Error correction using WOM codes

Patent metadata

Field	Value
Publication number	US-10133626-B2
Application number	US-201615232323-A
Country	US
Kind code	B2
Filing date	Aug 9, 2016
Priority date	Aug 14, 2015
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

A method is proposed for storing bits in memory cells of a memory, wherein in two successive write operations first and second wits are written to identical memory cells at an identical address, without the memory cells being erased after the first write operation, wherein first check bits are stored in further first memory cells and second check bits are stored in further second memory cells. A corresponding device is furthermore specified.

First claim

The invention claimed is:

1. A method for storing bits in memory cells of a memory, comprising:
writing first wits and second wits to identical memory cells at an identical address of the memory via two successive write operations, without the memory cells being erased after a first write operation of the two successive write operations; and
storing first check bits in further first memory cells and second check bits in further second memory cells,
wherein the number of first check bits is less than the number of the first wits, and
wherein the memory cells are physical memory cells susceptible to at least one of transient memory errors or permanent memory errors.

2. The method as claimed in claim 1 ,
wherein a first code word of a first error code is determined on the basis of the first wits and the first check bits,
wherein a second code word of a second error code is determined on the basis of the second wits and the second check bits.

3. The method as claimed in claim 1 , wherein the first wits and the second wits are determined on the basis of useful data bits by means of at least one WOM code.

4. The method as claimed in claim 3 , wherein at least one inversion bit is added to the useful data bits, wherein a subset of useful data bits is assigned to the at least one inversion bit.

5. The method as claimed in claim 4 , wherein the at least one inversion bit assumes a first value if the useful data bits assigned to the inversion bit are not inverted, and the inversion bit assumes a second value different therefrom if the useful data bits assigned to the inversion bit are inverted.

6. The method as claimed in claim 5 , wherein the respective wits are formed using the useful data bits and the inversion bit, wherein the useful data bits assigned to the at least one inversion bit are not inverted if the inversion bit has the first value, and wherein the useful data bits assigned to the at least one inversion bit are inverted if the inversion bit has the second value.

7. The method as claimed in claim 4 , wherein in the first write operation the first check bits are determined such that the first wits determined from the useful data bits and the at least one inversion bit and the first check bits are bits of the first code word, and wherein in the second write operation at the same address the second check bits are determined such that the second wits determined from the useful data bits and the at least one inversion bit and the second check bits are bits of the second code word.

8. The method as claimed in claim 4 , wherein the useful data bits assigned to the at least one inversion bit are all useful data bits.

9. The method as claimed in claim 3 ,
wherein before the write operation, in memory cells to be addressed, the data stored there are read out,
wherein the data read out are compared with the data to be stored,
wherein the data to be stored are inverted if the data to be stored cannot be written to the memory in an error-free manner,
wherein the first wits or the second wits are determined on the basis of the inverted data by means of the at least one WOM code and are stored in the memory cells.

10. The method as claimed in claim 3 , wherein the first wits coded by means of the at least one WOM code are different than the second wits coded by means of the at least one WOM code.

11. The method as claimed in claim 3 , wherein at least one memory cell of the memory is not completely erased before the first writing, wherein the first wits and the second wits are coded by means of the at least one WOM code on the basis of the at least one not completely erased memory cell.

12. The method as claimed in claim 3 , wherein the first wits and the second wits are organized in each case as blocks, wherein two different codings are assigned to each occupancy of the bits of one of the blocks of the first wits by the at least one WOM code.

13. The method as claimed in claim 1 , wherein the number of second check bits is less than the number of second wits.

14. The method as claimed in claim 1 , wherein the first error code is an error detecting code and/or an error correcting code, and wherein the second error code is an error detecting code and/or an error correcting code.

15. The method as claimed in claim 1 , wherein the first error code is a code having a code distance of greater than or equal to three.

16. The method as claimed in claim 1 , wherein the second error code is a code having a code distance of greater than or equal to three.

17. The method as claimed in claim 1 , wherein the further first memory cells and the further second memory cells are different than one another.

18. The method as claimed in claim 1 , wherein if no error occurs, the first wits and the first check bits are a proper subset of bits of the first code word, or the second wits and the second check bits are a proper subset of bits of the second code word, or the first wits and the first check bits are a proper subset of the bits of the first code word and also the second wits and the second check bits are a proper subset of the bits of the second code word.

19. The method as claimed in claim 1 , wherein the length of the first error code is greater than the number of first wits and first check bits, and wherein the length of the second error code is greater than the number of second wits and second check bits.

20. The method as claimed in claim 1 , wherein the first code word of the first error code is determined on the basis of the first wits, the first check bits and first additional bits, and wherein the second code word of the second error code is determined on the basis of the second wits, the second check bits and second additional bits.

21. The method as claimed in claim 20 , wherein when writing to the memory and/or when reading from the memory, the first wits, the first check bits and the first additional bits form the first code word and the second wits, the second check bits and the second additional bits form the second code word.

22. The method as claimed in claim 20 , wherein the first additional bits and/or the second additional bits are based in each case at least on one of the following items of information:
address information,
address information concerning a write address or a read address,
information derived from address information,
identifier information,
password information,
inversion information.

23. The method as claimed in claim 20 , wherein an error in the first additional bits is detected and/or corrected, and/or wherein an error in the second additional bits is detected and/or corrected.

24. The method as claimed in claim 1 , wherein a marking is stored in the first further memory cells provided that the second check bits are stored in the further second memory cells.

25. The method as claimed in claim 24 , wherein the marking is a bit sequence that does not constitute a valid bit sequence of the first check bits.

26. The method as claimed in claim 1 , wherein a partial area of the memory is erased which comprises the address to which a write operation was applied if the first or second wits to be stored at d N positions do not have the desired state after the write operation and/or if the first check bits or the second check bits at d P positions do not have the desired state after the write operation.

27. The method as claimed in claim 1 , wherein a seco

Assignees

• Infineon Technologies Ag

Inventors

• Kern Thomas
• Goessel Michael

Classifications

• G11C17/146

  Write once memory, i.e. allowing changing of memory content by writing additional bits · CPC title

• G11C2029/4402

  Internal storage of test result, quality data, chip identification, repair information · CPC title

• G11C2029/0411

  Online error correction · CPC title

• G11C29/52

  Protection of memory contents; Detection of errors in memory contents · CPC title

• G06F11/1044 Primary

  with specific ECC/EDC distribution · CPC title