Semiconductor Memory Reference Value Determination via Error Exclusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining reference values in memory cells are inefficient and prone to errors due to the inclusion of defective or erroneous memory cells, which can lead to incorrect digital value assignments, especially in cases where the separation between reference values is small.
Innovation Solution
A method and device for determining a reference value by selecting a subset of half reference values from memory cells, excluding erroneous or deviating values, and using operations such as averaging or median calculation to derive a meaningful reference value, which can also consider external signals for correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all half reference values from memory cells are used to determine the reference value, then the determination process is simple, but the reliability is reduced due to inclusion of defective or erroneous memory cells
Solution Approach 1:
The patent extracts and excludes erroneous or defective half reference values from the set of all half reference values before determining the reference value. This is achieved by comparing half reference values and identifying outliers that deviate from expected ranges, then removing these problematic values from consideration. This extraction principle directly resolves the contradiction by improving reliability through selective exclusion while managing complexity through systematic comparison procedures.
2Reliability
If memory cells are read frequently to determine reference values, then the reference value is up-to-date, but the energy consumption and time loss increase
Solution Approach 1:
The patent performs preliminary reading and evaluation of memory cells to identify and exclude defective cells before the actual reference value determination is needed. By pre-characterizing memory cells and storing information about their reliability, the system avoids repeated reading of the same cells, thus reducing time loss and energy consumption while maintaining reference value accuracy through selective use of reliable cells.
3Productivity
If a small number of half reference values are used to determine the reference value, then the process is fast, but the precision is reduced due to insufficient sampling
Solution Approach 1:
The patent changes the parameter of sample selection from random or sequential to error-based selection. Instead of using a fixed small number of half reference values, the system dynamically selects a subset based on error detection criteria, ensuring that the selected values are free from defects. This parameter change allows for faster processing with sufficient precision by focusing computational resources on high-quality samples rather than increasing the total number of samples.
Data Source
AI summary
A method and associated apparatus to determine a reference value on the basis of a plurality of half reference values stored in memory cells is disclosed, wherein the plurality of half reference values are read from the memory cells, wherein a subset of half reference values is determined from the plurality of half reference values, and wherein the reference value is determined on the basis of the subset of half reference values.