Dual sense bin balancing in NAND flash
A controller utilizes dual sense bin balancing (DSBB) to adjust a read level between memory states of an array of NAND flash memory cells. One or more iterations of DSBB may be performed to provide a read level. Each iteration of the DSBB includes performing a first sense read, performing a second sense read, determining a read error, and adjusting the initial read level. The first sense read is performed at a first offset of an initial read level of memory cells. The second sense read is performed at a second offset of the initial read level of memory cells. A read error is determined from the first sense read and the second sense read. The read level is adjusted by the read error. A read of the randomized data pattern is conducted with the adjusted read level of a last iteration of the DSBB.
1. A method of reading a randomized data pattern stored in an array of memory cells, comprising:
performing one or more iterations of a dual sense bin balancing (DSBB) to provide a read level, each iteration of the dual sense bin balancing comprising:
performing a first sense read at a first offset of an initial read level of memory cells to determine a first number of memory cells relative to the first offset, wherein the first offset is at a first nominal bin width from the initial read level;
performing a second sense read at a second offset of the initial read level of memory cells to determine a second number of memory cells relative to the second offset, wherein the second offset is at a second nominal bin width from the initial read level, wherein the second nominal bin width and the first nominal bin width are equal, wherein the second offset is opposite of the first offset from the initial read level, and wherein the first nominal bin width and the second nominal bin width is a quarter of a bin width;
determining a read error of the initial read level from the first sense read and the second sense read; and
adjusting the initial read level by the read error to provide an adjusted read level; and
conducting a read of the randomized data pattern with the read level of the adjusted read level of a last iteration of the one or more iterations of the dual sense bin balancing.
2. The method of claim 1 , wherein the adjusted read level is used as the initial read level in a successive iteration of the DSBB.
3. The method of claim 1 , wherein the adjusted read level is at an unfixed location.
4. The method of claim 1 , wherein the read error of the one or more iterations the dual sense bin balancing forms a read error gradient that is substantially linear.
5. The method of claim 1 , further comprising providing a plurality of read levels by performing the one or more iterations of the dual sense bin balancing for each of the plurality of read levels.
6. The method of claim 1 , wherein the read level is between two distributions of cell voltage thresholds of two memory states in which the two distributions have dissimilar curvatures.
7. The method of claim 1 , wherein the method is performed by a logic circuit of a flash die containing the array of memory cells.
8. The method of claim 1 , further comprising accumulating an accumulated count of the first number of memory cells relative to the first sense read and the second number of memory cells relative to the second sense read.
9. The method of claim 8 , further comprising subtracting an expectation value from the accumulated count to determine the read error.
10. A flash memory die, comprising:
an array of memory cells configured to store a randomized data pattern; and
a logic circuit configured to perform one or more iterations of a dual sense bin balancing (DSBB) to provide a read level, each iteration of the dual sense bin balancing comprising:
performing a first sense read at a first offset of an initial read level of memory cells to determine a first number of memory cells relative to the first offset, wherein the first offset is at a first nominal bin width from the initial read level;
performing a second sense read at a second offset of the initial read level of memory cells to determine a second number of memory cells relative to the second offset, wherein the second offset is at a second nominal bin width from the initial read level, wherein the second nominal bin width and the first nominal bin width are equal, wherein the second offset is opposite of the first offset from the initial read level, and wherein the first nominal bin width and the second nominal bin width is a quarter of a bin width;
determining a read error of the initial read level from the first sense read and the second sense read; and
adjusting the initial read level by the read error to provide an adjusted read level.
11. The flash memory die of claim 10 , wherein the adjusted read level is used as the initial read level in a successive iteration of the DSBB.
12. The flash memory die of claim 10 , wherein the adjusted read level is at an unfixed location.
13. The flash memory die of claim 10 , wherein the read error of the one or more iterations the dual sense bin balancing forms a read error gradient that is substantially linear.
14. The flash memory die of claim 10 , wherein the logic circuit is configured to provide a plurality of read levels by performing the one or more iterations of the dual sense bin balancing for each of the plurality of read levels.
15. The flash memory die of claim 10 , wherein the array of memory cells is selected from a group consisting of single-level cells (SLC), multiple-level cells (MLC), triple-level cells (TLC), quad-level cells (QLC), or five-level cells.
16. The flash memory die of claim 10 , wherein the logic circuit comprises an accumulator to accumulate an accumulated count of the first number of memory cells relative to the first sense read and the second number of memory cells relative to the second sense read.
17. The flash memory die of claim 16 , wherein the logic circuit further comprises an expectation unit storing an expectation value of balanced bins for the read level, wherein the logic circuit is configured to subtract the expectation value from the accumulated count to determine the read error.
18. A method of reading a randomized data pattern stored in an array of memory cells, comprising:
performing a first plurality of iterations to determine a first read level, each iteration comprising:
performing a first sense read at a first offset of an initial first read level of memory cells to determine a first number of memory cells relative to the first offset, wherein the first offset is at a first nominal bin width from the initial read level;
performing a second sense read at a second offset of the initial first read level of memory cells to determine a second number of memory cells relative to the second offset, wherein the second offset is at a second nominal bin width from the initial read level, wherein the second nominal bin width and the first nominal bin width are equal, wherein the second offset is opposite of the first offset from the initial read level, and wherein the first nominal bin width and the second nominal bin width is a quarter of a bin width;
determining a read error of the initial first read level from the first sense read and the second sense read; and
adjusting the initial first read level by the read error to provide an adjusted first read level;
performing a second plurality of iterations to determine a second read level, each iteration comprising:
performing a first sense read at a first offset of an initial second read level of memory cells to determine a first number of memory cells relative to the first offset;
performing a second sense read at a second offset of the initial second read level of memory cells to determine a second number of memory cells relative to the second offset;
determining a read error of the initial second read level from the first sense read and the second sense read; and
adjusting the initial second read level by the read error to provide an adjusted second read level; and
conducting a read of the randomized data pattern with the adjusted first read level of a last iteration of the first plurality of iterations and with the adjusted second read level of a last iteration of the second plurality of iterations.
19. The method of claim 18 , wherein the second plurality of iterations is performed after the first plurality of iterations.
20. The method of claim 18 , wherein the method is performed by a logic circuit of a flash die containing the array of memory cells.