A Fast Aging Test Method and Device for a Memory Array

Abstract

The present invention discloses a rapid aging test method and device for a memory array, relating to the technical field of memory array testing. It includes capturing the instantaneous current waveforms of each storage unit in the memory array during read and write operations, extracting current features therefrom, constructing a difference matrix using the extracted current features, and then dynamically marking the storage units that exceed the threshold. By analyzing the difference in the amount of stored data before and after capturing the instantaneous current waveforms of the storage units, the present invention evaluates the decline in the capacity of the storage units, and combines the capacity decline data as an auxiliary evaluation index with the aging rate of the storage units to achieve a comprehensive evaluation of the aging degree of the storage units without applying external influence to the memory array. This not only improves the accuracy of the aging test, but also can effectively identify the aging differences between storage units, ensuring that critical storage units are promptly processed and maintained.

Description

Technical Field

[0001] The present invention relates to the technical field of memory array testing, and specifically to a fast aging test method and device for a memory array. Background Art

[0002] A memory array is a storage structure composed of multiple storage units arranged in rows and columns. In a memory array, a large number of storage units are organized together to form a dense storage network, making the storage and reading of data efficient and fast. Since the performance and reliability of the memory array are crucial for the operation of the entire electronic device, a fast aging test method and device for the memory array need to be designed.

[0003] After retrieval, the Chinese invention patent with the publication number "CN113380312A" discloses a memory array test method and system. This application heats the NVM array to a target temperature, and while heating the NVM array to the target temperature, obtains a current distribution by measuring multiple currents of a subset of NVM cells in the NVM array, programs each NVM cell in the NVM array to one of a logic high state or a logic low state, and performs a first pass / fail (P / F) test and a second pass / fail (P / F) test on each NVM cell in the NVM array, and calculates the bit error rate based on the current distribution and the first P / F test and the second P / F test.

[0004] In addition, the Chinese invention patent with the publication number "CN114187955A" discloses a test method, device, equipment and storage medium for a memory array. This application, after lowering the capacitor plate voltage, performs a preset read / write operation on the memory array after storing charges greater than the pre-charge amount, so that the capacitor voltages of the storage units in the memory array change continuously, increasing the exposure probability of capacitor plate leakage failure conditions, and easily detecting potential capacitor plate leakage failure conditions in the memory array.

[0005] In order to ensure the normal operation of an electronic device, there is an urgent need for a fast aging test method and device that can be carried out without affecting the normal operation of the electronic device. However, according to some existing test means, for example, the methods shown in the two aforementioned related invention patents and similar related patents, although they can detect certain performance changes of the memory array to a certain extent, these methods require applying external conditions such as temperature and voltage that may damage the memory array. Therefore, this study proposes an innovative fast aging test method and device, aiming to complete the fast aging test of the memory array without causing additional damage to the memory array. Summary of the Invention

[0006] The object of the present invention is to provide a fast aging test method and device for a memory array to solve the problems raised in the above-mentioned background art.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] In a first aspect, a fast aging test method for a memory array is proposed, including:

[0009] During the read and write operations, capture the instantaneous current waveforms of each storage cell in the memory array and extract the current characteristics therefrom;

[0010] Construct a difference matrix using the extracted current characteristics, and then dynamically mark the storage cells exceeding the threshold;

[0011] Construct an aging trajectory model, with the instantaneous current waveforms of the storage cells with different marks in the time series as the input and the current characteristics as the output, so as to obtain the aging rate. The aging trajectory model is used to judge the aging degree of the storage cells through the waveforms of the instantaneous current, so as to predict in advance the subsequent fault points of the memory array;

[0012] Allocate priority weights to the marked storage cells according to the aging rate;

[0013] Obtain the storage capacities of the storage cells under different priority weights;

[0014] Apply a pressure gradient proportional to the storage capacity according to the current characteristic ratio;

[0015] Evaluate the capacity attenuation of the storage cells according to the change degree of the pressure gradient within a single storage cell in the time series;

[0016] Associate the evaluation results with the priority weights to obtain the aging replacement requirement degree of the storage cells in the memory.

[0017] As a further preference of this technical solution, the current characteristics include: the absolute deviation value of the current peak and the integral charge amount;

[0018] The extraction method of the current characteristics includes:

[0019] Obtain the arrangement characteristics of the storage cells in the memory array;

[0020] Based on the arrangement characteristics, connect a high-frequency sampling circuit to each storage cell;

[0021] According to the read and write operation frequency of the storage array, serially add an adjustable resistor in the high-frequency sampling circuit. The adjustable resistor is used to adjust the sampling frequency to keep synchronous with the read and write operation frequency;

[0022] The high-frequency sampling circuit obtains the absolute deviation value of the current peak value, the integrated charge quantity, the rising edge slope, and the time-frequency domain energy distribution characteristics of the integrated memory cell according to the extraction algorithm.

[0023] As a further preference of this technical solution, the number of the extraction algorithms matches the number of current feature types, and includes an absolute deviation value extraction algorithm and an integrated charge quantity extraction algorithm;

[0024] The absolute deviation value extraction algorithm is: , where is the current peak value. By using the high-frequency sampling circuit to sample the current waveform within the time period (t1, t2), a series of discrete points are obtained, where k ranges from 1 to M, and M is the number of sampling points. is the absolute deviation value of the current peak value, and the current peak value is the maximum value within the th sampling period, that is , is the average value of the current peak value, which is obtained by averaging the current peak values within multiple sampling periods. n is the maximum value collected, that is ;

[0025] The integrated charge quantity extraction algorithm is: , where and are the start time and end time of the read-write operation respectively. is the sampling interval. is the integrated charge quantity at the th time. represents the current value at the time point . K represents the number of sampling points, that is, how many times the current value is measured.

[0026] As a further preference of this technical solution, the threshold set based on the absolute deviation value of the current peak value is 1.5 to 2 times the average value of the absolute deviation value of the current peak value, and the threshold set based on the integrated charge quantity is 1.2 to 1.8 times the average value of the integrated charge quantity. When the current feature of the memory cell exceeds any of the above thresholds, it is marked as an abnormal memory cell;

[0027] The method for constructing the difference matrix is: taking each memory cell as a row of the matrix and the current feature as a column of the matrix, and filling the current feature values of each memory cell into the corresponding row and column positions to form a difference matrix.

[0028] As a further preference of this technical solution, the method for constructing the aging trajectory model includes:

[0029] Segmented sampling is performed on the historical current waveform in the marked storage unit;

[0030] Based on the segmented sampling results, a time series data set is extracted;

[0031] According to the time series data set, instantaneous current eigenvalue is extracted;

[0032] According to the material properties of the storage array, matching material degradation factor values and aging rate coefficients are retrieved from the Internet;

[0033] A function correlation formula is established between the instantaneous current eigenvalue, the material degradation factor value and the aging rate coefficient under the time series.

[0034] As a further preference of this technical solution, the function correlation formula is ;

[0035] where is the cumulative degradation amount within time t, indicating the degree of performance attenuation of the storage unit, is the material degradation factor value, and are the aging rate coefficients respectively, is the peak current in the i-th time period, used to represent the instantaneous current eigenvalue, is the current change rate in the i-th time period, is used to represent the time segmentation interval.

[0036] As a further preference of this technical solution, the pressure gradient acting on the storage capacity is used to represent the total gradient of data information that the storage unit can store;

[0037] The evaluation method for the capacity attenuation of the storage unit includes:

[0038] Obtain the difference in the total amount of data information in the time periods before and after the instantaneous current waveform;

[0039] Based on the difference in the total amount of data information, the storage capacity and the current characteristics of the instantaneous current waveform, a capacity attenuation evaluation formula is constructed;

[0040] According to the capacity attenuation evaluation formula, the capacity attenuation rate of the storage unit is obtained. The capacity attenuation rate is used to quantify the performance degradation of the storage unit within a unit time period. The higher the capacity attenuation rate, the more serious the aging degree of the storage unit.

[0041] As a further preference of this technical solution, the capacity attenuation evaluation formula is ;

[0042] where is used to represent the relative data loss degree of the storage unit in the aging test, Used to represent the contribution of the quantified current to aging. The greater the current or the longer the action time, the more significant the accelerated aging effect. Used to represent the difference in the total amount of data information before and after the instantaneous current waveform. The charge integration of the instantaneous current waveform within the test time period. Is the maximum storage capacity of the storage unit. Used to represent the capacity attenuation rate.

[0043] In a second aspect, to improve the above-disclosed fast aging test method for a memory array, a fast aging test device for a memory array is also proposed. It should be noted that a fast aging test device for a memory array uses the above-disclosed fast aging test method for a memory array and includes:

[0044] A current capture module, used to capture the instantaneous current waveform of each storage unit in the memory array during the read and write operations;

[0045] A feature extraction module, connected to the current capture module, used to extract current features from the captured instantaneous current waveform;

[0046] A difference matrix construction module, connected to the feature extraction module, used to construct a difference matrix based on the extracted current features and dynamically mark the storage units that exceed the threshold;

[0047] An aging trajectory model construction module, connected to the difference matrix construction module, used to take the instantaneous current waveforms of the storage units with different marks in the time series as inputs and the current features as outputs to construct an aging trajectory model, thereby obtaining the aging rate;

[0048] A priority weight assignment module, connected to the aging trajectory model construction module, used to assign priority weights to the marked storage units according to the aging rate;

[0049] A storage capacity acquisition module, connected to the priority weight assignment module, used to obtain the storage capacity of the storage units under different priority weights;

[0050] A pressure gradient application module, connected to the storage capacity acquisition module, used to apply a pressure gradient proportional to the current feature ratio to the storage capacity;

[0051] A capacity attenuation evaluation module, connected to the pressure gradient application module, used to evaluate the capacity attenuation of the storage unit according to the change degree of the pressure gradient within a single storage unit in the time series, and associate the evaluation result with the priority weight to obtain the aging replacement requirement degree of the storage units in the memory.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] The fast aging test method and device for a memory array monitor the current characteristics of the memory array during read and write operations to identify the aging rate of memory cells, and based on the quantified data of the aging rate, divide the memory cells into priority weights to ensure targeted attention and processing of the memory cells according to the priority;

[0054] Furthermore, by analyzing the difference in the amount of stored data before and after instantaneous current waveform capture of the memory cells, the decline of the memory cell capacity is evaluated, and combined with the capacity decline data as an auxiliary evaluation index, and combined with the aging rate of the memory cells, a comprehensive evaluation of the aging degree of the memory cells is achieved without applying external influences to the memory array, which not only improves the accuracy of the aging test, but also can effectively identify the aging differences between memory cells to ensure that critical memory cells are processed and maintained in a timely manner. Brief Description of the Drawings

[0055] Figure 1 is a flowchart of the steps of the method disclosed in the present invention;

[0056] Figure 2 is a schematic diagram of the high-frequency sampling circuit connected in series in the memory array of the present invention;

[0057] Figure 3 is a schematic diagram of the high-frequency sampling circuit connected in parallel in the memory array of the present invention;

[0058] Figure 4 is a block diagram of the module composition of the device disclosed in the present invention. Detailed Embodiment

[0059] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] Before understanding the technical solutions proposed by the present invention, it should be clear that the memory array, as the core component for data storage, its performance stability and data reliability are crucial. However, with the increase in usage time, the memory cells will gradually age due to various factors, resulting in a decline in storage performance and even data loss. In order to effectively monitor and manage the aging problem of memory cells, the present invention proposes an innovative fast aging test method, aiming to evaluate the aging degree of memory cells without applying external conditions that may damage the memory array to the outside of the memory array, so as to provide a scientific basis for subsequent maintenance and processing.

[0061] Specifically, the present invention proposes a fast aging test method for a memory array. Referring to Figure 1 it can be seen that it includes: step S100 - step S800.

[0062] Step S100: During the read and write operations, capture the instantaneous current waveform of each storage cell in the memory array, and extract the current characteristics therefrom.

[0063] It should be clear that step S100 in the present invention is used to collect the current characteristics of the storage cells during normal operation. The current characteristics can help accurately judge whether the storage cells show signs of aging and the degree of aging, and can be used to accurately evaluate the aging degree of the storage cells based on these basic data in the subsequent work.

[0064] It should be added that the current characteristics in step S100 include: the absolute deviation value of the current peak and the integral charge amount. The absolute deviation value of the current peak is used to represent the fluctuation degree of the current waveform at the peak, reflecting the stability of the current during the read and write operations of the storage cell, and the integral charge amount is used to represent the charge accumulation amount during the read and write operations of the storage cell. The size of the charge accumulation amount is directly related to the performance and lifespan of the storage cell.

[0065] Specifically, the method for capturing the instantaneous current waveform in step S100 is realized by connecting a high-frequency sampling circuit to each storage cell in the memory array. It should be added that the high-frequency sampling circuit captures the instantaneous current change of the storage cell during read and write because the high-frequency sampling circuit quickly captures and converts the instantaneous current, so that the continuous current change is recorded as discrete digital signals, and finally these digital signals are stored and recorded by the storage cell.

[0066] Specifically, referring to Figure 2 and Figure 3 it can be seen that in the present invention, the high-frequency sampling circuit has two specific forms. One form is applicable to the memory array formed by series-connected storage cells, and the other form is applicable to the memory array formed by parallel-connected storage cells. In the series form, the high-frequency sampling circuit is designed to be connected in series with each storage cell, so as to directly measure the instantaneous current of each storage cell. In the parallel form, the high-frequency sampling circuit is connected to each storage cell through a multiplexer to realize the time-sharing measurement of the instantaneous current of different storage cells. Both forms of the high-frequency sampling circuit can capture the instantaneous current waveform of the storage cell during the read and write operations.

[0067] In addition, it should be added that the method for extracting the current characteristics in step S100 includes: step S101 - step S104.

[0068] Step S101: Obtain the arrangement characteristics of the storage cells in the memory array.

[0069] Step S102: Based on the arrangement characteristics, connect a high-frequency sampling circuit to each storage cell.

[0070] Step S103: According to the read / write operation frequency of the storage array, serially add an adjustable resistor in the high-frequency sampling circuit.

[0071] It should be clear that in Step S103, the adjustable resistor is used to adjust the sampling frequency to keep in sync with the read / write operation frequency. In addition, it should be noted that a high-frequency current waveform sampler for obtaining the instantaneous current waveform is provided in the high-frequency sampling circuit. Since the high-frequency current waveform sampler is a mature technical product in the current electrical acquisition field, the specific structure and working principle of the high-frequency current waveform sampler will not be elaborated in this invention.

[0072] Step S104: The high-frequency sampling circuit obtains the absolute deviation value of the current peak and the integrated charge quantity of the integrated storage cell according to the extraction algorithm.

[0073] It should be clear that during the actual operation of Steps S101 - S104, the instantaneous current waveform data captured by the high-frequency sampling circuit is transmitted to the feature extraction module through the data line. The feature extraction module analyzes the received instantaneous current waveform data and calculates the absolute deviation value of the current peak and the integrated charge quantity respectively according to the preset extraction algorithm. These feature values are recorded and stored in the feature database in real time for subsequent steps to call.

[0074] As a preferred implementation manner, the number of extraction algorithms in Step S100 matches the number of current feature types, and includes an absolute deviation value extraction algorithm and an integrated charge quantity extraction algorithm.

[0075] Specifically, the absolute deviation value extraction algorithm is: , where is the current peak. By using the high-frequency sampling circuit to sample the current waveform within the time period (t1, t2), a series of discrete points are obtained, where k ranges from 1 to M, and M is the number of sampling points, is the absolute deviation value of the current peak, and the current peak is the maximum value within the th sampling period, that is , is the average value of the current peak, obtained by averaging the current peaks within multiple sampling periods. n is the maximum value collected, that is ;

[0076] In addition, the integrated charge quantity extraction algorithm is: , where and are the start time and end time of the read / write operation respectively, is the sampling interval, is the integrated charge quantity, represents the current value at the time point , K represents the number of sampling points, that is, how many times the current value is measured.

[0077] It should be added that when the absolute deviation value extraction algorithm and the integrated charge quantity extraction algorithm are put into actual use, assuming that the start time of sampling is 0s, the end time is 1s, the sampling interval is 0.01s, and the number of sampling points N is 100, at this time the current sampling value is , is 0.3A.

[0078] At this time, M = 10 sampling periods are carried out, and the number of current peaks obtained is , so , the absolute deviation value is .

[0079] In addition, assuming that within a sampling period, the current sampling value is , so the integrated charge quantity is .

[0080] Step S200: Construct a difference matrix using the extracted current features, and then dynamically mark the storage units exceeding the threshold.

[0081] It should be clear that step S200 in this application is used to perform state monitoring and preliminary determination of the aging degree of the storage unit based on the current feature data. Specifically, the construction of the difference matrix uses each storage unit as the row identifier and the current features (such as the absolute deviation value of the current peak and the integrated charge quantity) as the column identifier, and fills the extracted current feature values into the corresponding row and column positions one by one to form a complete difference matrix. This matrix intuitively shows the performance differences of each storage unit in different current features, which is convenient for subsequent analysis.

[0082] Furthermore, in order to accurately identify the aging storage units, a reasonable threshold needs to be set. Based on a large amount of experimental data and statistical analysis, the present invention preferably uses 1.5 to 2 times the average value of the absolute deviation value of the current peak as the threshold of the absolute deviation value, and 1.2 to 1.8 times the average value of the integrated charge quantity as the threshold of the integrated charge quantity. When any current feature value of the storage unit exceeds the above-set threshold, it is automatically marked as an abnormal storage unit.

[0083] In addition, it should be supplemented that the implementation of step S200 not only realizes the preliminary screening of the aging state of the storage unit, but also provides an important data basis for the construction of the subsequent aging trajectory model, ensuring the efficiency and accuracy of the aging test.

[0084] It also needs to be further supplemented that the construction method of the difference matrix in step S200 is as follows: taking each storage unit as a row of the matrix and the current characteristics as columns of the matrix, and filling the current characteristic values of each storage unit into the corresponding row and column positions to form a difference matrix.

[0085] Step S300: Construct an aging trajectory model, using the instantaneous current waveforms of storage units with different tags in the time series as inputs and the current characteristics as outputs, so as to obtain the aging rate.

[0086] It should be clear that the aging trajectory model is used to judge the aging degree of the storage unit through the waveform of the instantaneous current, so as to predict in advance the subsequent fault points in the memory array.

[0087] In addition, what needs to be supplemented for step S300 is that the construction method of the aging trajectory model includes: step S301 - step S305.

[0088] Step S301: Perform segmented sampling on the historical current waveforms in the tagged storage units.

[0089] It should be clear that segmented sampling means dividing the entire historical current waveform of the storage unit into multiple time periods, and the current waveforms within each time period are independently sampled to analyze in detail the changes in the current characteristics within each time period.

[0090] Step S302: Extract a time series data set based on the results of the segmented sampling.

[0091] Step S303: Extract the instantaneous current characteristic values according to the time series data set.

[0092] Step S304: According to the material properties of the storage array, retrieve the matching material degradation factor values and aging rate coefficients on the Internet.

[0093] Step S305: Establish a function association formula between the instantaneous current characteristic values in the time series and the material degradation factor values and aging rate coefficients.

[0094] It should be noted that the function association formula is ;

[0095] Where is the cumulative degradation amount within time t, indicating the degree of performance attenuation of the storage unit, is the material degradation factor value, and are the aging rate coefficients respectively, is the peak current in the i-th time period, used to represent the instantaneous current eigenvalue, is the current change rate in the i-th time period, used to represent the time segmentation interval.

[0096] It should be added that the function correlation formula comprehensively considers multiple factors to accurately reflect the aging of the storage unit. Among them, the cumulative degradation amount as the core index, intuitively reflects the degree of performance degradation of the storage unit over time. The value M of the material degradation factor, as an internal factor affecting aging, is directly related to the durability and stability of the storage unit material, while the aging rate coefficient and reflect the aging speed of the storage unit under different conditions and are key parameters for evaluating the lifespan of the storage unit. In addition, in terms of the instantaneous current eigenvalue, the peak current and the current change rate in the i-th time period are incorporated into the function correlation formula, representing the current intensity and current fluctuation of the storage unit within a unit time period respectively, and can sensitively capture the subtle characteristics of the performance change of the storage unit. In addition, the introduction of the time segmentation interval enables the function correlation formula to analyze the aging of the storage unit in different time periods more carefully, improving the accuracy and reliability of the test. It should be noted that when this application is actually used, the unit time can be limited to 1 minute.

[0097] Step S400: Assign priority weights to the marked storage units according to the aging rate.

[0098] It should be noted that the priority weights in step S400 are assigned from high to low according to the magnitude of the aging rate.

[0099] Step S500: Obtain the storage capacities of the storage units under different priority weights.

[0100] Step S600: Apply a pressure gradient proportional to the current characteristic ratio to the storage capacity.

[0101] It should be clear that the pressure gradient acting on the storage capacity is used to represent the total gradient of the data information that the storage unit can store.

[0102] Step S700: Evaluate the capacity attenuation of the storage unit according to the change degree of the pressure gradient within a single storage unit in the time series.

[0103] It should be clear that the evaluation method for the capacity attenuation of the storage unit in step S700 includes: step S701 - step S703.

[0104] Step S701: Obtain the difference in the total amount of data information in the time periods before and after the instantaneous current waveform.

[0105] Step S702: Based on the difference in the total amount of data information, the storage capacity, and the current characteristics of the instantaneous current waveform, construct a capacity attenuation evaluation formula.

[0106] Step S703: Obtain the capacity attenuation rate of the storage unit according to the capacity attenuation evaluation formula.

[0107] It should be clear that the capacity attenuation rate is used to quantify the performance degradation of the storage unit within a unit time period. The higher the capacity attenuation rate, the more serious the aging degree of the storage unit.

[0108] As a supplement to Step S702, the capacity attenuation evaluation formula is ;

[0109] Where is used to represent the relative data loss degree of the storage unit in the aging test, is used to represent the contribution of the quantization current to aging. The larger the current or the longer the action time, the more significant the accelerated aging effect, is used to represent the difference in the total amount of data information before and after the instantaneous current waveform, the charge integration of the instantaneous current waveform within the test time period, is the maximum storage capacity of the storage unit, is used to represent the capacity attenuation rate.

[0110] As a preferred implementation strategy, the following preparatory work must be completed before applying the capacity attenuation evaluation formula to actual operations.

[0111] First, conduct test preparations

[0112] Select a memory array: Assume there is a memory array with a capacity of 100 GB (i.e., = 100 GB).

[0113] Build a test environment: Set up a test platform to ensure that the capacity and current of the memory can be accurately measured.

[0114] Set test parameters: Apply a voltage of 3.3 V, and the duration of the aging test is 1000 hours.

[0115] Second, execute the aging test process

[0116] Initial capacity determination: Before the start of the aging test, determine the initial capacity of the memory array to be 100 GB.

[0117] Third, conduct current monitoring and integral calculation:

[0118] Assume that during the aging test, the current changes with time and can be monitored and recorded in real time by a data recording device.

[0119] Fourth, use the numerical integration method to integrate the current to obtain the total charge within 1000 hours =500 Ah.

[0120] Fifth, capacity measurement after aging: After the aging test is completed, measure the current capacity of the memory array to be 90 GB.

[0121] Sixth, calculate the capacity decay rate

[0122] Calculate the change in capacity: =100 GB - 90 GB = 10 GB.

[0123] Seventh, substitute the data into the formula to calculate the capacity decay rate:

[0124] ;

[0125] Finally, conduct result analysis and evaluation

[0126] Capacity decay rate: The calculated capacity decay rate is 50%, indicating that after 1000 hours of aging test, the capacity of the memory array has decreased by half of the initial capacity.

[0127] Step S800: Associate the evaluation result with the priority weight to obtain the degree of aging replacement requirement of the storage units in the memory.

[0128] It should be noted that when associating the evaluation result with the priority weight, the priority weight is taken as the core consideration factor and the evaluation result is used as a reference auxiliary to achieve this. In this way, it can be accurately determined which storage units in the memory are more urgently in need of replacement due to aging problems, thereby providing a favorable basis for subsequent maintenance and update operations of the memory, ensuring the stability of the overall performance of the memory and the security and reliability of data storage.

[0129] As a preferred implementation manner, referring to Figure 4 it can be known that the present invention also proposes a rapid aging test device for a memory array. It should be noted that a rapid aging test device for a memory array uses a rapid aging test method for a memory array and includes:

[0130] A current capture module for capturing the instantaneous current waveform of each storage unit in the memory array during read and write operations;

[0131] A feature extraction module, connected to the current capture module, for extracting current features from the captured instantaneous current waveform;

[0132] A difference matrix construction module, connected to the feature extraction module, for constructing a difference matrix based on the extracted current features and dynamically marking the storage units exceeding the threshold;

[0133] An aging trajectory model construction module, connected to the difference matrix construction module, for taking the instantaneous current waveforms of the storage units with different marks in time series as inputs and the current features as outputs to construct an aging trajectory model, so as to obtain the aging rate;

[0134] A priority weight assignment module, connected to the aging trajectory model construction module, for assigning priority weights to the marked storage units according to the aging rate;

[0135] A storage capacity acquisition module, connected to the priority weight assignment module, for acquiring the storage capacities of the storage units under different priority weights;

[0136] A pressure gradient application module, connected to the storage capacity acquisition module, for applying a pressure gradient proportional to the current feature ratio to the storage capacity;

[0137] A capacity attenuation evaluation module, connected to the pressure gradient application module, for evaluating the capacity attenuation of the storage unit according to the change degree of the pressure gradient within a single storage unit in the time series, and associating the evaluation result with the priority weight to obtain the aging replacement requirement degree of the storage units in the memory.

[0138] It should be noted that when a rapid aging test device for a memory array is actually used, first, the current capture module captures the instantaneous current waveforms of each storage cell in the memory array. The feature extraction module then extracts key current features from these waveforms. Next, the difference matrix construction module uses these features to construct a difference matrix and marks the storage cells whose current features exceed the preset threshold. These cells may be in a state of accelerated aging. Subsequently, the aging trajectory model construction module is connected and uses the instantaneous current waveforms under the time series as input and the current features as output to construct an aging trajectory model, thereby calculating the aging rate of the storage cells. With the information on the aging rate, the priority weight assignment module can assign priority weights to the marked storage cells according to this information. Subsequently, the storage capacity acquisition module measures the storage capacity of the storage cells under different priority weights to provide basic data for further testing. In addition, the pressure gradient application module applies a pressure gradient proportional to the current feature ratio to the storage capacity according to the current feature ratio to simulate the pressure environment of the storage cells in actual use. Finally, the capacity decay evaluation module evaluates the capacity decay of the storage cells according to the degree of change of the pressure gradient within a single storage cell in the time series and associates this evaluation result with the previous priority weights, thereby obtaining the degree of aging replacement requirements of the storage cells in the memory.

[0139] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended embodiments and their equivalents.

Claims

1. A fast aging test method for a memory array, characterized in that, Including: During the read and write operations, capture the instantaneous current waveforms of each storage cell in the memory array and extract current characteristics therefrom; Construct a difference matrix using the extracted current characteristics, and then dynamically mark the storage cells exceeding the threshold; Construct an aging trajectory model, taking the instantaneous current waveforms of storage cells with different marks in a time series as input and the current characteristics as output, so as to obtain the aging rate. The aging trajectory model is used to judge the aging degree of storage cells through the waveforms of instantaneous current, so as to predict in advance the subsequent fault points in the memory array; Allocate priority weights to the marked storage cells according to the aging rate; Obtain the storage capacities of storage cells under different priority weights; Apply a pressure gradient proportional to the current characteristic ratio to the storage capacity; Evaluate the capacity attenuation of the storage cell according to the change degree of the pressure gradient within a single storage cell in the time series; Associate the evaluation result with the priority weight to obtain the aging replacement demand degree of the storage cells in the memory; 2. A fast aging test method for a memory array according to claim 1, characterized in that: The current characteristics include: the absolute deviation value of the current peak and the integrated charge quantity; The extraction method of the current characteristics includes: Obtain the arrangement characteristics of the storage cells in the memory array; Based on the arrangement characteristics, connect a high-frequency sampling circuit to each storage cell; According to the read and write operation frequency of the storage array, serially add an adjustable resistor in the high-frequency sampling circuit. The adjustable resistor is used to adjust the sampling frequency to keep synchronous with the read and write operation frequency; The high-frequency sampling circuit obtains the absolute deviation value of the current peak and the integrated charge quantity of the integrated storage cell according to the extraction algorithm; 3. A rapid aging test method for a memory array according to claim 2, characterized in that: The number of the extraction algorithms matches the number of types of current characteristics, and includes an absolute deviation value extraction algorithm and an integrated charge quantity extraction algorithm; The absolute deviation value extraction algorithm is as follows: , where is the current peak value. By using a high-frequency sampling circuit to sample the current waveform within the time period (t1, t2), a series of discrete points are obtained, where k ranges from 1 to M, and M is the number of sampling points. is the absolute deviation value of the current peak value, and the current peak value is the maximum value within the th sampling period, that is . is the average value of the current peak value, which is obtained by averaging the current peak values within multiple sampling periods. n is the maximum value collected, that is ; The integral charge extraction algorithm is as follows: , where and are the start time and end time of the read / write operation respectively, is the sampling interval, is the integral charge within the th sampling period, represents the current value at the time point , and K represents the number of sampling points, i.e., the number of current value measurements.

4. A rapid aging test method for a memory array according to claim 2, characterized in that: The threshold set based on the absolute deviation value of the current peak is 1.5 to 2 times the average value of the absolute deviation value of the current peak, and the threshold set based on the integrated charge quantity is 1.2 to 1.8 times the average value of the integrated charge quantity. When the current characteristics of the storage cell exceed any of the above thresholds, it is marked as an abnormal storage cell; The construction method of the difference matrix is: taking each storage cell as a row of the matrix and the current characteristics as columns of the matrix, and filling the current characteristic values of each storage cell into the corresponding row and column positions to form a difference matrix; 5. A rapid aging test method for a memory array according to claim 1, characterized in that: The construction method of the aging trajectory model includes: Perform segmented sampling on the historical current waveforms in the marked storage cells; Extract the time series data set based on the segmented sampling results; Extract the instantaneous current characteristic values according to the time series data set; According to the material properties of the storage array, retrieve the matching material degradation factor values and aging rate coefficients on the Internet; Establish a function association formula between the instantaneous current characteristic values and the material degradation factor values and aging rate coefficients in the time series; 6. The rapid aging test method for a memory array according to claim 5, wherein: The function association formula is ; Among them is the cumulative degradation amount within time t, indicating the degree of performance attenuation of the storage unit, is the numerical value of the material degradation factor, and are the aging rate coefficients respectively, is the peak current in the i-th time period, used to represent the instantaneous current eigenvalue, is the current change rate in the i-th time period, is used to represent the time segmentation interval.

7. A fast aging test method for a memory array according to claim 1, characterized in that: The pressure gradient acting on the storage capacity is used to represent the total gradient of data information that the storage cell can store; The evaluation method of the capacity attenuation of the storage cell includes: Obtain the difference in the total amount of data information in the time periods before and after the instantaneous current waveform; Construct a capacity attenuation evaluation formula based on the difference in the total amount of data information, the storage capacity, and the current characteristics of the instantaneous current waveform; Obtain the capacity attenuation rate of the storage unit according to the capacity attenuation evaluation formula. The capacity attenuation rate is used to quantify the performance degradation of the storage unit within a unit time period. The higher the capacity attenuation rate, the more serious the aging degree of the storage unit.

8. A fast aging test method for a memory array according to claim 7, characterized in that: The capacity attenuation evaluation formula is ; wherein is used to represent the relative data loss degree of the storage unit during the aging test, is used to represent the contribution of the quantization current to aging. The larger the current or the longer the action time, the more significant the acceleration of the aging effect, is used to represent the difference in the total amount of data information before and after the instantaneous current waveform, is the charge integration of the instantaneous current waveform within the test time period, is the maximum storage capacity of the storage unit, is used to represent the capacity attenuation rate.

9. A rapid aging test device for a memory array, which uses the rapid aging test method for a memory array according to any one of claims 1-8, characterized in that, Including: A current capture module, which is used to capture the instantaneous current waveform of each storage unit in the memory array during the read and write operations; A feature extraction module, connected to the current capture module, which is used to extract current features from the captured instantaneous current waveform; A difference matrix construction module, connected to the feature extraction module, which is used to construct a difference matrix according to the extracted current features and dynamically mark the storage units exceeding the threshold; An aging trajectory model construction module, connected to the difference matrix construction module, which is used to construct an aging trajectory model with the instantaneous current waveforms of the storage units with different marks in the time series as the input and the current features as the output, so as to obtain the aging rate; A priority weight allocation module, connected to the aging trajectory model construction module, which is used to allocate priority weights to the marked storage units according to the aging rate; A storage capacity acquisition module, connected to the priority weight allocation module, which is used to obtain the storage capacity of the storage unit under different priority weights; A pressure gradient application module, connected to the storage capacity acquisition module, which is used to apply a pressure gradient proportional to the current feature ratio to the storage capacity; A capacity attenuation evaluation module, connected to the pressure gradient application module, which is used to evaluate the capacity attenuation of the storage unit according to the change degree of the pressure gradient within a single storage unit in the time series, and associate the evaluation result with the priority weight to obtain the aging replacement requirement degree of the storage units in the memory.

Images