Flash Memory Read Threshold Adjustment for Aging Cells
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Nonvolatile memory systems face challenges in accurately reading data due to errors caused by voltage shifts and physical defects, leading to incorrect bit readings and reduced data reliability, especially as the memory ages, where traditional Error Correction Codes (ECCs) become less effective with increasing errors.
Innovation Solution
A flash memory system that maps memory states to adjustable threshold windows, using a reading circuit to compare threshold voltages to predetermined voltages, and an ECC decoder to adjust these voltages based on decoding information, allowing for more precise data reading and correction by employing Soft-Input Soft-Output (SISO) decoding and adjusting discrimination voltages to balance error corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional fixed threshold voltage reading is used, then the reading process is simple, but data reading accuracy deteriorates as memory ages and usage increases due to voltage shifts and physical defects
Solution Approach 1:
The patent implements dynamic threshold voltage adjustment by mapping memory states to adjustable threshold windows instead of using fixed thresholds. The reading circuit compares threshold voltages to predetermined voltages within these windows, and the ECC decoder adjusts the threshold voltages based on decoding information from previous reads. This dynamic adaptation allows the system to maintain high reading accuracy despite memory aging, voltage shifts, and physical defects.
Solution Approach 2:
The patent employs feedback mechanisms where the ECC decoder uses decoding information from previously read data to adjust the threshold voltages for subsequent reads. This feedback loop enables the system to learn from past reading errors and corrections, continuously optimizing the threshold voltage mapping to compensate for memory degradation and maintain accurate data retrieval over time.
2Reliability
If traditional Error Correction Codes are used, then the system structure is simple, but error correction capability is insufficient when likelihood distributions widen due to memory aging
Solution Approach 1:
The patent transforms the static ECC system into a dynamic one by implementing adjustable threshold windows that adapt to memory aging. Instead of using fixed threshold voltages, the system dynamically remaps memory states to new threshold windows based on observed error patterns and likelihood distribution changes. This dynamic remapping extends the effective error correction capability beyond what traditional fixed-threshold ECC can achieve.
Solution Approach 2:
The patent changes the operating parameters of the ECC system by adjusting threshold voltages and window mappings based on memory state analysis. When likelihood distributions widen due to aging, the system modifies the threshold voltage parameters and remaps states to maintain adequate separation between logical states, thereby preserving error correction effectiveness despite increased noise and variability.
3Reliability
If fixed threshold windows are used throughout memory life, then the system operation is simple, but data quality deteriorates as memory ages and usage increases
Solution Approach 1:
The patent implements dynamic threshold window adjustment that adapts to memory aging and usage patterns. Instead of maintaining fixed threshold windows throughout the memory's life, the system periodically analyzes read data quality and remaps memory states to new threshold windows that compensate for degradation. This dynamic adaptation maintains high data quality while the automated nature of the process minimizes operational complexity.
Solution Approach 2:
The patent performs preliminary analysis of read data quality and proactively adjusts threshold windows before significant degradation occurs. By monitoring error patterns and likelihood distributions during normal operation, the system can remap states to preventive threshold windows that maintain data quality, rather than waiting for performance to deteriorate and then attempting correction.
Data Source
AI summary
Data is read from a nonvolatile memory array using one or more read voltages that are adjusted during memory life. Programming target voltages and read voltages may be adjusted together over memory life to map memory states to an increasingly wide threshold window. Individual memory states are mapped to sub-ranges that are made wider, reducing errors.