Multilevel Memory Coding for Overlapping Threshold Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory devices face challenges in increasing storage density due to intrinsic tolerances that lead to overlapping program distributions, making it difficult to accurately associate threshold voltage ranges with specific programming states, resulting in errors during data retrieval.
Innovation Solution
Implementing a memory device with multi-level memory cells that use convolutional coding and soft decision decoding, along with error correcting codes, to encode and store information in a way that reduces the probability of wrong readings by partitioning program distributions into subsets and using a trellis diagram to determine the most probable encoding path.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multilevel programming scheme is employed to increase storage density, then storage density is improved, but program distributions overlap making accurate retrieval difficult
Solution Approach 1:
The patent divides the continuous threshold voltage range into multiple discrete programming states by segmenting the voltage axis into distinct regions (e.g., first programming state with lowest threshold voltage, second programming state with intermediate threshold voltage, third programming state with highest threshold voltage). This segmentation allows accurate identification of programming states despite overlapping distributions by establishing clear decision boundaries between states.
Solution Approach 2:
The patent changes the parameter of threshold voltage measurement by introducing multiple read voltage levels and comparing the resulting current measurements to determine the programming state. Instead of relying on a single threshold measurement, the system uses multiple voltage points and comparative analysis to accurately identify the programming state, effectively transforming the measurement approach to overcome distribution overlap.
2Quantity of substance
If more programming states are used to increase storage density, then storage density is improved, but errors during data retrieval increase
Solution Approach 1:
The patent implements a feedback mechanism where read operations measure the current at multiple voltage points and compare these measurements against reference values or decision boundaries to determine the programming state. This feedback loop allows the system to accurately identify the intended programming state even when distributions overlap, thereby maintaining high retrieval accuracy despite using multiple programming states per cell.
Solution Approach 2:
The patent performs more measurements than the minimum required by using multiple read voltage levels and multiple current measurements per read operation. This excessive action (more measurements than strictly necessary) provides redundant information that improves the accuracy of programming state identification, allowing the system to tolerate the increased complexity of multiple programming states while maintaining high retrieval reliability.
Data Source
AI summary
Subject matter disclosed herein relates to semiconductor memories and, more particularly, to multilevel non-volatile or volatile memories.