Multi-Level Memory Encoding for Partially Defective PCM Cells
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-level phase change memory (PCM) technology faces challenges in writing multiple resistance levels due to process variability and cell-to-cell variability, leading to limitations in writing all nominal levels in partially-defective cells, which restricts multilevel coding and affects storage capacity.
Innovation Solution
A method and encoder/decoder system that adaptively encode data words as either standard or excess codewords based on cell defectiveness, using q nominal levels or (q-d) levels, allowing for reliable writing and decoding even in partially-defective cells by employing a two-level encoding scheme with translation-stable codes and error correcting codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multi-level cell functionality is implemented to increase storage capacity, then cost per bit is reduced, but process variability and cell-to-cell variability cause deviations from intended resistance levels
Solution Approach 1:
The patent changes the coding parameters by using two different code types: a first code with higher rate for cells without defects, and a second code with lower rate and reduced alphabet size for cells with defects. This parameter adaptation allows the system to accommodate variability while maintaining storage capacity.
Solution Approach 2:
The system dynamically selects between different coding schemes based on the actual state of the memory cells. The encoder adapts the code type according to whether cells can support all q levels or only a subset, making the coding scheme flexible and responsive to cell capabilities.
2Measurement precision
If adaptive iterative write scheme is used to compensate for variability, then reading accuracy improves, but partially-defective cells cannot support all nominal levels
Solution Approach 1:
The patent segments the coding space into two distinct code types: one for cells supporting all q levels and another for cells supporting only (q-d) levels. This segmentation allows partial-defective cells to be utilized with appropriate coding, increasing overall system adaptability.
Solution Approach 2:
Different coding qualities are applied to different cells based on their individual capabilities. Cells without defects receive higher-rate coding, while partially-defective cells receive lower-rate coding with reduced alphabet size, optimizing performance for each cell's local characteristics.
3Productivity
If standard codewords with q symbol values are used, then storage capacity is maximized, but partially-defective cells cannot be reliably written
Solution Approach 1:
The system dynamically adapts the coding scheme based on cell capabilities. When a cell can support all q levels, the higher-rate code is used for maximum capacity. When a cell is partially-defective, the system switches to a lower-rate code with (q-d) symbol values, ensuring reliable writing while maintaining overall productivity.
4Reliability
If excess codewords with (q-d) symbol values are used for partially-defective cells, then writing reliability improves, but storage capacity per cell is reduced
Solution Approach 1:
The patent changes the coding parameters (alphabet size and code rate) based on cell capabilities. For partially-defective cells, the alphabet size is reduced from q to (q-d) symbol values, ensuring reliable writing. Although this reduces capacity per defective cell, it enables the use of previously unusable cells, increasing overall system capacity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The proposed solution increases the storage capacity of solid state memory by enabling the use of partially-defective cells and improves reliability by using excess codewords to write data in cells that cannot support all nominal levels, while maintaining high Euclidean distances for error detection and correction.
Implementation Method 1
PCM is a non-volatile solid state memory technology that exploits a reversible, thermally-assisted switching of certain chalcogenide and non-chalcogenide compounds between certain states of different electrical conductivity. Multiple resistance levels or simply levels correspond to partial-amorphous and partial-crystalline phase distributions of the phase-change material of the PCM cell.
Implementation Method 2
Phase transformation, i.e., memory programming, can be enabled by Joule heating. In this regard, Joule heating can be controlled by a programming current or voltage pulse.
Data Source
AI summary
A method for encoding a data word for writing an encoded data word in N cells of a solid state memory. Each of the N cells can be programmed in one of q nominal levels. The method includes encoding the data word as a codeword of a first codeword type having q symbol values or as a codeword of a second codeword type having (q-d) symbol values, d ε [1, . . . , q−1], depending on a state of the N cells.