Method and system for acquiring optimal reading voltage of flash memory

The method of multi-peak cutting and Gaussian fitting accurately determines the optimal read voltage range for NAND flash memory, improving performance and reliability by achieving 99.9% accuracy.

CN119091936AInactive Publication Date: 2024-12-06合肥康芯威存储技术有限公司
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411570629.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-06
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing methods for determining the optimal read voltage of NAND flash memory are inaccurate and subjective, leading to read errors and affecting device performance and reliability.

Method used

A method involving multi-peak cutting, Gaussian fitting, and standard normal distribution to objectively determine the optimal read voltage range, using threshold voltage distribution data and quantile tables to achieve high accuracy.

Benefits of technology

Improves the accuracy of optimal read voltage determination to 99.9%, enhancing the read performance and reliability of flash memory.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091936A_ABST
    Figure CN119091936A_ABST
Patent Text Reader

Abstract

The invention provides a method and a system for acquiring the optimal read voltage of a flash memory, and relates to the field of memorizing.The method at least comprises the following steps: providing threshold voltage distribution source data of the flash memory, the source data comprising the number of memory units conducted under a plurality of voltage shifts; according to the type of the flash memory, performing multi-peak cutting on the source data to obtain multiple groups of basic data of flash memory threshold voltage distribution, and each group of basic data obeying normal distribution; fitting each group of basic data to obtain an expected value and a variance, and converting the normal distribution into standard normal distribution; and determining the interval range of the optimal read-write voltage of the flash memory according to a quantile table of standard normal distribution, the expected value and the variance. According to the method and the system for acquiring the optimal reading voltage of the flash memory, provided by the invention, the accuracy and objectivity of the optimal reading voltage can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of storage, and in particular to a method and system for obtaining an optimal read voltage of a flash memory. Background Art

[0002] As a high-density and low-cost storage medium, non-volatile storage media (NAND Flash) is widely used in various devices such as USB storage devices such as U disks, SD cards, mobile phones, and solid-state drives. As an important parameter in NAND flash memory, the optimal read voltage can affect the read performance and reliability of NAND flash memory. If the optimal read voltage is estimated inaccurately, it may cause read errors, which in turn affects the normal operation of the device. At present, the method of obtaining the optimal read voltage is to observe it with the naked eye, which leads to inaccurate and subjective determination of the optimal read voltage. Summary of the invention

[0003] The object of the present invention is to provide a method and system for obtaining an optimal read voltage of a flash memory, which can accurately and objectively obtain the optimal read voltage, thereby improving the read performance and reliability of the flash memory.

[0004] In order to solve the above technical problems, the present invention is achieved through the following technical solutions: The present invention provides a method for obtaining an optimal read voltage of a flash memory, which at least comprises the following steps: Providing threshold voltage distribution source data of the flash memory, the source data including the number of memory cells that are turned on at multiple voltage levels; According to the type of the flash memory, the source data is subjected to multi-peak cutting to obtain multiple groups of basic data of flash memory threshold voltage distribution, and each group of the basic data obeys a normal distribution; Fitting each set of the basic data to obtain expected values ​​and variances, and converting the normal distribution into a standard normal distribution; and The interval range of the optimal read and write voltage of the flash memory is determined according to the quantile table of the standard normal distribution, the expected value and the variance.

[0005] In one embodiment of the present invention, before performing the multi-peak cutting, a threshold voltage distribution curve of the flash memory is obtained, and obtaining the threshold voltage distribution curve comprises at least the following steps: Mapping the voltage levels to continuous natural numbers one-to-one; and A relationship curve between the continuous natural numbers and the number of storage cells is obtained as the threshold voltage distribution curve.

[0006] In one embodiment of the present invention, performing multi-peak cutting on the source data at least comprises the following steps: Obtaining an average value of the number of storage units under a plurality of the consecutive natural numbers; Obtaining a plurality of intersection points between a straight line whose ordinate is the average value and the threshold voltage distribution curve; Starting from a preset intersection point, along the direction in which the continuous natural number increases, obtain the first point where the number of storage units is less than or equal to the preset value, and record it as the right boundary point; Starting from the preset intersection, along the direction in which the continuous natural number decreases, obtain the first point where the number of storage units is less than or equal to the preset value, record it as the left boundary point, and the threshold voltage distribution curve between the right boundary point and the left boundary point is an original peak; Changing the preset intersection point, and repeating the steps of obtaining the right boundary point, the left boundary point, and the original peak until the right boundary point and the left boundary point of each intersection point are found; and The threshold voltage distribution curve is cut into a plurality of the original peaks, all data points in each of the original peaks are a group of basic data of flash memory threshold voltage distribution, and each group of the basic data obeys a normal distribution.

[0007] In one embodiment of the present invention, after the multi-peak cutting is completed, each of the original peaks is further translated to make each of the original peaks symmetrical about the origin.

[0008] In one embodiment of the present invention, the expected value and the variance are obtained according to the following relationship: ; = ; Where μ is the expected value, is the variance, c is the offset of the starting point of each original peak to the right about the origin, and b1 and b2 are fitting parameters.

[0009] In one embodiment of the present invention, the fitting parameters are obtained by the following relationship: L ; ; ; ; Among them, z i is the logarithmic value of the number of storage units in the ith data point, x i is the continuous natural number in the i-th data point, z(x i ) is the ideal value fitted to the ith data point, L is the sum of squares of deviations between the ideal value and the number of storage units, and b0, b1 and b2 are the fitting parameters.

[0010] In one embodiment of the present invention, the normal distribution is converted into the standard normal distribution according to the following relationship: = ; Where μ is the expected value, is the standard deviation, y is the number of storage cells, and Y is the normal value of the number of flash storage cells y.

[0011] In one embodiment of the present invention, in the quantile table of the standard normal distribution, the following relationship is satisfied: P(Y≤a)≥0.999; Wherein, Y is the normal value, a is the value of the normal value, and P(Y≤a) represents the probability of the normal value Y≤a in the threshold voltage distribution of the flash memory.

[0012] In one embodiment of the present invention, for two adjacent original peaks The interval [y1, y2] is the interval of the optimal read voltage of the flash memory, where y1 and y2 are obtained by the following relationship: ≥a* + ; ≤-a* +u; Where μ is the expected value, is the standard deviation, and a is the value of the normal value Y when P(Y≤a)≥0.999.

[0013] The present invention also provides a system for obtaining an optimal read voltage of a flash memory, comprising at least: flash memory; A data reading module, used to obtain threshold voltage distribution source data of the flash memory; and The calculation module executes the above-mentioned method for obtaining the optimal read voltage of the flash memory according to the information of the data reading module.

[0014] As described above, the present invention provides a method and system for obtaining the optimal read voltage of a flash memory, which can objectively and accurately obtain the interval range of the optimal read voltage through Gaussian fitting and least squares method, and the accuracy of the optimal read voltage can reach more than 99.9%, thereby improving the read performance and reliability of the flash memory. Moreover, the acquisition method and system provided by the present invention can shorten the calculation amount of the fitting process, improve the fitting efficiency, and conveniently and quickly obtain the optimal read voltage.

[0015] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0017] Figure 1 The figure is a flow chart of a method for obtaining an optimal read voltage of a flash memory according to an embodiment of the present invention.

[0018] Figure 2 for Figure 1 Flowchart of multi-peak cutting.

[0019] Figure 3 for Figure 1 Flowchart for obtaining expected value and variance in .

[0020] Figure 4 for Figure 1 Schematic diagram of determining the interval range of the optimal reading voltage.

[0021] Figure 5 FIG. 4 is a schematic diagram of a system for obtaining an optimal read voltage of a flash memory according to an embodiment of the present invention.

[0022] Description of labels: 100, flash memory; 200, data reading module; 300, calculation module. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1 As shown, the present invention provides a method for obtaining an optimal read voltage of a flash memory, which at least includes steps S11-S14.

[0025] Step S11, obtaining threshold voltage distribution source data of the flash memory, the source data including the number of memory cells that are turned on at multiple voltage levels.

[0026] Step S12: Perform multi-peak cutting on the source data according to the type of the flash memory to obtain multiple groups of basic data, and each group of basic data obeys a normal distribution.

[0027] Step S13: Fit each set of basic data to obtain expected value and variance, and transform the normal distribution into a standard normal distribution.

[0028] Step S14: Determine the interval range of the optimal reading voltage according to the quantile table, expected value and variance of the standard normal distribution.

[0029] See also Figure 1 As shown, in one embodiment of the present invention, in step S11, the threshold voltage distribution source data of the flash memory is obtained, and the source data includes the number y of memory cells that are turned on under multiple voltage levels x'. Among them, the source data is the number y of memory cells that are turned on under a set of decreasing and increasing voltage levels x', and one voltage level x' corresponds to the number y of memory cells. The present invention does not limit the method for obtaining the threshold voltage distribution source data, which can be selected according to actual conditions. In this embodiment, the threshold voltage distribution source data is obtained, for example, by a threshold voltage tool, and the threshold voltage distribution source data is stored in the array y_coor in the form of the number y of memory cells.

[0030] See also Figure 1 As shown, in one embodiment of the present invention, after obtaining the threshold voltage distribution source data, a threshold voltage distribution curve is also drawn. Specifically, since the voltage level x' corresponds to the number of storage cells y, after the voltage level x' is mapped one-to-one with the continuous natural number x, the continuous natural number x corresponds to the number of storage cells y, and multiple data points are obtained, each of which represents the number of storage cells y under a continuous natural number x. Then, based on the multiple data points, the threshold voltage distribution curve is drawn with the continuous natural number x as the horizontal coordinate and the number of storage cells y as the vertical coordinate. In this embodiment, the continuous natural number x starts from 1, for example, and x is 1, 2, 3, ..., n, and the step of mapping the voltage level x' to the continuous natural number x one-to-one specifically includes: the first voltage level x' corresponds to the continuous natural number x=1, the second voltage level x' corresponds to the continuous natural number x=2, and so on. In the threshold voltage distribution source data, the voltage level x' is a nonlinearly increasing string. By mapping the voltage level x' to the continuous natural number x one-to-one, subsequent fitting can be facilitated, the amount of data calculation can be reduced, and the fitting efficiency can be improved.

[0031] See also Figure 1 to Figure 2As shown, in one embodiment of the present invention, after drawing the threshold voltage distribution curve, in step S12, the threshold voltage distribution source data is subjected to multi-peak cutting according to the type of flash memory. According to the number of storage cells, flash memory is divided into multiple types such as SLC (Single-Level Cell) and TLC (Triple-Level Cell), wherein the threshold voltage distribution curve of the SLC type obeys a bimodal normal distribution, and the threshold voltage distribution curve of the TLC type includes 8 peaks and obeys a multi-peak normal distribution. Different types of flash memory have different numbers of peaks, so it is necessary to perform multi-peak cutting on the threshold voltage distribution source data according to the type of flash memory. In this embodiment, the steps of multi-peak cutting are described by taking the TLC type as an example. Specifically, in this embodiment, the steps of multi-peak cutting include, for example, steps S121-S125.

[0032] Step S121, calculate the average value y of all storage unit quantities y under different consecutive natural numbers x avg .

[0033] Step S122: set y=y avg Intersecting with the threshold voltage distribution curve, multiple intersection points are obtained.

[0034] Step S123, starting from a preset intersection point, along the direction of increasing continuous natural number x, obtain the first point where the number of storage units y ≤ the preset value, and record it as the right boundary point.

[0035] Step S124, starting from the preset intersection point, along the direction of decreasing x by a continuous natural number, obtain the first point where the number of storage cells y ≤ the preset value, record it as the left boundary point, and the threshold voltage distribution curve between the right boundary point and the left boundary point is an original peak.

[0036] Step S125, changing the preset intersection point, and repeating the operation steps of obtaining the right boundary point, the left boundary point and the original peak until the right boundary point and the left boundary point of each intersection point are found.

[0037] Step S126, cutting the threshold voltage distribution curve into a plurality of original peaks, all data points in each original peak being a group of basic data of the flash memory threshold voltage distribution, and each group of basic data obeys a normal distribution.

[0038] See also Figure 1 to Figure 2 As shown, in one embodiment of the present invention, in step S121, the average value y of all storage unit quantities y under different consecutive natural numbers x is calculated. avg . See Figure 1 to Figure 2 As shown, in one embodiment of the present invention, the average value y is obtained. avg Then, in step S122, let y=y avgand the threshold voltage distribution curve to obtain multiple intersections. In this embodiment, after obtaining multiple intersections, the horizontal coordinates of all intersections are recorded, and the values ​​of all horizontal coordinates are recorded as a global array x_coor according to the direction of increasing or decreasing horizontal coordinates. The global array x_coor includes multiple values, namely, x_coor[1], x_coor[2]...x_coor[i]...x_coor[i+1]..., x_coor[i] is the value of the horizontal coordinate of the i-th intersection.

[0039] See also Figure 1 to Figure 2 As shown, in one embodiment of the present invention, after the intersection is obtained, in step S123, starting from a preset intersection, along the direction of increasing continuous natural number x, the first point where the number of storage cells y ≤ the preset value is obtained, which is recorded as the right boundary point. Specifically, in the threshold voltage distribution curve, starting from x_coor[i], along the direction of increasing continuous natural number x, the first point where the number of storage cells y ≤ the preset value is found, which is recorded as the right boundary point. The preset value is, for example, 0 or 1.

[0040] See also Figure 1 to Figure 2 As shown, in one embodiment of the present invention, after obtaining the right boundary point, in step S124, starting from the preset intersection point, along the direction of the continuous natural number x decreasing, the first point where the number of storage cells y ≤ the preset value is obtained, which is recorded as the left boundary point, and the threshold voltage distribution curve between the right boundary point and the left boundary point is an original peak. Specifically, in the threshold voltage distribution curve, starting from x_coor[i], along the direction of the continuous natural number x decreasing, the first point where the number of storage cells y ≤ the preset value is found, which is recorded as the left boundary point. The preset value is, for example, 0 or 1.

[0041] See also Figure 1 to Figure 2 As shown, in one embodiment of the present invention, after obtaining an original peak, in step S125, the preset intersection is changed, and the operation steps of obtaining the right boundary point, the left boundary point and the original peak are repeated until all values ​​in the global array x_coor have gone through the operation steps, and the right boundary point and the left boundary point of each intersection are found. In step S126, the entire threshold voltage distribution curve is cut into multiple original peaks, and all data points in each original peak are a group of basic data of the flash memory threshold voltage distribution, and each group of basic data obeys a normal distribution.

[0042] See also Figure 1 As shown, in one embodiment of the present invention, after multi-peak cutting, each original peak is translated so that each original peak is symmetrical about the origin, and the translated original peak still obeys the normal distribution and satisfies the normal distribution formula ~N(μ, ), a and b are the parameters of the normal distribution, c is the offset of the starting point of each original peak to the right of the origin, x is a continuous natural number, y is the number of storage units, μ is the expected value, is the variance. Taking the logarithm of both sides of the normal distribution formula, we can get lny=lna-(1 / b)(x 2 -2cx+c 2 )=(lna-c 2 / b)+(2c / b)x-(1 / b)x 2 , let b0=(lna-c 2 / b), b1=(2c / b), b2=-(1 / b), ln(y)=z, get z i =b0+b1*x+b2*x i 2 Among them, b0, b1 and b2 are unknown numbers, z i is the logarithm of the ordinate of the i-th data point, that is, the logarithm of the stored quantity y, x i is the horizontal coordinate value of the ith data point, that is, a continuous natural number x. In this embodiment, by translating the peak, the calculation amount of subsequent fitting can be simplified and the fitting efficiency can be improved.

[0043] See also Figure 1 and Figure 3 As shown, in one embodiment of the present invention, after the original peak is translated, in step S13, each set of basic data that conforms to the normal distribution is fitted to obtain the expected value and variance, and the normal distribution is converted into a standard normal distribution. Specifically, the step of obtaining the expected value and variance includes, for example, steps S131-S134.

[0044] Step S131: Perform Gaussian fitting on the data points in each group of basic data to obtain ideal values ​​of Gaussian fitting.

[0045] Step S132: Calculate the sum of squares of deviations between the ideal value and the number of storage units y.

[0046] Step S133, obtaining Gaussian fitting parameters when the sum of squares is minimized.

[0047] Step S134: Obtain the expected value according to the Gaussian fitting parameters and variance = .

[0048] See also Figure 1 and Figure 3 As shown, in one embodiment of the present invention, in step S131, according to the formula z(x)= + *x+ , perform Gaussian fitting on the data points in each set of basic data, where z(x) is the ideal value of Gaussian fitting, x is a continuous natural number, and b0, b1 and b2 are the parameters of Gaussian fitting.

[0049] See also Figure 1 and Figure 3 As shown, in one embodiment of the present invention, after obtaining the ideal value z(x) after Gaussian fitting, in step S132, the fitting deviation e of each data point is calculated. i =z i -z(x i ), e i is the fitted deviation of the ith data point, z i is the logarithm of the ordinate of the i-th data point, that is, the logarithm of the stored quantity y, x i is the horizontal coordinate value of the i-th data point, that is, the continuous natural number x, z (x i ) is the ideal value for Gaussian fitting of the i-th data point. The fitting deviation of all data points can be expressed in matrix form E=Z-XB, that is, , and calculate the sum of squares of the fitted deviations L for all data points .

[0050] See also Figure 1 and Figure 3 As shown, in one embodiment of the present invention, after obtaining the sum of squares L of the fitting deviations of all data points, in step S133, the b0, b1 and b2 corresponding to when L is minimum are obtained according to the least squares method, the matrix rank and Gaussian elimination method. Specifically, L is partially derived with respect to b0, b1 and b2 respectively, and L is minimum when the derivative is 0, that is: , , , Transform the above formula into: , , , Written in matrix form: , Let X = ,B= , Z = , Then the matrix form can be written as X)B=XZ, the matrix B is the solution of the linear equations. If the equations have a unique solution, then r X)=r( , ), where r represents the rank of the matrix. Moreover, since any two rows of the matrix X are not proportional, r(X)=3, so r X)=r(X)=r( , )=3, the linear equation system has a unique solution, namely and is unique. Because r X)=3, the number of rows and columns of the matrix is ​​relatively small. For example, Gaussian elimination method is used to solve B and obtain the ideal fitting parameters. and .

[0051] See also Figure 1 and Figure 3 As shown, in one embodiment of the present invention, the parameters are obtained and Then, in step S134, according to z(x)= + *x+ , we can calculate the axis of symmetry of the quadratic polynomial to be x= +c, this axis of symmetry is also the axis of symmetry of the normal distribution curve , and the translation of the normal distribution curve will not change the volatility of the data, so unchanged, thus calculating the expected value μ= +c and variance = , then according to the expected value μ and variance , transform the basic data that obeys the normal distribution into the standard normal distribution Y= , where Y is the normal value of the number of flash memory cells y.

[0052] See also Figure 1 As shown, in one embodiment of the present invention, after the normal distribution is converted into a standard normal distribution, in step S14, the range of the optimal read and write voltage is determined according to the quantile table, expected value and variance of the standard normal distribution. Among them, some quantile tables of the standard normal distribution are shown in Tables 1 and 2. In Table 1, when a=1.65, P(Y≤1.65)=0.9505, where Y is the normal value of the number of flash memory storage cells y, a is the value of the normal value Y, and P(Y≤a) represents the probability of the normal value Y≤a in the threshold voltage distribution of the flash memory. If P(X≤a)≥0.999, this means that the probability of X≤a is at least 0.999, that is, it can be considered that X≤a will occur with a high probability. Comparing the standard normal distribution quantile table in Table 2, it is obtained that when a≥3.1, P(X≤a)≥0.999.

[0053] Table 1. Partial quantile table of standard normal distribution

[0054] Table 2. Another part of the quantile table of the standard normal distribution

[0055] See also Figure 1 and Figure 4 As shown, in one embodiment of the present invention, after determining a≥3.1, according to the standard normal distribution Y= , for any peak, P( )≥0.999, then >=a, >=a* + , for another adjacent peak, P( >= )≥0.999, then -≤a, ≤-a* +u, the interval [y1, y2] is the trough position of the two peaks, that is, the interval of the optimal read voltage. In this embodiment, by limiting a≥3.1, the probability that the optimal read voltage falls in the interval [y1, y2] is increased to 99.9%, thereby improving the accuracy of the optimal read voltage. In other embodiments of the present invention, the value of a can also be further increased to increase the probability of P(X≤a) and further improve the accuracy of the optimal read voltage.

[0056] See also Figure 5 As shown, based on the above-mentioned method for obtaining the optimal read voltage of a flash memory, the present invention also provides a system for obtaining the optimal read voltage of a flash memory, and the acquisition system at least includes: a flash memory 100, a data reading module 200, and a calculation module 300. Among them, the data reading module 200 is used to obtain the threshold voltage distribution source data of the flash memory 100, and the source data includes the number of storage cells that are turned on under multiple voltage levels. The calculation module 300 executes the above-mentioned method for obtaining the optimal read voltage of the flash memory according to the information of the data reading module 200.

[0057] The present invention provides a method and system for obtaining the optimal read voltage of a flash memory, which obtains the optimal read voltage by multi-peak cutting, normal distribution standardization and a quantile table of a standard normal distribution, and can improve the accuracy of the optimal read voltage to 99.9%, improve the accuracy and objectivity of the optimal read voltage, and thus improve the read performance and reliability of the flash memory. Moreover, the present invention can reduce the amount of calculation and improve the efficiency of obtaining the optimal read voltage by mapping the threshold voltage to a continuous natural number one-to-one.

[0058] The embodiments of the present invention disclosed above are only used to help illustrate the present invention. The embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for obtaining an optimal read voltage of a flash memory, characterized in that: At least the following steps are included: Providing threshold voltage distribution source data of the flash memory, the source data including the number of memory cells that are turned on at multiple voltage levels; According to the type of the flash memory, the source data is subjected to multi-peak cutting to obtain multiple groups of basic data of flash memory threshold voltage distribution, and each group of the basic data obeys a normal distribution; Fitting each set of the basic data to obtain expected values ​​and variances, and converting the normal distribution into a standard normal distribution; and The interval range of the optimal read and write voltage of the flash memory is determined according to the quantile table of the standard normal distribution, the expected value and the variance.

2. The acquisition method according to claim 1, characterized in that: Before performing the multi-peak cutting, a threshold voltage distribution curve of the flash memory is obtained, and obtaining the threshold voltage distribution curve comprises at least the following steps: Mapping the voltage levels to continuous natural numbers one-to-one; and A relationship curve between the continuous natural numbers and the number of storage cells is obtained as the threshold voltage distribution curve.

3. The acquisition method according to claim 2, characterized in that: Performing multi-peak cutting on the source data at least comprises the following steps: Obtaining an average value of the number of storage units under a plurality of the consecutive natural numbers; Obtaining a plurality of intersection points between a straight line whose ordinate is the average value and the threshold voltage distribution curve; Starting from a preset intersection point, along the direction in which the continuous natural number increases, obtain the first point where the number of storage units is less than or equal to the preset value, and record it as the right boundary point; Starting from the preset intersection, along the direction in which the continuous natural number decreases, obtain the first point where the number of storage units is less than or equal to the preset value, record it as the left boundary point, and the threshold voltage distribution curve between the right boundary point and the left boundary point is an original peak; Changing the preset intersection point, and repeating the steps of obtaining the right boundary point, the left boundary point, and the original peak until the right boundary point and the left boundary point of each intersection point are found; and The threshold voltage distribution curve is cut into a plurality of the original peaks, all data points in each of the original peaks are a group of basic data of flash memory threshold voltage distribution, and each group of the basic data obeys a normal distribution.

4. The acquisition method according to claim 3, characterized in that: After completing the multi-peak cutting, each of the original peaks needs to be translated so that each of the original peaks is symmetrical about the origin.

5. The acquisition method according to claim 4, characterized in that: The expected value and the variance are obtained according to the following relationship: ; = ; Where μ is the expected value, is the variance, c is the offset of the starting point of each original peak to the right about the origin, and b1 and b2 are fitting parameters.

6. The acquisition method according to claim 5, characterized in that: The fitting parameters are obtained by the following relationship: L ; ; ; ; Among them, z i is the logarithmic value of the number of storage units in the ith data point, x i is the continuous natural number in the i-th data point, z(x i ) is the ideal value fitted to the ith data point, L is the sum of squares of deviations between the ideal value and the number of storage units, and b0, b1 and b2 are the fitting parameters.

7. The acquisition method according to claim 6, characterized in that: The normal distribution is converted to the standard normal distribution according to the following relationship: Y= ; Where μ is the expected value, is the standard deviation, y is the number of storage cells, and Y is the normal value of the number of flash storage cells y.

8. The acquisition method according to claim 7, characterized in that: In the quantile table of the standard normal distribution, the following relationship is satisfied: P(Y≤a)≥0.999; Wherein, Y is the normal value, a is the value of the normal value, and P(Y≤a) represents the probability of the normal value Y≤a in the threshold voltage distribution of the flash memory.

9. The acquisition method according to claim 8, characterized in that: For two adjacent original peaks, the interval [y1, y2] is the interval of the optimal read voltage of the flash memory, where y1 and y2 are obtained by the following relationship: ≥a* + ; ≤-a* +u; Where μ is the expected value, is the standard deviation, and a is the value of the normal value Y when P(Y≤a)≥0.

999.

10. A system for obtaining an optimal read voltage of a flash memory, characterized in that: At least: flash memory; A data reading module, used for acquiring threshold voltage distribution source data of the flash memory; as well as The calculation module executes the method for obtaining the optimal read voltage of the flash memory according to any one of claims 1 to 9 according to the information of the data reading module.

Citation Information

Patent Citations

  • Threshold distribution fitting method, device and system for flash memory

    CN112735502A

  • Deep learning-based programming verification modeling and voltage estimation for memory devices

    CN114792546A

  • Method, system and equipment for processing threshold voltage data of memory and medium

    CN118644580A