value sets GU={GU1,GU2,…,GU p ,…,GU P}, GU p For AT pThe associated set of target test values; from which GU is obtained p The steps include:
[0007] S710, calculate the mean μ and standard deviation σ of the effective test value set after the (r-1)th noise filtering, and obtain the expected range (μ-hσ, μ+hσ) of the rth filtering based on μ and σ, where r is greater than 1 and h is a preset coefficient.
[0008] S730, when all test values in the effective test value set after the (r-1)th noise filtering are within the expected range, the target test value set GU is obtained. p Otherwise, filter out test values outside the expected range to obtain the set of valid test values after the r-th noise filtering; execute S710 and S730 again based on the set of valid test values after the r-th noise filtering.
[0009] S900, obtain the test value distribution index DIS = {DIS1, DIS2, ..., DIS} of P related test items based on GU. p ,…,DIS P}, DIS p For AT p The test value distribution index, DIS p Including GU p Trend metric ME p and GU p The standard deviation; generate a mapping table of associated test items and test value distribution indicators based on AT and DIS.
[0010] Compared with the prior art, the present invention has significant advantages. Through the above technical solution, the test data extraction method and system provided by the present invention achieve considerable technical progress and practicality, and have broad industrial application value. It has at least the following advantages:
[0011] This invention extracts associated test items from test entries and preprocesses all test value data associated with each type of associated test item to obtain a valid test value set. Noise in the valid test value set is filtered multiple times according to the expected range. When all test values are within the expected range, the target test value set is obtained. The trend measure and standard deviation of the target test value set are calculated. A mapping relationship table is generated based on the associated test items, trend measure, and standard deviation to present the test results of each associated test item in the test data. This allows users to intuitively grasp the valid data of the test results and thus understand the trend of the test results. Attached Figure Description
[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0013] Figure 1 This is a flowchart of a test data extraction method provided in an embodiment of the present invention. Detailed Implementation
[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0015] Please see Figure 1 It illustrates a flowchart of a test data extraction method, which includes:
[0016] S100, acquire test data Data, wherein Data includes N test items E = {E1, E2, ..., E...} i ,…,E N}, E i For the i-th test item, the value of i ranges from 1 to N; E i Including the qth test metric TR q The name of the f-th test object is TA. f and TR at M test points q The test value.
[0017] It should be noted that the test point is the location of each die probe on the wafer, and a test value is obtained by testing each test point. It can also be different test points on other test components.
[0018] Optionally, the test parameters can be electrical parameters such as voltage, current, or impedance.
[0019] Optionally, the test object name is a unique identifier for each wafer.
[0020] S300, extract the test object name and test index from E to obtain P kinds of related test items AT = {AT1,AT2,…,AT...} p ,…,AT P}, AT p ={TA f TR q},ATp Let p be the p-th association test item, where p ranges from 1 to p.
[0021] It should be understood that a test item of the same type can be associated with multiple test items, but a test item cannot be associated with multiple test items.
[0022] As a preferred embodiment, the step of extracting the test object name and test index from E to obtain P kinds of associated test items AT includes:
[0023] S310: Extract the test object name from each test item based on the first specified line, and extract the test metric from each test item based on the second specified line, to obtain candidate associated test items consisting of the test object name and test metric corresponding to each test item. It should be noted that the first and second specified lines are specified by the user.
[0024] S320, compare whether there are identical candidate association test items among the E candidate association test items, and remove the identical candidate association test items to obtain P kinds of association test items AT.
[0025] Those skilled in the art will understand that any method for extracting associated test items in the prior art falls within the protection scope of this invention.
[0026] It should be understood that, in addition to the test object name and test metrics, associated test items may also include other test parameters specified by the user in the test entries. As an example, other test parameters may be the test type of the test metrics, etc. In this case, associated test items are categorized according to the test object name, test metrics, and other test parameters.
[0027] S500 preprocesses all test values associated with each associated test item in AT, filtering out invalid data to obtain a preprocessed set of P valid test values.
[0028] Preprocessing is used to filter out invalid test values and retain valid test values.
[0029] Optional, the S500 includes:
[0030] S510, obtain the minimum threshold of the test metric for each associated test item in AT {THN1,THN2,…,THN p ,…,THN P} and the maximum threshold {THX1,THX2,…,THX} p ,…,THX P}, where THN p For AT p China TR q Minimum threshold, THX q For ATp China TR q The maximum threshold.
[0031] S530, AT p All associated test values are respectively related to THN p and THX p Compare and remove those less than THN p And greater than THX p The test value is used to obtain AT. p The set of valid test values associated with the data.
[0032] In a preferred embodiment, S500 includes:
[0033] S520, obtain the absolute threshold of the test metric {TH1,TH2,…,TH} for each associated test item in AT. p ,…,TH P}, where TH p For AT p China TR q The absolute threshold.
[0034] S540, AT p The absolute value of all associated test values and TH p Compare and remove those greater than TH p The test value is used to obtain AT. p The set of valid test values associated with the data.
[0035] It should be noted that data with test values greater than the absolute threshold are considered invalid and will not be included in subsequent calculations. For example, voltage test values greater than the absolute threshold of 10 are considered invalid. Preprocessing the test values removes invalid data, preventing pre-assigned invalid values from being used in the calculation of the test distribution index, which could lead to significant deviations in the final results. This ensures that the final test distribution index accurately reflects the overall distribution.
[0036] Those skilled in the art will understand that any prior art method for filtering invalid data by threshold falls within the protection scope of this invention.
[0037] S700, filter the noise in the P types of valid test value sets to obtain the P types of target test value sets GU={GU1,GU2,…,GU p ,…,GU P}, GU p For AT p The associated set of target test values. Among them, obtaining GU... p The steps include:
[0038] S710, calculate the mean μ and standard deviation σ of the effective test value set after the (r-1)th noise filtering. Based on μ and σ, obtain the expected range (μ-hσ, μ+hσ) for the r-th filtering, where r is greater than 1 and h is a preset coefficient. Optionally, h is 3 or 5. Preferably, h is 5.
[0039] S730, when all test values in the effective test value set after the (r-1)th noise filtering are within the expected range, the target test value set GU is obtained. p Otherwise, filter out test values outside the expected range to obtain the set of valid test values after the r-th noise filtering; execute S710 and S730 again based on the set of valid test values after the r-th noise filtering.
[0040] It should be noted that by calculating the expected range of each set of valid test values, test values that are noise in the set of valid test values can be filtered out. Then, the expected range of the filtered data is calculated again, and so on, until all test values are within the expected range, thus obtaining a set of valid test values without noise.
[0041] As a preferred embodiment, S700 further includes:
[0042] S810, calculate the proportion of the total number of test values in the set of P target test values to the total number of test values contained in Data.
[0043] S820: When the percentage is less than the preset percentage threshold, a preset prompt icon is generated.
[0044] Preferably, the preset percentage threshold is 1 / 3.
[0045] It should be noted that when the percentage is less than the preset percentage threshold, it means that the original data contains a lot of invalid and noisy data, resulting in fewer valid test values.
[0046] Preferably, the preset prompt is a preset text message. Specifically, the preset text message may be a message suggesting that the user has insufficient valid test data and recommends retesting.
[0047] S900, obtain the test value distribution index DIS = {DIS1, DIS2, ..., DIS} of P related test items based on GU. p ,…,DIS P}, DIS p For AT p The test value distribution index, DIS p Including GU p Trend metric ME p and GU pThe standard deviation; generate a mapping table of associated test items and test value distribution indicators based on AT and DIS.
[0048] Understandably, the trend metric is used to reflect the distance from GU. p Proximity to the center.
[0049] Optional, trend metric ME p For GU p The median or mean. Those skilled in the art will understand that any existing technology used to calculate distance GU... p Methods for measuring the trend of proximity to the center all fall within the protection scope of this invention.
[0050] As a preferred embodiment, the ME p For GU p of the median.
[0051] As a preferred embodiment, the DIS p It also includes the noise item NOI. p The NOI p Including obtaining GU p The number of noises filtered out, L p and noise deviation nod p Optional, nod p For L p Noise and GU p The average difference between the means, or nod p By calculating L p Noise and GU p The variance between the means is obtained. Specifically, when filtering noise, the filtered noise data is stored in a noise set associated with each test item, and the number of elements in the noise set is the number of noise items. By generating noise items, users can intuitively grasp the overall situation of the actual test data.
[0052] In summary, this embodiment of the invention extracts associated test items from the test entries and preprocesses all test value data associated with each type of associated test item to obtain a valid test value set. Noise in the valid test value set is filtered multiple times according to the expected range. When all test values are within the expected range, a target test value set is obtained. The trend measure and standard deviation of the target test value set are calculated. A mapping table is generated based on the associated test items, trend measure, and standard deviation to present the test results for each associated test item in the test data. Compared to the original test results themselves, this allows users to intuitively grasp the valid data of the test results, thereby understanding the overall trend of the test results and improving the readability of the test results.
[0053] Based on the same inventive concept as the above-described method embodiments, this invention also provides a test data extraction system. The system includes a processor and a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores at least one instruction or at least one program. The at least one instruction or the at least one program is loaded and executed by the processor to implement a test data extraction method. One such test data extraction method has been described in detail in the above embodiments and will not be repeated here.
[0054] While specific embodiments of the invention have been described in detail by way of example, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. It should also be understood that various modifications can be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.
Claims
1. A method of extracting test data, characterized by, The extraction method comprises: S100, acquire test data Data, wherein Data includes N test items E={E1,E2,…,E...} i ,…,E N }, E i For the i-th test item, the value of i ranges from 1 to N; E i Including the qth test metric TR q The name of the f-th test object is TA. f and TR at M test points q The test value; S300, extracting the test object name and test index in E to obtain P associated test items AT={AT1, AT2,..., AT p ,…,AT P} and AT p ={TA f ,TR q}, AT p is the pth associated test item, the value range of p is 1 to P; wherein P is the total number of associated test items, each associated test item is uniquely determined by a test object name and a test index, and P is less than or equal to N; S500, preprocessing all test values associated with each associated test item in the AT, filtering out invalid data to obtain a set of P pre-processed valid test values; S700, filtering the noise in each of the P sets of effective test values respectively to obtain P sets of target test values GU={GU1, GU2, …, GU p P P}, GU p is the target test value set associated with the AT p ; wherein the step of obtaining GU p includes: S710, calculating the mean μ and the standard deviation σ of the set of valid test values after the r-1th filtering of noise, and obtaining the expected range (μ-hσ, μ+hσ) of the rth filtering according to μ and σ, wherein r is greater than 1, and h is a preset coefficient; S730, when all the test values in the effective test value set after the r-1th filtering of noise are within the expected range, obtaining a target test value set GU p ; otherwise, filtering out the test values outside the expected range to obtain an effective test value set after the rth filtering of noise; performing S710 and S730 again according to the effective test value set after the rth filtering of noise; S900, according to the GU to obtain P kind of correlation test item test value distribution index DIS = {DIS1, DIS2, …, DIS p ,…,DIS P} DIS p is the test value distribution index of AT p , DIS p includes the trend measure value ME p of GU p and the standard deviation of GU p ; according to AT and DIS, the mapping relationship table of the correlation test item and the test value distribution index is generated; The S300 comprises: S310, extracting the test object name in each test entry according to the first specified row, and extracting the test index in each test entry according to the second specified row, to obtain a candidate associated test item composed of the test object name and the test index corresponding to each test entry; S320, comparing whether there are same candidate associated test items in the E candidate associated test items, and eliminating the same candidate associated test items to obtain P associated test items AT.
2. The method of claim 1, wherein, S500 comprises: S510, obtain the minimum threshold of the test metric for each associated test item in AT {THN1,THN2,…,THN p ,…,THN P } and the maximum threshold {THX1,THX2,…,THX} p ,…,THX P }, where THN p For AT p China TR q Minimum threshold, THX q For AT p China TR q The maximum threshold; S530, AT p All test values associated with THN p and THX p are compared, and test values less than THN p and greater than THX p are removed, leaving a set of valid test values associated with AT p .
3. The method of claim 1, wherein, S500 comprises: S520, obtaining absolute threshold values {TH1, TH2, …, THN} of the test indexes in each association test item in the AT, wherein THN is the absolute threshold value of the TRN in the AT. p ,…,TH P} are the absolute threshold values of the TRs in the AT. p p q S540, AT p The absolute value of all test values associated with TH p are compared and test values greater than TH p are removed to obtain AT p The set of valid test values associated.
4. The method of claim 1, wherein, S700 further comprises: S810, calculating the proportion of the total number of test values in the P target test value set in the total number of all test values contained in the Data; S820, when the proportion is less than a preset proportion threshold, a preset prompt identifier is generated.
5. The method of claim 1, wherein, The ME p For GU p of the median.
6. The method of claim 1, wherein, The h is 3 or 5.
7. The method of claim 1, wherein, The DIS p Further comprises a noise term NOI p , the NOI p comprises a GU p number of filtered noise L p and noise deviation nod p .
8. The method of claim 7, wherein, The noise deviation nod p For L p Noise and GU p The average difference between the means, or nod p By calculating L p Noise and GU p The variance between the means is obtained.
9. A test data extraction system, the system comprising a processor and a non-transitory computer readable storage medium, the non-transitory computer readable storage medium having stored therein at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the method of any one of claims 1-8.
Citation Information
Patent Citations
Test method and device
CN110209577A
Software development kit SDK test method and related equipment
CN112433940A