Read Margin Health Evaluation via Cell Activation Counting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory systems face challenges in accurately evaluating read margin health due to shifts in threshold voltage levels over time, leading to false positive errors and unnecessary resource wastage, particularly in high-performance applications like AI, AR, VR, and gaming, which require efficient processing and reduced latency.
Innovation Solution
A memory system evaluates read margin health by determining the quantity of activated memory cells in response to specific bias voltages applied to distinguish between logic states, thereby assessing the health of memory cells without data decoding or error count detection, reducing latency and false detection rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional read margin health evaluation methods are used, then data can be read from memory cells, but false positive errors occur due to threshold voltage shifts over time
Solution Approach 1:
The patent changes the evaluation parameter from traditional bit error count (BEC) methods to activation quantity-based evaluation. By applying different bias voltages and counting the number of activated memory cells, the system creates a new metric that is not affected by threshold voltage shifts, thereby resolving the contradiction between reliability and measurement precision.
Solution Approach 2:
The patent substitutes the traditional electrical readout mechanism with a counting mechanism. Instead of detecting voltage levels to determine logic states and counting bit errors, the system applies bias voltages and counts the quantity of activated cells, replacing complex voltage-based detection with a simpler activation counting approach that avoids threshold voltage shift issues.
2Reliability
If media scan operations are performed to detect read errors, then error detection is achieved, but latency increases and resources are wasted due to false positives
Solution Approach 1:
The patent performs preliminary evaluation of read margin health by continuously monitoring activation quantities at different bias voltages. This preliminary action allows the system to detect actual health degradation before it causes false positives in traditional BEC-based error detection, enabling proactive maintenance without unnecessary media scans and reducing latency.
Solution Approach 2:
The system uses its own operational data (activation quantities at different bias voltages) to self-diagnose read margin health without requiring external intervention or complex error detection algorithms. This self-service approach eliminates the need for resource-intensive media scan operations and reduces latency by using already-available measurement data.
3Measurement precision
If multiple read voltage levels are used to distinguish logic states, then reading accuracy is improved, but complexity of the evaluation system increases
Solution Approach 1:
The patent segments the evaluation process into distinct steps: applying a first bias voltage to activate memory cells, counting activated cells, applying a second bias voltage, and counting activated cells again. This segmentation simplifies the evaluation logic compared to managing multiple read voltage levels, while maintaining accurate distinction between logic states through systematic voltage application and counting.
Data Source
AI summary
Methods, systems, and devices for read margin health evaluations for memory systems are described. A memory system may evaluate a read margin health for a set of memory cells (e.g., a page of memory cells, a block of memory cells) based on quantities of memory cells that activate in response to applying voltages to the set of memory cells. For example, the memory system may determine, for a pair of bias voltages between a first nominal voltage associated with a first logic state and a second nominal voltage associated with a second logic state, a respective quantity of memory cells of the set of memory cells that are activated in response to biasing the set of memory cells with each of the pair of voltages. Based on the quantities of activated memory cells, the memory system may determine whether to refresh the set of memory cells.