Multi-bit Memory Device Adaptive Decision Value Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-level cell memory devices face challenges in reducing data reading time and error rates due to fine electrical characteristics and limited voltage windows, leading to overlapping distributions and increased read-failure rates as the number of bits stored in a single cell increases.
Innovation Solution
A memory device with a multi-bit cell array, a decision unit, an error detector, and a determination unit that detects threshold voltages, decides data using first and second decision values, and adjusts based on error bits to minimize errors and optimize data reading, employing ECC decoding and considering mechanisms like floating gate coupling and charge loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multi-level cell memory devices store more bits in a single cell, then storage density increases, but read-failure rate increases due to overlapping voltage distributions
Solution Approach 1:
The patent segments the data reading process into multiple stages: first reading data from cells with fewer error bits using initial decision values, then reading data from cells with more error bits using adjusted decision values. This segmentation allows the system to handle different error scenarios separately, improving overall reliability while maintaining high storage density.
Solution Approach 2:
The patent dynamically adjusts decision values based on the number of error bits detected in previously read data. When error bits exceed a threshold, the system modifies decision values for subsequent readings. This dynamic adaptation enables the system to optimize reading accuracy in real-time, resolving the contradiction between high storage density and read reliability.
2Reliability
If multiple threshold voltage detections are performed to reduce error bits, then error correctability improves, but data reading time increases
Solution Approach 1:
The patent applies partial action by performing multiple threshold voltage detections only when necessary - specifically, when error bits in the first read data exceed a predetermined threshold. For cells with acceptable error rates, the system uses only the initial reading, avoiding unnecessary additional detections. This selective approach improves error correctability when needed while minimizing impact on reading time.
Solution Approach 2:
The system uses feedback from the error detection stage to control subsequent reading operations. When error bits exceed the threshold, the system triggers additional threshold voltage detections with adjusted decision values. This feedback mechanism ensures that multiple detections are performed only when actually needed, balancing error correctability with reading time efficiency.
3Reliability
If adjusted decision values are used for all cells, then error bits are reduced, but hardware complexity and processing overhead increase
Solution Approach 1:
The patent applies local quality by adjusting decision values only for specific memory cells that exhibit high error rates, rather than uniformly adjusting decision values for all cells. The system identifies cells with error bits exceeding the threshold and applies modified decision values only to those cells in subsequent readings. This localized approach reduces error bits while minimizing the increase in hardware complexity and processing overhead.
Data Source
AI summary
Provided are memory devices and memory data read methods. A method device may include: a multi-bit cell array; a decision unit that may detect threshold voltages of multi-bit cells of the multi-bit cell array to decide first data from the detected threshold voltages, using a first decision value; an error detector that may detect an error bit of the first data; and a determination unit that may determine whether the decision unit decides second data from the detected threshold voltages using a second decision value, based on a number of detected error bits, the second decision value being different from the first decision value. Through this, it is possible to reduce time spent for reading data stored in the multi-bit cell.