Memory Cell State Determination Using Multiple Reference Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memories with small reading windows, such as RRAMs and MRAMs, face challenges in reliably distinguishing between states 0 and 1 due to narrow or shifting separation gaps, leading to potential errors in data retrieval.
Innovation Solution
A method involving the use of multiple reference values to determine data words for memory cells, where each cell is compared against a range of reference values to assess its state, allowing for error-tolerant handling and improved reliability even with small reading windows, by assigning different values based on comparisons with smallest, largest, and intermediate reference values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single reference value is used for reading memory cells, then the reading process is simple and fast, but the reliability of state determination deteriorates when the reading window is small
Solution Approach 1:
The patent divides the reading process into multiple independent comparison operations, where each memory cell is compared with multiple reference values (first reference value and second reference value) to generate multiple data words. This segmentation allows reliable state determination even with small reading windows by examining multiple reference points rather than relying on a single threshold.
Solution Approach 2:
The patent transitions from a single-dimensional reading approach (one reference value) to a multi-dimensional approach by introducing multiple reference values and generating multiple data words (first data word and second data word). This dimensional expansion enables the system to distinguish between states more reliably when the reading window is small by creating a richer comparison space.
2Measurement precision
If multiple reference values are used to improve reading reliability, then the accuracy of state determination improves, but the reading time and processing complexity increase
Solution Approach 1:
The patent merges the results from multiple comparison operations by generating both a first data word and a second data word from the same memory cell readings using different reference values. These merged data words are then used together to determine the final state, allowing the system to achieve high accuracy through combined information while optimizing the reading process.
Solution Approach 2:
The patent creates copies of the reading process by performing multiple comparisons with different reference values. Each comparison generates a data word that is a copy of the original reading interpreted through a different reference frame. These copies are then synthesized to produce the final accurate state determination, enabling parallel processing and reduced time loss.
3Quantity of substance
If the reading window is made smaller to improve memory density, then more memory cells can be packed, but the ability to distinguish between states 0 and 1 deteriorates
Solution Approach 1:
The patent introduces multiple reference values as intermediaries between the memory cell states and the reading process. These reference values act as mediators that help distinguish between states 0 and 1 even when the reading window is small. By comparing the memory cell reading against multiple reference points, the system can reliably identify states despite the reduced separation gap caused by higher density.
Data Source
AI summary
A method for determining a resultant data word when accessing memory cells includes reading a set of memory cells, and determining first and second data words therefrom. Each memory cell is assigned a component of the first and second data words. The first and second data words for the respective memory cell assume a first value if a first comparison with a first reference value and a second comparison with a second reference value show that the two reference values are greater and assume a second value if the first comparison with the first reference value and the second comparison with the second reference value show that the two reference values are smaller. The first and second data words assume at least one third value if neither condition is satisfied. The resultant data word is determined based on the first or second data words. A corresponding device is also proposed.