Flash Memory Compensation Using Quality Metrics for Threshold Shift
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As the number of data levels per memory cell increases in flash memory, so does the probability of errors during data readback, due to the need for higher accuracy in resolving between closely positioned data thresholds, leading to distortion mechanisms that can shift data values across thresholds, causing improper read values.
Innovation Solution
A method and apparatus that utilize a memory controller to apply quality metric-based compensation by determining statistical averages, variances, and error rates to correct distortions in data, using pre- and post-compensation techniques, and curve fitting to adjust data values, ensuring accurate data retrieval and storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the number of data levels per memory cell is increased to enhance storage capacity, then storage capacity is improved, but error probability increases due to distortion mechanisms shifting data values across thresholds
Solution Approach 1:
The patent applies preliminary compensation by determining compensation values based on historical quality metrics before actual data read operations. The memory controller pre-calculates compensation values using quality metrics from previous reads and stores these values for subsequent use, thereby proactively counteracting distortion effects before they cause read errors.
Solution Approach 2:
The patent implements a feedback mechanism where quality metrics are continuously monitored during data read operations, and these metrics are used to adjust compensation values. The memory controller uses the observed quality of retrieved data to refine future compensation applications, creating a closed-loop system that adapts to changing distortion conditions.
2Reliability
If error correction codes are used to compensate for increased error probability, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent changes the parameter approach by shifting from correcting errors after they occur to pre-compensating for expected distortion. Instead of using complex error correction codes that operate on binary data, the system applies analog compensation values to adjust data levels before threshold comparison, thereby reducing the burden on error correction mechanisms.
Solution Approach 2:
The patent introduces compensation values as an intermediary element between the stored data and the threshold comparison process. These compensation values act as a mediator that adjusts the effective threshold positions or data levels, thereby reducing the direct impact of distortion without requiring complex error correction logic.
Data Source
AI summary
In one embodiment, an encoder reads a set of data from memory cells to obtain retrieved data influenced by one or more distortion mechanisms as a result of having been stored. A quality metric is generated responsive to the retrieved data that changes in value responsive to differences between the user data and the associated retrieved data. A quality monitor establishes a relationship between a current value of the quality metric and a threshold value and monitors the relationship as being indicative of a degradation of the quality of the retrieved data, and selectively initiates an error response. In another embodiment, a correction value is iterated through a set of values as a quality metric is monitored such that the value of the quality metric which most closely approaches the value of the quality metric immediately subsequent to an initial writing of the data can be selected.