KR20170079451A - Repair analysis method for the memory device and apparatus thereof - Google Patents

Abstract

A repair analysis method and apparatus for finding an optimal repair result using spares of various structures in response to an error occurring in a memory device of a multi-block structure are disclosed. The repair analysis method of the present invention includes an address of pivot errors, A memory unit including a first memory for storing addresses and a second memory for storing one of a column and a row representing the address of each of the non-pivot errors and a block address of the non-pivot errors; And a comparison result for comparing the non-pivot error information with the pivot error information of the selected pivot errors to determine whether the spare pivot error information can be repaired by the assigned spare, The resulting single analyzer, including .

Description

TECHNICAL FIELD [0001] The present invention relates to a repair analysis method and apparatus,

The present invention relates to a memory device, and more particularly, to a repair analysis method and apparatus for obtaining an optimal repair result by using spares having various structures in response to an error occurring in a memory device of a multi-block structure.

Repair is a necessary method to improve the yield of a memory device. As the memory capacity of a memory device increases, recent memory devices are designed to include a plurality of memory blocks and a spare (spare memory) for repairing memory cell errors in the memory block.

Currently, a large number of repair analysis algorithms performing redundancy analysis are being developed for repairing memory devices. However, most repair analysis algorithms are being developed to perform analysis of errors in a single memory block. In this regard, Korean Patent No. 10-1133689 (repair analysis apparatus and method) discloses a method and apparatus for analyzing errors of a single memory block.

As the degree of integration of the memory device increases, a multi-block structure including a plurality of memory blocks or a three-dimensional memory device is being developed. There is a limitation in applying the repair analysis algorithm that performs the repair analysis on the error of a single memory block to the multi-block structure or the three-dimensional memory device.

Therefore, it is necessary to propose a solution that can obtain an optimal repair rate by using a spare of various structures in a memory device of a multi-block structure.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a repair analysis method and apparatus capable of obtaining an optimal repair rate using a spare of various structures for a memory cell error of a memory device.

Another problem to be solved by the present invention is to provide a memory device capable of storing error information of a memory device with a small memory capacity and analyzing stored error information to analyze a memory cell error of a spare or two or more memory blocks, A repair analysis method and apparatus capable of obtaining an optimum repair result for resolving an error by using a spare having a variety of structures.

A further object of the present invention is to provide a repair analysis method and apparatus capable of obtaining an optimum repair result for error correction of a memory device having a multi-block structure by using block information related to an error of a memory cell .

The repair analysis apparatus of the present invention includes a first memory for storing an address and a block address of pivot errors, a second memory for storing a block address of one of columns and rows representing the address of each of the non-pivot errors, Wherein the pivot error information includes an address of the pivot errors and a block address, and the non-pivot error information including the address of the pivot errors and the remaining address representing the address of the non- A memory unit for providing non-pivot error information including an address of the pivot errors and a block address; And selecting a column address or a row address for each of the pivot errors in the pivot error information to orient the spare to repair errors of the memory blocks, comparing the non-pivot error information with the pivot error information of the selected pivot errors The first case in which the non-pivot error and the selected pivot error have the same address of the same memory block and the second case in which the non-pivot error and the selected pivot error are in different memory blocks arranged in one direction and having the same address Determines that the first case is repairable, and generates a comparison result for determining that the second case is repairable when the spare having a length corresponding to all of the other memory blocks arranged in the one direction is allocated , By the non-pivot error You the group comparison single analyzer providing a pivoting cover result; characterized in that it comprises a.

Also, the repair analysis method of the present invention includes the steps of storing an address and a block address of pivot errors in a first memory; Storing one of a column and a row representing the address of each of the non-pivot errors and a block address of the non-pivot errors in a second memory; Selecting a column address and a row address of pivot error information including an address of the pivot errors and a block address to define a spare direction; Comparing the non-pivot error information with the pivot error information of the selected pivot errors to determine a first case where the non-pivot error and the selected pivot error have the same address of the same memory block, The first case is determined to be repairable, and the second case is determined to be a length corresponding to all of the other memory blocks arranged in the one direction Generating a comparison result that determines that repair is possible if the spare having the spare is allocated; And providing a comparison result of the non-pivot error information as a non-pivot cover result.

The present invention can obtain optimal repair rates for memory cell errors in memory devices using spurs of various structures.

In addition, the present invention can store error information of a memory device with a small memory capacity, analyze a stored error information, and store a spare that is shared in an adjacent memory block or a variety of memory cells that can repair a memory cell error of two or more memory blocks at a time Spaces of structures can be used.

Further, the present invention can obtain an optimum repair result for eliminating a memory cell error of a memory device of a multi-block structure having two or more memory blocks.

1 is a layout of a memory device according to the present invention;
2 is an illustration of an error line;
3 illustrates a structure of a PCAM that stores pivot errors of the present invention;
4 is a diagram illustrating a structure of an NPCAM that stores a non-pivot error of the present invention;
5 illustrates memory cell errors occurring in multiple blocks;
FIG. 6 is a diagram illustrating a state in which the error of FIG. 5 is classified into a pivot error and a non-pivot error and stored in the PCAM and the NPCAM;
FIGS. 7 to 16 are diagrams illustrating a method of determining an error as a pivot error and a non-pivot error; FIG.
17 to 19 are diagrams for explaining a method of repairing a cell line of an adjacent block;
20 is a block diagram showing an embodiment of the repair analysis apparatus of the present invention;
21 is a detailed circuit diagram showing an embodiment of a single analyzer;
22 is a detailed circuit diagram of CAC.
23 is a detailed circuit diagram of RAC;
24 is a table showing the result of error collection;

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. Like reference numerals in the drawings denote like elements.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

The terms first, second, etc. may be used to describe various components, but the components are not limited by the terms, and the terms are used only for the purpose of distinguishing one component from another Is used.

The repair analysis method of the present invention uses various types of spare.

Referring to FIG. 1, the repair analysis method of the present invention can be applied not only to a single memory block but also to a multi-block structure. A memory device of a multi-block structure can be arranged such that the memory blocks have a block address in rows and columns, and Fig. 1 illustrates four memory blocks B00, B01, B10 and B11. Each of the memory blocks B00, B01, B10, and B11 can have a number of row cell lines and column cell lines so as to have memory cells of the same capacity, and the number of memory cells included in the row cell lines and the column cell lines is equal to or different from each other .

As shown in FIG. 1, a spare may be classified into a local spare, a common spare, and a global spare. In FIG. 1, the local spare, the common spare, and the global spare are configured as a line type in which memory cells are arranged in a line. At least one of a local spare, a common spare, and a global spare may be divided into a short corresponding to one memory block and a wide corresponding to two or more memory blocks, respectively.

That is, FIG. 1 illustrates a short local spare line, a wide local spare line, a short commons spare line, a short global spare line, and a wide global spare line, and various spares as described above may be arranged in various formats have. In FIG. 1, the shortcommon spare line may be arranged to form a column or row line between the memory blocks B00, B01, B10 and B11, and may be a short local spare line, a wide local spare line, a short global spare line, The spare line may be arranged to form a column or row line at the top, bottom, left, and right of the memory blocks B00, B01, B10, and B11.

Among the spares illustrated in FIG. 1, the short local spare line is for repairing an error of a specific memory block, the wide local spare line is for repairing an error of a specific memory block line (block line), and the short local spare line The short global spare line is for repairing an error of one of the plurality of memory blocks, and the wide global spare line is for repairing an error of the memory block line (block line) will be.

Errors that may occur in the memory blocks B00, B01, B10, and B11 may be classified into various types as shown in FIG.

Referring to FIG. 2, errors may be classified into a single fault, a pivot fault, a sparse fault line, and a must-repare line.

A single error is an error that does not share row and column addresses with other errors.

And a pivot error is an error with a row address and a column address that are different from previously found errors. A single error is always a pivot error because it does not share row and column addresses with other errors.

A sparse error line can be divided into a low spar error line and a column spar error line depending on the direction, and a low spar error line indicates that the low line has one or more errors or the same number of low spare lines Or when it has a small error. Column sparse error lines can also be defined similar to low sparse error lines.

And the must repair line is a line where there are more errors than the number of spare lines that can be used to repair the cell line. For example, if there are three errors in one row cell line, each error may be repaired using three column spare lines, respectively. However, if the number of spare lines that can be used for repairing the row cell line is two, the row cell line in which three errors exist corresponds to the repair word line.

The present invention has a memory unit for analyzing a memory device having the above-described error, and the memory unit can be composed of a content addressable memory (CAM). CAM refers to the memory that accesses the data by addressing the contents of the actual data.

The memory unit of the present invention may include a first memory for storing a pivot error and a second memory for storing a non-pivot error. The first memory may be defined as PCAM (Pivot CAM), and the second memory may be defined as NPCAM (Non-Pivot CAM). As described above, the reason why the error is divided into pivots and non-pivots according to the present invention is to utilize the characteristics of pivot errors in a single analyzer, which will be described later.

The PCAM can be illustrated as in FIG.

The PCAM includes a plurality of lines, and each line includes an enable flag, a raw address, a column address, a block address, and a must-repair information .

The PCAM stores an error having an independent row address and column address for previously stored errors. That is, errors stored in the PCAM are defined as pivot errors.

In the PCAM, the enable flag indicates whether the line of the PCAM is caused by an error, and if the line is enabled by an error, "1" is stored as an enable flag, In the disabled state, "0" is stored as an enable flag.

And, in the PCAM, the row address and the column address are for storing the address of the pivot error, and the block address is for storing the block address of the storing pivot error.

Also, the PCAM stores the so-called repair information in order to show the state of the so-called repair on the memory device. The must update information includes a block low matter flag, a block column must flag, an adjacent block low matter flag, and an adjacent block column must flag. The repair line is set when an error occurs in excess of the number of spare blocks available for each memory block and can be set separately for each memory block. Therefore, the must-repair information includes flags for the rows and columns of the current memory block (block low-mast flag, block column must-flag), a row for the adjacent memory block block and a flag for the column of another adjacent memory block block Flag, adjacent block column must flag), and each flag of the soccer repair information stores "1" in the case of a soccer repair and "0" in the case of not a soccer repair.

The number of lines of the PCAM is equal to the total number of spares. The pivot error can be repaired by at least one repair. If the number of pivot errors is greater than the total number of spares, the error pattern can not be repaired with a given number of spares.

The NPCAM can be illustrated as shown in FIG.

The NPCAM stores an error having the same row address or column address as the pivot error previously stored in the PCAM line by line. Each line of the NPCAM includes an enable flag, a PCAM pointer (PCAM Pointer), a Descriptor, an address, and a block address.

In the NPCAM, the enable flag indicates whether the line of the NPCAM is caused by an error, "1" is stored in the enable flag when it is caused by an error, and & .

 The PCAM pointer represents the position of the PCAM line shared by the NPCAM line, and the position of the PCAM line can be sequentially counted from 0. The descriptor represents whether the address is a row address or a column address. When the address is low, "0" is stored in the descriptor. When the address is a column, "1" is stored in the descriptor.

In NPCAM, the address is the column address or row address specified by the descriptor in the address of the non-pivot error, and the block address is the block address of the stored non-pivot error. You may experience pivot errors in other memory blocks than in pivot errors, in which case the block address (PCAM block address) of the pivot error and the block address (NPCAM block address) of the non-pivot error are different.

The number of lines of the NPCAM can be defined by the state of the MAST REPEAR, and the memory including the PCAM and the NPCAM has all of the error information except for the MAST REPAIR line.

Figure 5 illustrates two memory blocks B0, B1 with errors. The two memory blocks B0 and B1 each contain an error, and the location where the error occurred is indicated by "X ". The numbers # 1 to # 11 shown in the two memory blocks B0 and B1 in FIG. 5 represent the order of analyzing errors for convenience of explanation. The address lines are displayed by writing 0 to 5 in the column and the row direction, respectively, in order to display the address of the cell line in each block. The address line is not actually implemented but is shown for the convenience of explanation.

Column sparing lines and row global spare lines are illustrated corresponding to the two blocks B0, B1 in FIG. 5, and the column spare lines are understood to include a center column short common spare line and a column short local spare line on both sides .

FIG. 6 shows PCAM and NPCAM corresponding to errors of the two memory blocks B0 and B1 in FIG.

Figure 5 represents two memory blocks B0, B1 forming one row block line. Therefore, the PCAM of FIG. 6 has an enable flag for each line, a row and column address (Addr.R / C), a block address (Block Addr.), A block low- (Must R / C) and an adjacent block lowermost flag (A-Must). Correspondingly, each line of the NPCAM includes an enable flag, a PCAM pointer, a descriptor, an address Addr, and a block address (Block Addr.).

Hereinafter, errors of the two memory blocks B0 and B1 of FIG. 5 are expressed in the PCAM and the NPCAM by the analysis of the present invention with reference to FIGS. 7 to 16. FIG.

The present invention analyzes errors in the order shown in FIG. 5 and stores error information in the PCAM and the NPCAM.

First, errors corresponding to # 1 in FIG. 5 are analyzed as shown in FIG. Since error # 1 is formed in memory block B0, it has a block address of "0 ", and has addresses (1, 2) in memory block B0. Error # 1 is a pivot error. Therefore, the error information of error # 1 is stored in line 0 of the PCAM according to the analysis of the present invention as shown in FIG.

1 ", row address" 1 ", column address" 2 ", block address" 0 ", and column address" 0 "are written to the line 0 of the PCAM in response to the error # A block low must flag (Must R) "0", a block column must flag (Must C) "0" and an adjacent block must flag "0". Since the line 0 is generated by the error, "1" is stored in the enable flag, and since there is no repair line in the current memory block and the adjacent memory block, the block low must flag (Must R), the block column must flag ) And "0" are stored in the adjacent block must flag.

Thereafter, as shown in Fig. 9, the error corresponding to # 2 in Fig. 5 is analyzed. Since the error # 2 is formed in the memory block B0, it has a block address of "0 " and has addresses (1, 3) in the memory block B0. Error # 2 is an error with the same row address as the pivot error of line 0 of the previously stored PCAM, i.e., a non-pivot error. Therefore, the error information of error # 2 is stored in line 0 of the NPCAM according to the analysis of the present invention as shown in FIG.

That is, a line is generated at the position "0" of the NPCAM in response to the error # 2, and the line 0 of the NPCAM has the enable flag "1 ", the PCAM pointer" 0 ", the descriptor "1 & "0" is stored. Since the line 0 of the NPCAM is generated by the error, "1" is stored in the enable flag, "0" is stored in the PCAM pointer since the position of the PCAM line shared by the line 0 of the NPCAM is "0" Quot; 1 "is stored in the descriptor and the address (column address)" 3 "specified in the descriptor is stored in the address. Then, the block address of the error # 2 is stored in the block address of the NPCAM line 0 as "0 ".

 Thereafter, as shown in FIG. 11, the error corresponding to # 3 in FIG. 5 is analyzed. Since error # 3 is formed in memory block B0, it has a block address of "0 ", and has the address of (4, 1) in memory block B0. Error # 3 is a pivot error. Therefore, the error information of error # 3 is stored in line 1 of the PCAM according to the analysis of the present invention as shown in FIG.

Thereafter, errors corresponding to # 4, # 5 and # 6 in FIG. 5 are sequentially analyzed as shown in FIG. Since errors # 4 and # 5 are formed in the memory block B1, they have a block address of "1" and have addresses (1, 1) and (1, 2) in the memory block B1. Errors # 4 and # 5 are errors that have the same row address as the pivot error of line 0 of the previously stored PCAM, that is, a non-pivot error. Therefore, the error information of errors # 4 and # 5 is stored in line 1 and line 2 of the NPCAM according to the analysis of the present invention as shown in FIG.

Thereafter, as shown in FIG. 13, the error corresponding to # 6 in FIG. 5 is analyzed. Since the error # 6 is formed in the memory block B1, it has a block address of "1" and has addresses (1, 4) in the memory block B1. Error # 6 is a pivot error. Therefore, the error information of error # 6 is stored in line 3 of the PCAM according to the analysis of the present invention as shown in FIG.

However, as shown in Fig. 13, the memory block B1 has two column spare lines for repairing an error formed in one row cell line. Therefore, the row cell line in which # 4, # 5 and # 6 are formed is a must repair line having an error exceeding the usable spare line. In this case, error information of Errors # 4, # 5 and # 6, that is, lines 1, 2 and 3 of NPCAM are deleted and error information of line 0 of PCAM shared by NPCAM lines 1, 2 and 3 Quot; 1 "is stored in the adjacent block must flag included in the adjacent block must flag.

Thereafter, errors corresponding to # 7, # 8, and # 9 in FIG. 5 are analyzed as shown in FIG. Error # 7 is analyzed as a pivot error, and error information of error # 7 is stored on line 2 of the PCAM as shown in FIG. Errors # 8 and # 9 are analyzed as a non-pivot error, and error information of Errors # 8 and # 9 is stored in Lines 1 and 2 of the NPCAM as shown in FIG.

Thereafter, errors corresponding to # 10 and # 11 in FIG. 5 are analyzed. Errors # 10 and # 11 are analyzed as pivot error or non-pivot error, and error information of Errors # 10 and # 11 are stored in line 3 of the PCAM and line 3 of the PCAM as shown in FIG.

In the above description, errors of repair lines such as errors # 4, # 5, and # 6 are not stored in PCAM and NPCAM. Therefore, the storage capacity required for PCAM and NPCAM to store error information can be reduced.

The above described methods of repairing the errors # 4, # 5, and # 6 can be described with reference to FIGS.

FIG. 17 illustrates that a short spare line (local spare line, common spare line, or global spare line) and a wide spare line (local spare line or global spare line) are available. In the memory block B0 in Fig. 17, there is an error in the same row address as that of the block B1.

In this case, according to the analysis result, the must repair line of block B1 may be repaired using a short spare line as shown in FIG. 18, or may be repaired as a block B0 using a wide spare line as shown in FIG.

 The results of the above analysis can be stored in the PCAM and the NPCAM included in the memory unit 10 of the repair analysis apparatus of the present invention as shown in FIG.

The repair analysis apparatus of FIG. 20 includes a memory unit 10, a single analyzer 20, a signal validity checker 30, and a redundant analyzer 40 do.

The memory unit 10 includes a PCAM and an NPCAM, and provides the single analyzer 20 with pivot error information and non-pivot error information of the PCAM. The pivot error information of the PCAM includes the row / column address and the block address of the pivot error, and the non-pivot error information of the NPCAM includes the address, the descriptor and the PCAM pointer information. Then, the memory unit 10 provides the block validity and the repair information of the PCAM to the signal validity checker 30.

The single analyzer 20 calculates the pace of the PCAM according to the Spare Direction Selection Signal (DSSS), the Raw Length Selection Signal (RLSS), and the Column Length Selection Signal (CLSS) And compares the error information with the non-pivot error information of the NPCAM to provide a non-pivot cover result and a non-pivot address. The single analyzer 20 extracts the row address, the column address, and the block address corresponding to the non-covering error information for comparing with the pivot error information using the address, the descriptor and the PCAM pointer information of the non-pivot error.

The signal validity checker 30 uses the block address and the must repair information of the pivot error of the PCAM to determine whether the spare assigned by the spare direction select signal DSSS, the row length select signal RLSS and the column length select signal CLSS is Check the validity of the spare to see if it can be used on a block-by-block basis.

The redundant analyzer 40 outputs the signal validity checker 30 with the spare allocated by the spare direction selection signal DSSS, the row length selection signal RLSS, and the column length selection signal CLSS, (Unused Spare, US) that is provided by the user to determine whether the uncovered must repair line and the unrepaired non-pivot error can be repaired.

First, the single analyzer 20 will be described in more detail with reference to FIG.

The single analyzer 20 includes a selection unit for selecting row addresses (Pr1 to Pr6) of a pivot error to be repaired in the PCAM and a column address in consideration of the direction of the spare, a selection unit for selecting the PCAM Compare the pivot error information and the selected length (short or wide) of the row and column to see whether the non-pivot error and the pivot error can be repaired by a short corresponding to one memory block or a wide spare line corresponding to two or more memory blocks And an analysis unit for analyzing the comparison result and providing non-pivot cover result. The single analyzer 20 can provide the redundant analyzer 40 with the address of the non-pivot error included in the non-pivot error information used in the comparison unit, such as the non-pivot cover information of the analysis unit.

First, the selection unit selects a pivot error to be repaired. The selection of the pivot error and the direction of the spare line for repairing the pivot error can be selected by a spare direction selection signal (DSSS).

The selection unit includes a column part selection unit MUXC and a low part selection unit MUXR. All column addresses of each line of the PCAM are provided in the column section selection section MUXC and all row addresses of each line of the PCAM are provided in the row selection section MUXR.

The column part selection part MUXC selects the column address of the PCAM pivot error corresponding to the spare direction selection signal DSSS among all the column addresses of the pivot errors of the PCAM, and the low part selection part MUXR selects the pivot error of the PCAM The row address of the pivot error of the PCAM corresponding to the spare direction selection signal DSSS is selected from among all the row addresses of the spare row direction selection signal DSSS. The spare direction selection signal DSSS is provided in the same bit as the number of lines of the PCAM, that is, the number of spare lines (rows and columns) formed corresponding to the memory blocks B00, B01, B10 and B11, Select the column.

That is, if there are five spare lines and the PCAM has five lines corresponding to the pivot error, the spare direction selection signal DSSS is provided in 5 bits. For example, if the spare direction select signal DSSS is "11000 ", it means that the first and second pivot errors of the PCAM are repaired to the low spare and the remaining pivot errors are repaired to the column spare. To this end, the low-part selection unit MUXR provides a row address corresponding to the first and second pivot errors, and the column part selection unit MUXC provides the column address corresponding to the remaining pivot errors.

The comparator includes comparators (CAC) and low-part comparators (RAC) of the column part. Each of the comparators (CAC) of the column part can understand the detailed configuration with reference to FIG. 22, and each of the low-part comparators (RAC) can understand the detailed configuration with reference to FIG. The number of comparators (CAC) for one non-pivot error of the column equals the number of column spare lines, and the number of comparators (RAC) for one non-pivot error of the row equals the number of low spare lines. The comparators CAC and RAC in FIGS. 22 and 23 are designed to include four memory blocks BOO, B01, B10 and B11 as shown in FIG. 1, and the design is changed when the number of memory blocks is increased or decreased .

The comparator (CAC) and the comparator (RAC) compare the first case where the pivot error and the non-pivot error have the same address (column or row) of the same memory block and the first case where the pivot error and the non- (Column or row) in the same block line or adjacent memory blocks), and the first case is determined to be repairable regardless of the row / column length select signals RLSS and CLSS , And the second case is determined to be repairable when the row / column length select signals RLSS and CLSS are "1" (wide).

More specifically, the comparators CAC and RAC include three XOR gates, one NOR gate, and one AND gate. In the case of a comparator (CAC), the first XOR gate is for comparing the address of the pivot error and the non-pivot error, and the second XOR gate is for determining whether the pivot error and the non- And the third XOR gate is for checking whether the pivot error and the pivot error are in the same memory block if the pivot error and the non-pivot error are in the same block line in the row or column direction .

An example will be described with reference to FIG. In the display of the memory block, the first "0" of "00" represents the row block address B [1] and the second "0" represents the column block address B [0]. Where RRX is the row address of the pivot error, NPry is the row address of the non-pivot error, RRX B [1] is the row block address of the pivot error, NPry B [ [0] is the column block address of the pivot error, and NPry B [0] is the column block address of the non-pivot error.

When the pimetal error and the pivot error have the same memory block and row address of the memory block B00, the first XOR gate outputs "1" because the row address RRX of the pivot error is equal to the row address NPry of the non-pivot error, The XOR gate outputs "1" because the column block address of the pivot error is equal to the column block address of the non-pivot error. The third XOR gate outputs "1" because the row block address of the pivot error is the same as the column block address of the non- " In this case, regardless of the row length selection signal RLSS, the AND gate inputs all "1", and the AND gate outputs the comparison result as "1", ie, repairable.

When the pivot error and the non-pivot error have the same row address of the memory block B00 and the memory block B01, the first XOR gate outputs "1" because the row address RRX of the pivot error is equal to the row address NPry of the non- The second XOR gate outputs "1" because the column block address of the pivot error is equal to the column block address of the non-pivot error, and the third XOR gate has the column block address of the pivot error different from the column block address of the pivot error, 0 " In this case, the output of the gate is determined by the row length selection signal RLSS. That is, when the row length selection signal RLSS is provided as "1" (wide) in order to repair the same row address of adjacent memory blocks, the AND gate inputs all "1" 1 ", i.e., repairable.

22 may be similarly described, and a duplicate description thereof will be omitted.

As described above, the comparators (RAC, CAC) compares the pivot error with the non-pivot error using the block address, and when there is a pivot error and a non-pivot error at the same address of the adjacent memory block, Pivot errors and you can repair pivot errors.

The single analyzer 20 has a function of comparing the results of the comparison of the comparators RAC and CAC with respect to the respective non-pivot errors NPrx and NPcx, And an OR gate (ARC) for outputting the comparison result to non-pivot cover results NPC0 to NP2.

More specifically, the comparators (RAC, CAC) described above can compare repairable states against all of the selected pivot errors in the same number as the spare line in the non-pivot error (NPRx). The comparison results of the comparators (RAC, CAC) for each non-pivot error (NPrx) are input to one of the gates (AR, AC).

Thereafter, the results of the comparison of the row and column of the same non-pivot error are output to the non-pivot cover results NPC0 to NP2 via the gate (ARC).

The redundant analyzer 40 may perform a redundant analysis on a non-pivot address for which the non-pivot cover result provided in the analyzer 20 is not provided as a repairable ("1").

A detailed operation of the repair analysis apparatus of FIG. 20 will be described in detail with reference to FIG. FIG. 24 illustrates the results obtained by a total of eight repair attempts.

A spare direction selection signal DSSS having the same number of bits as the number of lines of the pivot error is applied to the selection unit including the column part selection part MUXC and the low part selection part MUXR, ("1") can be obtained while changing the spare direction selection signal DSSS.

FIG. 24 illustrates the case of two memory blocks included in the same row block line as shown in FIGS. 5 to 15, and the block address of the memory block is divided into [0] and [1]. Then, the spare direction selection signal DSSS is set to have 5 bits, and the row length selection preference (RLSS) can be set to 2 bits corresponding to having two row spacers.

The operation of the embodiment of the repair analysis apparatus of the present invention shown in Figs. 20 to 23 will be described with reference to # 4-1 and # 4-2 of Fig. The Spare Direction Select signal (DSSS) applied to # 4-1 and # 4-2 is defined as "11000" and means that the first and second pivot errors are repaired as low spare and the remaining errors are repaired as column spare.

Accordingly, the pivot error column address defined by "Parent Column cover Address & block" is selected as 5 [1] for RC1 and 4 [0] for Rc2 corresponding to the bit of "1" Quot; 1 " is selected for PR1 corresponding to the bit of "0 ", and 4 [0] is selected for PR2.

In response to the pivot error, the non-pivot error column addresses 3 [0] and 4 [1] are selected, and the non-pivot error row addresses 1 [0] and 4 [0] are selected. The last spare direction selection signal DSSS is a spare that is left as an invalid spare allocation.

# 4-1 assigns a short spare line to the row address of the first line of the non-pivot error, and # 4-2 assigns a wide spare line to the row address of the first line of the non-pivot error.

In the above steps # 4-1 and # 4-2, the non-pivot cover results (NPC) can be obtained as shown in FIG. 24 by the operation of the embodiment of FIG.

# 4-1 can not repair a low-spare line and an adjacent repair line.

In contrast, # 4-2 assigns an invalid assigned non-pivot error to the fourth non-pivot error that has not been repaired. The non-pivot cover results (NPC) of # 4-2 are passed to the redundant analyzer 40 as the corresponding non-pivot address.

The signal validity checker 30 uses the block address and the must repair information of the pivot error of the PCAM to determine whether the spare assigned by the spare direction select signal DSSS, the row length select signal RLSS and the column length select signal CLSS is Make sure that it is valid as a block-specific spare. The signal validity checker 30 provides the redundant analyzer 40 with the spare information US, the unburned must repair pivot error information UMPC, and the signal validity flag as a result of the checking.

The redundant analyzer 40 outputs a state in which the signal validity is secured, that is, the spare assigned by the spare direction select signal DSSS, the row length select signal RLSS, and the column length select signal CLSS, , It is judged whether or not the spare spare information (US) provided in the signal validity checker (30) and the uncovered must repaired pivot error information (INPC) can be used to repair a non-repair pivoted error And outputs the final result (FS).

As a result, # 4-2 can be repaired using the local spare that has been allocated to the fourth non-pivot error, and the final result is determined to be repairable ("0").

Therefore, as described above, the present invention can obtain optimum repair rates for memory cell errors of memory devices using spares having various structures as shown in FIG.

The present invention stores error information of a memory device with a small memory capacity because error information other than a must repair is stored in the PCAM and the NPCAM, analyzes the stored error information, and stores a spare shared in the adjacent memory block, The memory cell error of the memory cell can be repaired at a time.

Then, the present invention can obtain an optimum repair result for eliminating a memory cell error of a memory device of a multi-block structure having two or more memory blocks by obtaining the final result for repair using the block address.

Although the present invention has been described with reference to the embodiments shown in the drawings, it is to be understood that various modifications and equivalent embodiments may be made by those skilled in the art without departing from the scope of the present invention. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

Claims (10)

A first memory for storing addresses and block addresses of pivot errors, and a second memory for storing one of columns and rows representing the address of each of the non-pivot errors and a block address of the non-pivot errors, Defining the remaining one representing the address of each of the errors using the address of the pivot error at the specified location, and comparing the pivot error information including the address of the pivot errors and the block address and the address and block address of the non- A memory unit for providing non-pivoted error information; And
Selecting a column address or row address for each of the pivot errors in the pivot error information to determine a direction of a spare to repair errors of the memory blocks, comparing the non-pivot error information with the pivot error information of the selected pivot errors A first case in which the non-pivot error and the selected pivot error have the same address of the same memory block, and a second case in which the non-pivot error and the selected pivot error are in another memory block arranged in one direction and having the same address And generating a comparison result for determining that the second case is repairable when the spare having a length corresponding to all of the other memory blocks arranged in the one direction is assigned, The non-pivot error And a single analyzer for providing a comparison result as a non-pivot cover result. The method according to claim 1,
Wherein the pivot error information further comprises < RTI ID = 0.0 >
Further comprising a signal validity checker for verifying whether the spare allocated to the memory block using the block address and the length of the pivot error, Device. 3. The method of claim 2,
Further comprising a redundant analyzer,
The signal validity checker provides a signal validity flag, remaining spare information, and unburned must repair pivot error information to verify that the spare is valid,
Wherein the redundant analyzer determines whether the spare pivot error can be repaired by using the remaining spare information and the unburned body repair pivot error information in a state in which the spare validity flag determines that the spare is valid, Analysis device. The method of claim 3,
Wherein the single analyzer provides the address of the non-pivot error and the corresponding non-pivot cover result to the redundant analyzer. The method according to claim 1,
The single analyzer comprises:
A selector for selecting a column address or a row address for each of the pivot errors in the pivot error information to determine a direction of the spare;
Comparing the non-pivot error information with the pivot error information of the selected pivot errors so that the non-pivot error and the selected pivot error have the same address of the same memory block, The first case is determined to be repairable, and the second case is determined to be the other case in which the pivot error is detected in the other memory disposed in the one direction, A comparison unit for generating the comparison result for determining that repair is possible if the spare having a length corresponding to all of the blocks is allocated; And
Repair analysis apparatus including; analyzing unit for providing the non pivoting cover information by analyzing the comparison result. 6. The method of claim 5,
Wherein the comparison unit receives a length selection signal for selecting the length of the spare and generates the comparison result comparing the block address of the pivot error, the block address of the non-pivot error, and the length selection signal, A repair analysis that provides the comparison result as to whether the non-pivot error and the pivot error are repairable to a spare having a length corresponding to a shot corresponding to one memory block or a width corresponding to two or more memory blocks corresponding to a length select signal. Device. The method according to claim 6,
Wherein,
A first gate comparing the address of the pivot error with the address of the non-pivot error;
A second gate for checking whether the pivot error and the non-pivot error are in the adjacent memory block;
A third gate for checking if the pivot error and the non-pivot error are in the same memory block if the memory block is adjacent to the third memory; And
A fourth gate for combining the check result of the third gate with the length selection signal; And
And a fifth gate that combines the outputs of the first, second, and fourth gates with NAND and outputs the result as the comparison result. Storing an address and a block address of pivot errors in a first memory;
Storing one of a column and a row representing the address of each of the non-pivot errors and a block address of the non-pivot errors in a second memory;
Selecting a column address and a row address of pivot error information including an address of the pivot errors and a block address to define a spare direction;
Comparing the non-pivot error information with the pivot error information of the selected pivot errors to determine a first case where the non-pivot error and the selected pivot error have the same address of the same memory block, The first case is determined to be repairable, and the second case is determined to be a length corresponding to all of the other memory blocks arranged in the one direction Generating a comparison result that determines that repair is possible if the spare having the spare is allocated; And
And providing a comparison result of the non-pivot error information as a non-pivot cover result. 9. The method of claim 8,
Checking whether the spare assigned by using the block address and the must repair information of the pivot error can be used for each memory block, validating the spare and verifying whether the spare is valid, remaining spare information, Providing the must repair pivot error information; And
Determining whether the unspoiled non-repairable pivot error can be repaired using the remaining spare information and the unburned body repair pivot error information in a state in which it is determined that the spare is valid according to the signal validity flag Repair method. 9. The method of claim 8,
Wherein the step of generating the comparison result includes determining whether the non-pivot error and the pivot error are shorter than a length corresponding to a short of the memory block or a length of a wide corresponding to two or more memory blocks, And provides the comparison result.

Images