An intelligent binning method and system for storage chip burn-in testing
By dynamically adjusting the batching strategy through real-time data acquisition and probability distribution models, the problem of low resource utilization in memory chip aging tests was solved, achieving efficient adaptive batching decisions and waste material processing, thereby improving equipment utilization and production line efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN HONGJINGWEI TECHNOLOGY CO LTD
- Filing Date
- 2026-02-08
- Publication Date
- 2026-06-12
AI Technical Summary
In existing memory chip aging tests, traditional batch closure methods rely on fixed timeout mechanisms, resulting in low resource utilization and an inability to adapt to individual differences between different chips, leading to low equipment utilization and reduced production line throughput.
By collecting aging status data in real time, fitting a probability distribution model, and dynamically adjusting the batch closure trigger threshold and timeout countdown, adaptive batch closure decision-making is achieved, and tailings are automatically identified and processed.
It improves the turnover efficiency of aging equipment, reduces resource idleness, ensures test integrity, is suitable for large-scale mass production scenarios, requires no manual intervention, and is low-cost and easy to integrate.
Smart Images
Figure CN122201401A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor reliability testing technology, specifically to an intelligent batching method and system for aging tests of memory chips. Background Technology
[0002] In the reliability verification process of semiconductor memory chips (such as eMMC and UFS), high-temperature aging testing (Burn-in Test) is one of the key steps. This test runs the device under test (DUT) for an extended period under accelerated stress conditions such as high temperature and high voltage to induce potential early failures, thereby screening out high-reliability products. Since aging tests typically last tens to hundreds of hours and require processing hundreds or thousands of chips simultaneously, the core challenge in improving aging test efficiency is how to efficiently organize test batches, promptly release resources that have completed testing, and avoid delaying the overall progress due to individual "slow" or "abnormal" devices.
[0003] Traditional aging test systems typically employ a fixed timeout mechanism for batch control: a uniform aging time threshold is preset, and once this time is reached, the current batch is forcibly terminated regardless of whether all DUTs have completed testing. However, in actual production, the aging completion time varies significantly among different chips—some devices may complete all tests well before the set time, while a few devices may fail to complete the process for an extended period due to process deviations, defects, or communication anomalies. In this situation, premature batch aging may result in some qualified products not being adequately aged; conversely, delayed batch aging forces a large number of tested chips to remain untested, consuming valuable aging board resources and reducing equipment utilization and production line throughput.
[0004] To alleviate these problems, some improvement solutions have introduced static grouping or manual intervention strategies, such as grouping tests according to estimated performance or having operators periodically determine whether a batch should be closed. However, these methods either rely on prior knowledge and lack adaptability, or increase labor costs, making it difficult to meet the needs of large-scale, highly automated modern packaging and testing production lines.
[0005] Therefore, there is an urgent need for a technical solution that can dynamically sense test progress based on real-time aging data, intelligently predict completion trends, and adaptively trigger batch closure decisions, so as to maximize the utilization efficiency of aging resources while ensuring the integrity of the test. Summary of the Invention
[0006] To address the shortcomings of existing technologies, the present invention aims to provide an intelligent batching method and system for memory chip aging tests, thereby solving the problems of existing intelligent batching methods for memory chip aging tests relying on manual intervention, prior knowledge, and lacking adaptive capabilities.
[0007] To solve the above-mentioned technical problems, the present invention is implemented through the following solution:
[0008] The present invention provides an intelligent batching method for aging tests of memory chips, comprising the following steps:
[0009] S100: Real-time acquisition of aging status data of the device under test (DUT), and recording the completion time of each DUT when it is determined that it has completed all preset aging test items;
[0010] S200. Obtain a probability distribution model of the completion time by fitting the completion time, and dynamically determine the batch termination trigger threshold of the current batch based on the probability distribution model.
[0011] S300. When the number of devices under test (DUTs) that have completed aging tests reaches the batching trigger threshold, start the adaptive timeout countdown.
[0012] S400. For each device under test (DUT) that has not completed aging tests, calculate the remaining aging time based on its running time and the completion progress of the aging test items, and determine the duration of the adaptive timeout countdown based on all the remaining aging times.
[0013] S500: If all devices under test (DUTs) have completed aging tests before the adaptive timeout countdown ends, the batch closure operation is executed immediately; if there are still devices under test (DUTs) that have not completed aging tests when the adaptive timeout countdown ends, the batch closure operation is forcibly executed, and the devices under test (DUTs) that have not completed aging tests are marked as tailings for post-processing.
[0014] Preferably, in step S200, the batch termination trigger threshold for the current batch is dynamically determined based on the probability distribution model, including:
[0015] Generate a corresponding cumulative distribution function based on the probability distribution model; obtain the current system time and the preset time buffer amount and substitute them into the cumulative distribution function to calculate the proportion of devices under test (DUTs) expected to complete aging tests within a short future window, which is defined as the expected completion proportion; multiply the expected completion proportion by the total number of DUTs in the current batch, and use the result as the batch termination trigger threshold.
[0016] Preferably, in step S400, the remaining aging time is estimated as follows: for each device under test (DUT) that has not completed aging tests, its running time is divided by the proportion of the number of aging test items currently completed to the total number of test items to obtain the estimated total aging time; the remaining aging time is the estimated total aging time minus the running time.
[0017] In step S400, determining the duration of the adaptive timeout countdown based on the remaining aging time of all unfinished DUTs includes: selecting the duration corresponding to a preset percentile from all the remaining aging times as the base timeout duration; comparing the base timeout duration with a preset safety lower limit time, and taking the larger of the two as the adaptive timeout countdown duration.
[0018] Preferably, the aging status data includes at least one of the following: pass or fail status of aging test items, read / write latency under a specific access mode, and ECC error correction count.
[0019] Preferably, the probability distribution model is fitted using a normal distribution or a Weiber distribution.
[0020] Preferably, the post-processing of the waste material includes automatically transferring the incomplete DUT to an extended aging analysis process or a failed product analysis process.
[0021] The present invention also provides an intelligent batching system for implementing the intelligent batching method as described above, comprising:
[0022] The real-time data acquisition module is used to acquire aging status data of each DUT and record the completion time when the DUT completes all preset aging test items;
[0023] The dynamic parameter calculation engine is used to fit a probability distribution model of completion time based on the completion time of completed DUTs, calculate the batch trigger threshold, and estimate the remaining aging time based on the running time and test progress of incomplete DUTs, thereby determining the adaptive timeout countdown duration.
[0024] The decision arbitrator is used to monitor the number of completed DUTs and the countdown status, and triggers the batch closing operation when the batch closing conditions are met.
[0025] The aging board controller interface is used to send batch closing instructions to the aging test hardware platform and to perform isolation or power-off control on DUTs marked as tailings.
[0026] Preferably, the intelligent batching system is deployed as an independent service and communicates with the test execution management system through an application programming interface.
[0027] Furthermore, the present invention also covers a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent batching method as described above.
[0028] In addition, the present invention also provides an aging test device, which integrates an intelligent batching system as described above, for performing aging tests on eMMC or UFS storage chips and implementing intelligent batching.
[0029] Compared with the prior art, the beneficial effects of the present invention are:
[0030] By collecting real-time data on the completion time of aging tests of devices under test (DUTs) and fitting a probability distribution model of the completion time, the system dynamically adjusts the batch closure trigger threshold and timeout countdown duration. This overcomes the resource idleness or insufficient testing problems caused by traditional fixed timeout mechanisms that fail to reflect actual testing progress. While ensuring that the vast majority of DUTs complete sufficient aging tests, the system promptly releases aging board channels and system resources occupied by tested devices, effectively shortening the average batch occupancy time and significantly improving the turnaround efficiency of aging equipment. The entire batch closure decision process requires no manual intervention; the system automatically predicts aging progress, determines the batch closure timing, and identifies and processes leftover materials, making it suitable for high-throughput, continuous, large-scale mass production scenarios. This solution can be directly integrated into existing aging test platforms, requiring only the addition of data analysis and intelligent decision-making modules at the software level. No structural modifications to the aging hardware or testing process are needed, resulting in low implementation costs and ease of promotion. For devices under test that fail to complete testing before the countdown ends, the system automatically marks them as leftovers and triggers a dedicated post-processing procedure to ensure that all abnormal or delayed devices are recorded, isolated, and included in subsequent analysis, thereby maintaining the quality traceability of the entire process. Attached Figure Description
[0031] Figure 1 This is a flowchart of the intelligent batch closing method in an embodiment of the present invention;
[0032] Figure 2 This is a structural block diagram of the intelligent batch processing system in an embodiment of the present invention. Detailed Implementation
[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the scope of protection of the present invention. Obviously, the embodiments described in this invention 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.
[0034] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0035] Example 1: The specific structure of the present invention is as follows:
[0036] like Figure 1 As shown, the present invention provides an intelligent batching method for aging tests of memory chips, comprising the following steps:
[0037] S100: Real-time acquisition of aging status data of the Device Under Test (DUT), and recording the completion time when each DUT completes all preset aging test items. It should be noted that during the aging test, the system continuously acquires the status information of each DUT through the monitoring circuit or test interface on the aging board, including the pass / fail status of each aging test item, read / write latency under specific access modes, ECC error correction count, etc. When a DUT successfully passes all preset aging test items consecutively without any unrecoverable errors, the system determines that it has completed the aging test and records its completion time. .
[0038] S200. A probability distribution model of the completion time is fitted based on the stated completion time, and the batch termination trigger threshold for the current batch is dynamically determined based on the probability distribution model. It should be noted that the system maintains a set of completion times for completed DUTs. When the number of samples in this set reaches a preset minimum threshold (e.g., 10), the probability distribution model of the completion time, such as a Weiber distribution or a normal distribution, is fitted using maximum likelihood estimation, and the corresponding cumulative distribution function (CDF) is generated. Subsequently, for every additional completed samples (e.g., every 5), the system refits the model to maintain the timeliness and accuracy of the prediction.
[0039] Based on the current system time and preset buffer time For example, over a 10-minute period, the cumulative distribution function (CDF) is used to calculate... That is, expected in the future The percentage of Device Under Test (DUTs) that completed aging tests within the specified time. Multiply this percentage by the total number of DUTs in the current batch. The result obtained is the dynamic batch trigger threshold. For example, if =0.95, then the threshold is set to 95% of the total quantity.
[0040] S300. When the number of devices under test (DUTs) that have completed aging tests reaches the batch termination trigger threshold, an adaptive timeout countdown is initiated. It should be noted that the batch termination trigger threshold is first reached when the number of completed DUTs reaches this threshold. When the test is completed, the system initiates an adaptive timeout countdown. This countdown is initiated only once and will not be reset even if more devices under test (DUTs) complete subsequently. The countdown duration is determined as follows: for each incomplete DUT, the remaining aging time is estimated based on its running time and the proportion of test item completion; the 90th percentile value of all remaining times (this percentile is configurable) is taken as the base timeout duration; the final countdown duration is the greater of this base value and a preset safety lower limit (e.g., 30 minutes).
[0041] S400. For each device under test (DUT) that has not completed aging tests, calculate the remaining aging time based on its running time and the completion progress of the aging test items, and determine the duration of the adaptive timeout countdown based on all the remaining aging times.
[0042] S500: If all devices under test (DUTs) have completed aging tests before the adaptive timeout countdown ends, the batch closure operation is executed immediately; if there are still devices under test (DUTs) that have not completed aging tests when the adaptive timeout countdown ends, the batch closure operation is forcibly executed, and the devices under test (DUTs) that have not completed aging tests are marked as tailings for post-processing.
[0043] If all devices under test (DUTs) complete the aging test during the countdown period, the system will immediately perform a batch closure operation and release the corresponding aging board resources. If there are still unfinished DUTs when the countdown ends, the batch will be forcibly closed and these DUTs will be marked as "leftovers". The Test Execution Management System (TEMS) will be notified via API to automatically transfer them to dedicated processing procedures such as extended aging, in-depth diagnostics, or scrap analysis.
[0044] In this embodiment, by real-time acquisition of the aging completion time of the devices under test (DUTs), and based on this, a probability distribution model of the completion time is fitted to dynamically adjust the batch closure trigger threshold and the timeout countdown duration. This overcomes the problem of resource idleness or insufficient testing caused by the traditional fixed timeout mechanism, which cannot reflect the actual testing progress. While ensuring that the vast majority of DUTs complete sufficient aging testing, the system can promptly release the aging board channels and system resources occupied by the tested devices, effectively shortening the average batch occupancy time and significantly improving the turnaround efficiency of the aging equipment. The entire batch closure decision process requires no manual intervention; the system can automatically predict aging progress, determine the batch closure timing, and identify and process leftover materials. It is suitable for high-throughput, continuous, large-scale mass production scenarios. This solution can be directly integrated into existing aging test platforms, requiring only the addition of data analysis and intelligent decision-making modules at the software level. No structural modifications to the aging hardware or testing process are needed, resulting in low implementation costs and easy promotion. For devices under test that fail to complete testing before the countdown ends, the system automatically marks them as leftovers and triggers a dedicated post-processing procedure to ensure that all abnormal or delayed devices are recorded, isolated, and included in subsequent analysis, thereby maintaining the quality traceability of the entire process.
[0045] Further, in step S200, the batch termination trigger threshold for the current batch is dynamically determined based on the probability distribution model, including:
[0046] Generate a corresponding cumulative distribution function based on the probability distribution model; obtain the current system time and the preset time buffer amount and substitute them into the cumulative distribution function to calculate the proportion of devices under test (DUTs) expected to complete aging tests within a short future window, which is defined as the expected completion proportion; multiply the expected completion proportion by the total number of DUTs in the current batch, and use the result as the batch termination trigger threshold.
[0047] Further, in step S400, the remaining aging time is estimated as follows: for each device under test (DUT) that has not completed the aging test, its running time is divided by the proportion of the number of aging test items that have been completed to the total number of test items to obtain the estimated total aging time; the remaining aging time is the estimated total aging time minus the running time.
[0048] In step S400, determining the duration of the adaptive timeout countdown based on the remaining aging time of all unfinished DUTs includes: selecting the duration corresponding to a preset percentile from all the remaining aging times as the base timeout duration; comparing the base timeout duration with a preset safety lower limit time, and taking the larger of the two as the adaptive timeout countdown duration.
[0049] Let the remaining aging time be Adaptive timeout countdown is The safe lower limit time is The formula for calculating the adaptive timeout countdown is: = That is, take the remaining aging time of the remaining device under test (DUT) predicted to be the first of the remaining aging time. percentiles (e.g.) =90), and not lower than a safety lower limit. .
[0050] Furthermore, the aging status data includes at least one of the following: the pass or failure status of aging test items, read / write latency under a specific access mode, and ECC error correction count.
[0051] Furthermore, the probability distribution model is fitted using a normal distribution or a Weiber distribution.
[0052] Furthermore, the post-processing of the waste material includes automatically transferring the incomplete DUT to an extended aging analysis process or a failed product analysis process.
[0053] like Figure 2 As shown, the present invention also provides an intelligent batching system for implementing the intelligent batching method described above, comprising:
[0054] The real-time data acquisition module is used to acquire aging status data of each DUT and record the completion time when the DUT completes all preset aging test items;
[0055] The dynamic parameter calculation engine is used to fit a probability distribution model of completion time based on the completion time of completed DUTs, calculate the batch trigger threshold, and estimate the remaining aging time based on the running time and test progress of incomplete DUTs, thereby determining the adaptive timeout countdown duration.
[0056] The decision arbitrator is used to monitor the number of completed DUTs and the countdown status, and triggers the batch closing operation when the batch closing conditions are met.
[0057] The aging board controller interface is used to send batch closing instructions to the aging test hardware platform and to perform isolation or power-off control on DUTs marked as tailings.
[0058] Furthermore, the intelligent batching system is deployed as an independent service and communicates with the test execution management system via an application programming interface.
[0059] Furthermore, the present invention also covers a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent batching method as described above.
[0060] In addition, the present invention also provides an aging test device, which integrates an intelligent batching system as described above, for performing aging tests on eMMC or UFS storage chips and implementing intelligent batching.
[0061] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A smart batching method for aging tests of memory chips, characterized in that, Includes the following steps: S100: Real-time acquisition of aging status data of the device under test (DUT), and recording the completion time of each DUT when it is determined that it has completed all preset aging test items; S200. Obtain a probability distribution model of the completion time by fitting the completion time, and dynamically determine the batch termination trigger threshold of the current batch based on the probability distribution model. S300. When the number of devices under test (DUTs) that have completed aging tests reaches the batching trigger threshold, start the adaptive timeout countdown. S400. For each device under test (DUT) that has not completed aging tests, calculate the remaining aging time based on its running time and the completion progress of the aging test items, and determine the duration of the adaptive timeout countdown based on all the remaining aging times. S500: If all devices under test (DUTs) have completed aging tests before the adaptive timeout countdown ends, the batch closure operation is executed immediately; if there are still devices under test (DUTs) that have not completed aging tests when the adaptive timeout countdown ends, the batch closure operation is forcibly executed, and the devices under test (DUTs) that have not completed aging tests are marked as tailings for post-processing.
2. The intelligent batch closing method as described in claim 1, characterized in that, In step S200, the batch closing trigger threshold for the current batch is dynamically determined based on the probability distribution model, including: Generate the corresponding cumulative distribution function based on the probability distribution model; The current system time and the preset time buffer are obtained and substituted into the cumulative distribution function to calculate the proportion of the device under test (DUT) expected to complete the aging test within a short future window, which is defined as the expected completion proportion. The expected completion rate is multiplied by the total number of devices under test (DUTs) in the current batch, and the result is used as the batch closure trigger threshold.
3. The intelligent batch closing method as described in claim 1 or 2, characterized in that, In step S400, the remaining aging time is estimated as follows: For each device under test (DUT) that has not completed aging tests, the estimated total aging time is obtained by dividing its running time by the ratio of the number of aging test items currently completed to the total number of test items. The remaining aging time is the estimated total aging time minus the already run time; In step S400, determining the duration of the adaptive timeout countdown based on the remaining aging time of all incomplete DUTs includes: Select the duration corresponding to a preset percentile from all remaining aging times as the base timeout duration; The basic timeout duration is compared with the preset safety lower limit time, and the larger of the two values is taken as the adaptive timeout countdown duration.
4. The intelligent batch closing method as described in claim 1, characterized in that, The aging status data includes at least one of the following: pass or fail status of aging test items, read / write latency under a specific access mode, and ECC error correction count.
5. The intelligent batch closing method as described in claim 1, characterized in that, The probability distribution model is fitted using a normal distribution or a Weiber distribution.
6. The intelligent batch closing method as described in claim 1, characterized in that, The post-processing of waste materials includes automatically transferring incomplete DUTs to extended aging analysis processes or failed product analysis processes.
7. An intelligent batching system implementing the intelligent batching method as described in any one of claims 1 to 6, characterized in that, include: The real-time data acquisition module is used to acquire aging status data of each DUT and record the completion time when the DUT completes all preset aging test items; The dynamic parameter calculation engine is used to fit a probability distribution model of completion time based on the completion time of completed DUTs, calculate the batch trigger threshold, and estimate the remaining aging time based on the running time and test progress of incomplete DUTs, thereby determining the adaptive timeout countdown duration. The decision arbitrator is used to monitor the number of completed DUTs and the countdown status, and triggers the batch closing operation when the batch closing conditions are met. The aging board controller interface is used to send batch closing instructions to the aging test hardware platform and to perform isolation or power-off control on DUTs marked as tailings.
8. The intelligent batch processing system as described in claim 7, characterized in that, The intelligent batch processing system is deployed as an independent service and communicates with the test execution management system through an application programming interface.
9. A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the intelligent batching method as described in any one of claims 1 to 7.
10. An aging test apparatus, integrating an intelligent batching system as described in claim 8 or 9, for performing aging tests on eMMC or UFS storage chips and implementing intelligent batching.