Dynamic Trim Management for Memory Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional memory systems face inefficiencies in processing level calibration due to unnecessary testing across memory devices, leading to increased manufacturing time and resource consumption, as they calibrate all trims until the slowest converging device satisfies the test condition, rather than dynamically managing the optimization trim list.
Innovation Solution
Implementing a dynamic trim management mechanism that continuously samples and calculates feedback measures for read levels, allowing for iterative calibration of individual read levels and optimizing the subset of trims, thereby reducing the number of reads necessary and eliminating calibrated trims from the testing process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional memory systems calibrate all trims until the slowest converging device satisfies the test condition, then calibration reliability is improved, but manufacturing time and resource consumption increase
Solution Approach 1:
The patent implements dynamic trim management where the calibration process adapts in real-time based on device characteristics. The system continuously monitors calibration progress and dynamically adjusts the trim list, removing calibrated trims and focusing resources on non-converged trims. This dynamic approach allows the system to maintain high calibration reliability while significantly reducing manufacturing time by avoiding unnecessary calibration of already-converged devices.
Solution Approach 2:
The patent applies local quality by treating different memory devices and trims individually based on their specific calibration needs. Instead of uniform calibration across all devices, the system identifies and focuses calibration efforts on specific trims that require attention (non-converged trims), while skipping those that have already met calibration criteria. This localized approach ensures reliability for critical trims while eliminating waste on already-calibrated ones.
2Reliability
If traditional memory systems calibrate all trims across all devices, then comprehensive calibration coverage is improved, but resource consumption increases
Solution Approach 1:
The patent extracts and removes calibrated trims from the active calibration list once they satisfy test conditions. By taking out these successfully calibrated trims from further processing, the system maintains comprehensive calibration coverage for remaining trims while eliminating unnecessary resource consumption on already-calibrated trims. This extraction mechanism ensures that resources are focused only where needed.
Solution Approach 2:
The patent implements partial action by calibrating only the necessary subset of trims that require calibration at any given time, rather than continuously calibrating all trims across all devices. The system performs calibration on non-converged trims and stops when calibration criteria are met, avoiding excessive calibration actions on already-satisfied devices and trims, thereby reducing resource consumption while maintaining adequate coverage.
3Measurement precision
If traditional memory systems test all trims until slowest device converges, then calibration accuracy is improved, but productivity decreases
Solution Approach 1:
The patent performs preliminary calibration actions on trims that are likely to converge quickly or have already converged, identifying and removing them from the calibration list before they would slow down the overall process. By taking preliminary actions to identify and exclude easily-calibrated trims, the system maintains calibration accuracy for difficult trims while improving overall manufacturing productivity by eliminating bottlenecks.
Solution Approach 2:
The patent implements skipping by rapidly identifying and bypassing trims that have already converged or are unlikely to require extensive calibration. The system rushes through the identification and removal of calibrated trims from the list, allowing the calibration process to focus efficiently on remaining non-converged trims. This skipping mechanism maintains accuracy for critical trims while dramatically improving productivity by eliminating time spent on already-satisfied devices.
Data Source
AI summary
A system comprises a memory device comprising a plurality of memory cells; and a processing device coupled to the memory device, the processing device configured to manage optimization target data that at least initially includes read levels in addition to a target trip, wherein the optimization data is managed based on iteratively calibrating the read levels and removing the calibrated levels from the optimization target data.