TWI467593B - Marking method and initializing method used for a non-volatile memory array - Google Patents

Description

Marking method and initialization method for a non-volatile memory array

The invention relates to a marking method and an initialization method, in particular to a marking method and an initialization method for a non-volatile memory.

Generally, a non-volatile memory array is composed of a plurality of blocks, each of which contains a plurality of pages, each of which contains a plurality of pages. Byte (byte).

A conventional memory array is a unit in which a page is written as a data, and a block is a data erased unit. That is to say, each page of a block can write a piece of data separately, but if one of the pages of the page is to be erased, the other pages of the block need to be erased together. Therefore, when a certain page of a block is abnormal (cannot be accessed normally), the block cannot be erased, and the block should not be used to avoid problems such as data reading errors or disappearance.

Usually, after a memory array is fabricated at the manufacturing end (for example, a fab), a test device is used to test or scan the memory array when a partial page or byte abnormality in a block is found. The block will be written to a bad mark. After the test, if the normal block ratio is higher than the customer's requirements, for example, the customer defines the good product as the normal block ratio is more than 90%, and the normal block of the memory array is as high as 95%, the manufacturing end can It is listed as a good product, and the memory array is shipped to the application end (or other user side). The application then connects the memory array to a controller (or other firmware). The controller will scan and read which blocks have damage flags to determine which blocks should not be used. This action is called initialization.

However, "Write a bad flag to an abnormal block" has at least the following missing:

1. The conventional detection method is that if a partial page or a byte is damaged, the entire block is judged to be abnormal, and a damage flag is written thereto. Since the block as a whole is judged to be abnormal and discarded, other pages or bytes that can be accessed normally in the block cannot be used. As a result, the memory array may not meet the standard of good products; even if the memory array is still judged to be in compliance with the standard, the usage rate will be low because the other pages or bytes that can be accessed normally are discarded.

2. When the block is judged to be abnormal and the damage mark needs to be written, the write operation may be invalid because the byte of the damaged mark is written, which happens to be an abnormal byte. Therefore, when the controller first accesses the block that is determined to be abnormal, because the damage flag cannot be read, the block is mistakenly judged to be normal.

3. If the data loss of the byte of the damaged mark is stored, the damage mark disappears, which may cause the controller to misjudge the original abnormality because it cannot read the damaged mark. The block is normal.

4. If the manufacturing end is not rigorous or error-prone during testing, there are some abnormal bytes in the block, but it is still misidentified as normal access, so no damage mark is written. The controller will also use the area that was misjudged as normal because the damage flag is not read.

5. The cross talk effect may occur between the byte groups. When the damage flag is written into the block determined to be abnormal, it also affects other byte groups. If the byte group happens to be The pre-marked position of other blocks may cause the controller to misjudge the normal block as an exception.

In view of this, it is an urgent problem to be solved in the industry to provide a method for improving at least one of the above-mentioned deficiencies.

The main object of the present invention is to provide a marking method and an initializing method for a non-volatile memory array, which may be improved when testing and marking the byte access status of the memory unit (or data writing unit). Misjudgment.

Another object of the present invention is to provide a marking method and an initialization method for a non-volatile memory array. By reducing the detection unit to the judgment and marking for each page, it is not caused by the damage of some of the bytes. Discard the entire block and increase the memory module usage.

It is still another object of the present invention to provide a marking method and an initialization method for a non-volatile memory array, which are respectively written with different corresponding marks for normal paging. When the controller initializes it, a more accurate comparison result can be obtained, and the occurrence of false positives can be reduced.

To achieve the above objective, the present invention discloses a marking method for a non-volatile memory, the non-volatile memory array comprising a plurality of memory cells, each of the memory cells comprising a plurality of byte groups, the marking method comprising the following Step: determining that one of the memory cells has one of the groups of bits that can be accessed normally; and writing a flag to the memory unit.

The present invention further discloses an initialization method for a non-volatile memory array, the non-volatile memory array comprising a plurality of memory cells, each of the memory cells comprising a plurality of bytes, wherein each of the memory cells has a mark Corresponding to the fact that the bits in the memory unit are determined to be normally accessible for writing, the initialization method includes the steps of: reading the mark of each memory unit; determining that the mark conforms to a preset a value; recording each of the memory cells that match the preset value into an identification table; and erasing the markers.

The above objects, technical features and advantages will be more apparent from the following description.

Please refer to FIG. 1 for a preferred embodiment of the marking method for a non-volatile memory array of the present invention. The volatile memory array applied by the marking method includes a plurality of memory units (or data writing units), and each memory unit includes a plurality of bytes.

Please refer to FIG. 2 for a non-volatile memory array 1 as an example. The non-volatile memory array 1 includes a plurality of memory cells 111, and each memory cell 111 includes a plurality of byte groups. For example, each memory unit 111 of this embodiment may be a page, each page contains 4224 bytes, and 128 pages (P ₁ ~P ₁₂₈ ) may be combined and defined as a block. 11, and 4096 blocks (B ₁ ~ B ₄₀₉₆₎ is further formed in the non-volatile memory array 1. In other words, the non-volatile memory array 1 contains a total of 4096 x 128 = 524288 memory cells 111 (pages).

Referring to FIG. 3, the non-volatile memory array 1 can be coupled to a test device 2. The steps of the marking method can be implemented by the testing device 2, but are not limited thereto. Each step will be described in detail as follows: First, step S101 is executed to test the non-volatile memory array 1 and determine whether one of the memory cells 111 has normal access to the bit groups. In detail, a data is first written into the byte of the memory unit 111, and then the data is read, and if the read data content has not changed or disappeared, the memory unit 111 The byte can be judged to be accessible normally. In other words, if the data cannot be written or the content is changed or disappeared after the data is written, it can be determined that the bits of the memory unit 111 cannot be accessed normally.

If it is determined in step S101 that normal access is possible, step S103 is executed to write a flag to the memory unit 111. In detail, a preset value (or preset content) is written to some or all of the bytes of the memory unit 111.

If it is determined in step S101 that the access is not normally performed, or when step S103 is performed, step S105 is performed to determine whether there are still other memory units 111 that have not performed step S101; if so, the memory unit 111 is repeatedly executed. The determining and writing steps of step S101 and step S103 are completed until all memory units 111 have been completed; if not, indicating that all memory units 111 have performed steps S101 and S103, the marking method can be completed.

The test device 2 can count the number of memory units 111 that can be normally used according to the above test results, thereby calculating the usable rate of the non-volatile memory array 1, and evaluating whether it can meet the standard. In addition, a lookup table 21 can also be generated or outputted, and the lookup table 21 can record the addresses of the memory cells 111 having the tags, and the preset values corresponding to the tags.

It is worth mentioning that the preset values of the markers on all the memory units 111 may be identical to each other or different from each other. In the preferred embodiment of the present invention, the markers are distinguished from one another by the memory unit 111. In other words, in step S103, different memory units 111 are written with different preset values. The labels that are different from each other may correspond to the address of the memory unit 111. For example, the label preset value of the 100th memory unit of the first block 11 may be 0001100, and the label of the 15th memory unit of the third block The preset value can be 0003015. In this way, each memory unit 111 has a unique mark. If a subsequent check is required, the memory unit of an address can be requested to be read, and whether the data is read or not. Consistent with the address. Alternatively, the preset value of the mark may be correspondingly numbered corresponding to the memory unit 111, so the mark preset value of the 100th memory unit of the first block may be 000100, and the 15th memory unit 111 of the third block 11 The tag preset can be 000271, and so on. It should be noted that the above method for distinguishing the mark is merely an example, and those who have ordinary knowledge in the art should be able to easily think about other marks, and are not limited herein.

Referring to FIG. 4, after performing the foregoing marking method, in the non-volatile memory array 1, it is determined that the memory unit 111 that can be accessed normally has a flag, and the memory unit 111 that determines that the memory cannot be accessed normally will not. With a tag. For convenience of the following description, the normally accessible memory units 111 are additionally numbered 111A, and the memory units 111 that are not normally accessible are otherwise numbered 111B.

The labeled non-playable memory array 1 can then be implemented by the initialization method for a non-volatile memory array of the present invention to recheck the memory cells 111A with the mark to confirm the memory cells 111A. Is there any loss of data loss, crosstalk effect, and access status judgment errors as described in the prior art.

Please refer to FIG. 5, which illustrates a preferred embodiment of a method for initializing a non-volatile memory array of the present invention. The volatile memory array to which the initialization method is applied is exemplified by the non-volatile memory array 1 shown in the foregoing embodiment and FIG.

Referring to FIG. 6, the non-volatile memory array 1 can be coupled to a controller 3. The steps of the initialization method can be performed by the controller 3, but are not limited thereto. In addition, the controller 3 can read the lookup table 21 generated by the test device 2 to obtain information such as the address of the memory unit 111A and the preset value of the mark. The steps of the initialization method will be described in detail as follows: First, step S501 is executed to read the flag of the memory unit 111A. Then, S503 is executed to determine whether the flag meets a preset value. The preset value record may be pre-stored in the lookup table 21 or determined by the controller 3. In addition, each memory unit 111A may have a common mark preset value or a mark preset value that is unique and distinct from each other.

If the read content of the mark does not meet the corresponding preset value, it indicates that the memory unit 111A still has problems such as data loss or difficulty in writing data, and is not suitable for use. In other words, the memory unit 111A is erroneously judged to be normally accessible, but is actually abnormal and should not be used. It should be noted that if the memory unit 111A does not read any mark, indicating that the mark disappears, the memory unit 111A is also regarded as a mark that does not conform to the preset value. That is to say, only when the read content of the mark meets the corresponding preset value, it means that the memory unit 111A should be normally accessible and can be used. The above technical means is because the technical means of the present invention marks a normal memory unit.

When the flag of the memory unit 111A is judged to be in accordance with the preset value, then step S505 is performed to record the memory unit 111A to an identification table 31, that is, the address of the memory unit 111A is recorded into the identification table 31. The identification table 31 can be disposed in the controller 3 or in other firmware coupled to the controller 3. After the end of step S505, step S507 is followed to erase the flag of the memory unit 111A, so that all the data of the memory unit 111A can be cleared.

When the flag determination of the memory unit 111A in step S503 does not meet the preset value, or when step S507 is performed, step S509 will next determine whether there are other memory units 111A not executed; if yes, repeat it. The above steps are performed to determine whether the read content of the mark of the other memory unit 111A meets the corresponding preset value of the memory unit 111A; if not, it indicates that all the memory units 111A have been executed, and the initialization method can be completed.

After the initial method is completed, the address information of the memory unit 111A whose mark matches the preset value is recorded in the identification table 31. The memory unit 111 (including the unmarked memory unit 111B and the memory unit 111A whose mark does not match the preset value) which is not recorded in the identification table 31 will not be used by the controller 3 (or other firmware).

Please refer to FIG. 7, which is a second preferred embodiment of the initialization method for a non-volatile memory array of the present invention. The difference between the second embodiment and the initialization method of the first embodiment is that a step S511 is added, so that other similar parts will not be described again.

When the step S511 is performed, the addresses of all the memory units 111 (including the unmarked memory unit 111B, the memory unit 111A marked with the preset value, and the memory unit 111A marked with the preset value) will be redefined. That is, the normal memory unit 111A recorded in the identification table 31 can redefine the block, and the unmarked memory unit 111B and the memory unit 111A whose mark does not conform to the preset value are discarded or are defined in the same Block 11 will not be used. In other words, all of the usable blocks 11 are formed by the definition of the memory unit 111A whose mark conforms to the preset value.

Thus, the unmarked memory unit 111B and the memory unit 111A whose marks do not conform to the preset value can be concentrated in a few blocks 11, thereby reducing the number of blocks 11 that cannot be used.

In addition, in the first or second embodiment of the above initialization method, step S507 may be changed to be executed after step S509, and all of the flags are eliminated once.

In summary, the marking method and the initialization method for a non-volatile memory array of the present invention have many features, which are listed below:

1. Using the paging as the unit of detection, the usage rate can be improved by using the block as a unit compared with the conventional technology.

2. The memory unit that is judged to be normally accessible can be written to the mark smoothly, and the problem of data loss is not easy. Therefore, the present invention marks the memory unit that can be normally accessed, and the conventional mark is difficult to write and mark. The problem of disappearance can be improved.

3. If there is an error or imprecision in the marking process, some memory cells that are abnormal (cannot be accessed normally) are judged to be normal access. However, an abnormal memory unit is more difficult to write a mark, and even if the mark is written, the abnormal memory unit is liable to have a data loss problem, so the mark easily disappears or changes. After initialization, the abnormal memory unit that is misjudged as being normally accessible will not be read to the mark or read to a mark that does not match the preset value. Thus, the misjudged abnormal memory unit can be found without being used.

4. When the preset values of the markers are different from each other, the loss due to the crosstalk effect can be improved. In detail, if the mark is written to a memory unit that can be accessed normally, the crosstalk effect is also erroneously written to a certain memory unit, but since the written content of the mark is based on the memory unit corresponding to the normal access The preset value, after initialization, the miswritten memory unit, even if there is a mark, will not be used because the read content of the mark does not conform to the corresponding preset value of the erroneous memory unit.

5. Redefine the composition of the block according to the identification table, discard the memory unit that cannot be used, and redefine the block with the usable memory unit, thus improving the usage rate of the non-volatile memory (the block that can be used) The number is relative to the number of all blocks).

The embodiments described above are only intended to illustrate the embodiments of the present invention, and to explain the technical features of the present invention, and are not intended to limit the scope of protection of the present invention. Any changes or equivalents that can be easily made by those skilled in the art are within the scope of the invention. The scope of the invention should be determined by the scope of the claims.

1. . . Non-volatile memory

11. . . Block

111, 111A, 111B. . . Memory unit

2. . . Test device

twenty one. . . Query list

3. . . Controller

31. . . Identification table

Figure 1 is a flow chart of a method of marking a non-volatile memory array of the present invention.

Figure 2 is a schematic illustration of a non-volatile memory array of the present invention.

Figure 3 is a schematic diagram of the non-volatile memory array coupled to the test device.

Figure 4 is a schematic diagram of the non-volatile memory array after labeling.

Figure 5 is a flow chart of a method for initializing a non-volatile memory array of the present invention.

Figure 6 is a schematic diagram of the non-volatile memory array coupled to a controller.

Figure 7 is a flow chart of a method of another preferred embodiment of the method for initializing a non-volatile memory array of the present invention.

Claims (12)

A marking method for a non-volatile memory array, the non-volatile memory array comprising a plurality of memory cells, each of the memory cells comprising a plurality of bytes, the marking method comprising the steps of: determining such One of the memory cells has one of the groups of bits that can be accessed normally; and a flag is written to the memory unit. The marking method as claimed in claim 1, further comprising performing the determining step and the writing step repeatedly for the other memory units. The marking method of claim 2, wherein the markings in the memory units are different from one another. The marking method of claim 3, wherein each of the markings corresponds to an address of each of the memory units. The marking method of claim 3, wherein the markings are numbered sequentially. The marking method of claim 2, wherein the markings are identical to each other. The marking method of claim 2, wherein each of the memory units is a page, and a plurality of pages can be defined to form a block. An initialization method for a non-volatile memory array, the non-volatile memory array comprising a plurality of memory cells, each of the memory cells comprising a plurality of bytes, wherein each of the memory cells has a mark The method is characterized in that the bit groups in the memory unit are determined to be normally accessible, and the initialization method includes the steps of: reading the mark of each memory unit; determining that the mark meets a preset value. Recording each of the memory cells that match the preset value into an identification table; and erasing the markers. The initialization method of claim 8, wherein the non-volatile memory array is connected to a controller. The initialization method of claim 8, wherein the recording step records one address of each of the memory units. The initialization method of claim 8, wherein the markers are different from each other. The initialization method of claim 8, further comprising the step of forming a plurality of definitions of the memory cells recorded in the identification table in a block.