Memory Characterization Using CDF Data for Read Voltage Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory sub-systems lack the ability to identify and address nuanced issues in memory units during manufacturing and operation, leading to inadequate early identification of problematic units and inefficient error recovery.
Innovation Solution
Characterization of memory units using Cumulative Distribution Function (CDF)-based data to identify exceptional units, with methods like Auto Read Calibration (ARC) and Continuous Read Level Calibration (cRLC) to determine optimal read voltage thresholds and error recovery strategies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional memory testing methods are used, then manufacturing process is simple, but ability to identify nuanced issues in memory units is insufficient
Solution Approach 1:
The patent segments the memory unit analysis into multiple dimensions by generating CDF-based data for different parameters (read errors, program failures, erase failures) and analyzing exceptional memory units through multiple characterization procedures (ARC, cRLC, PCCC). This segmentation allows precise identification of nuanced issues while organizing complexity into manageable analytical components.
Solution Approach 2:
The patent applies parameter changes by using CDF-based data to identify exceptional memory units and then applying different characterization procedures with varying parameters (read voltage thresholds, program voltage levels, erase voltage settings). This enables precise measurement of nuanced issues by adjusting and comparing multiple parameters across different test procedures.
2Reliability
If no early identification of problematic units is performed, then manufacturing process is faster, but error recovery efficiency is reduced
Solution Approach 1:
The patent implements preliminary action by performing comprehensive memory unit characterization during manufacturing before the memory devices are deployed. CDF-based data is generated and analyzed to identify exceptional memory units early, allowing problematic units to be flagged and handled differently before they cause errors in operation, thus improving reliability without significant time loss.
Solution Approach 2:
The patent uses feedback mechanisms where CDF-based data from manufacturing characterization is fed back into the production process to identify and flag exceptional memory units. This feedback loop enables early detection of problematic units, allowing for adjusted handling or allocation strategies that improve error recovery efficiency while minimizing time loss through targeted rather than universal re-testing.
3Manufacturing precision
If optimal read voltage thresholds are not set, then manufacturing is simpler, but read error rates increase
Solution Approach 1:
The patent applies self-service by implementing Auto Read Calibration (ARC) procedures that automatically determine optimal read voltage thresholds for each memory unit without requiring extensive manual intervention. The system uses CDF-based data from automated tests to self-calibrate voltage settings, achieving high manufacturing precision while maintaining ease of manufacture through automation.
Solution Approach 2:
The patent uses parameter changes in the ARC and cRLC procedures to systematically adjust and optimize read voltage thresholds. By varying read voltage parameters across a range of values and analyzing the resulting CDF-based error data, the system automatically identifies optimal threshold settings that maximize read accuracy while streamlining the calibration process through algorithmic parameter optimization.
Data Source
AI summary
A memory system can identify target memory units to characterize by generating Cumulative Distribution Function (CDF)-based data for each memory unit and analyzing the CDF-based data to identify target memory units that are exceptional. Such target memory units can be those with CDF-based data with extrinsic tails or that crosses an info limit threshold. The memory system can perform characterization processes for the target memory units, e.g. using an Auto Read Calibration (ARC) analysis or a Continuous Read Level Calibration (cRLC) analysis. A manufacturing process for the memory device can use results of the characterization processes, e.g. by mapping them to types of problems observed during testing. Alternatively, results of the characterization processes to can be used during operation of the memory device, e.g. to adjust the initial read voltage threshold, the read retry voltage values, or the order of read retry voltages used in data recovery.