Non-Volatile Memory Cell Grouping for Read Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-volatile memory systems, such as NAND-type flash memory, face errors in data reading due to variations in charge distribution across memory cells, leading to inaccuracies when using a single reference read voltage.
Innovation Solution
The memory system classifies memory cells into groups based on data retention characteristics and employs optimized read voltages for each group using Vth Tracking, merging data read with different operation parameters to minimize errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single reference read voltage is used for reading data from all memory cells in a page, then the reading operation is simple and fast, but reading errors increase due to variations in charge distribution across memory cells
Solution Approach 1:
The patent divides a page of memory cells into multiple sub-pages based on data retention characteristics. Each sub-page is read using a dedicated reference read voltage optimized for its specific charge distribution characteristics. This segmentation allows accurate reading for each sub-page while managing overall system complexity through organized分组 reading operations.
Solution Approach 2:
The patent applies different reference read voltages to different sub-pages according to their specific data retention characteristics. Instead of using a uniform reading approach, each sub-page receives a tailored read voltage that matches its charge distribution profile, thereby optimizing reading accuracy for local conditions while maintaining overall system functionality.
2Measurement precision
If multiple reference read voltages are used for different groups of memory cells, then reading accuracy improves, but the reading time increases due to sequential reading operations
Solution Approach 1:
The patent performs preliminary classification of memory cells into sub-pages based on data retention characteristics before the actual reading operation. Reference read voltages are pre-determined for each sub-page based on their charge distribution characteristics. This preliminary organization enables more efficient reading operations by avoiding dynamic voltage switching during the reading process.
Solution Approach 2:
The patent implements dynamic reading operations where the system adaptively selects and switches between different reference read voltages based on the specific sub-page being read. The reading operation dynamically adjusts which voltage is applied to which sub-page, optimizing the balance between reading accuracy and reading speed by processing sub-pages with similar characteristics using their dedicated voltages.
Data Source
AI summary
According to one embodiment, a controller groups a plurality of memory cells in each of the pages into a plurality of groups. The plurality of groups includes a first group and a second group. In a case of reading data from a first page, The controller performs first reading. The first reading includes reading data from the first page by using a first operation parameter for the first group. The controller performs second reading. The second reading includes reading data from the first page by using a second operation parameter for the second group. The controller merges first read data and second read data, and return the merged data as read data read from the first page. The first read data is acquired by the first reading. The second read data is acquired by the second reading.