Non-Volatile Memory Defect Detection and Data Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-volatile memory devices face challenges in accurately programming data, especially when defects such as broken word lines or weak word line-to-word line shorts are present, leading to read failures and data loss.
Innovation Solution
The implementation of a method that assigns unused data latch combinations to locked out cells, allowing for distinction between different data states after programming, and uses read levels to count mismatches and initiate data recovery when errors exceed a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional programming methods are used without defect detection, then programming speed is maintained, but data accuracy deteriorates due to defects like broken word lines or weak word line-to-word line shorts
Solution Approach 1:
The patent applies preliminary action by performing defect detection and classification before the actual programming operation. The system identifies defective cells and determines their type (stuck-at-0, stuck-at-1, or floating gate defects) in advance, then uses this information to select appropriate programming strategies, thereby preventing data loss without requiring complex real-time defect correction mechanisms
Solution Approach 2:
The patent implements feedback by reading the programmed data back from the memory device and comparing it with the original input data. Based on the comparison results, the system determines whether programming was successful or if defects were present, and uses this feedback information to trigger data recovery operations only when necessary, maintaining simplicity for successful programming cases
2Reliability
If defect detection and data recovery operations are implemented, then data accuracy is improved, but programming time increases due to additional read operations and mismatch counting
Solution Approach 1:
The patent applies partial action by implementing defect detection and data recovery only when necessary, not for all programming operations. The system performs read operations and mismatch counting only after programming to identify defects, and only initiates data recovery when mismatches are detected, thereby avoiding the time penalty of full recovery procedures for successful programming cases
Solution Approach 2:
The patent implements self-service by using the memory device's own read operations to detect defects and verify programming accuracy. The system leverages the existing read infrastructure to perform defect detection without requiring separate dedicated detection hardware or complex external testing equipment, thereby minimizing additional time overhead
3Measurement precision
If multiple read operations are performed to detect defects and recover data, then detection precision is improved, but the number of operations increases
Solution Approach 1:
The patent applies universality by designing the read operation to serve multiple functions simultaneously: reading the programmed data for verification, detecting defects through mismatch comparison, and enabling data recovery when necessary. This multi-functional approach eliminates the need for separate dedicated defect detection operations, thereby maintaining high throughput while achieving precise defect detection
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting the programming strategy based on defect type and location. The system changes programming parameters such as programming strength, timing, and sequence based on the detected defect characteristics, allowing precise defect detection and correction without requiring a fixed large number of operations for all cases
Data Source
AI summary
A programming operation for a set of non-volatile storage elements determines whether the storage elements have been programmed properly after a program-verify test is passed and a program status=pass is issued. Write data is reconstructed from sets of latches associated with the storage elements using logical operations optionally one or more reconstruction read operations. Normal read operations are also performed to obtain read data. A number of mismatches between the read data and the reconstructed write data is determined, and determination is made as to whether re-writing of the write data is required based on the number of the mismatches.