On-the-fly Memory Repair Using Redundancy Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
After being shipped to customers, non-volatile memories can develop bad blocks due to usage, leading to operational failures, especially since users lack the necessary equipment for on-site repair, necessitating an on-the-fly repair method to maintain memory functionality.
Innovation Solution
The method employs redundancy blocks within the memory to automatically replace failed blocks by programming error correction data into redundancy information regions, allowing for on-the-fly repair even in systems without a memory controller, thereby maintaining memory functionality and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If redundancy blocks are used to repair bad blocks before shipping, then manufacturing yield is improved, but redundancy blocks are consumed and cannot repair newly-found bad blocks after shipping
Solution Approach 1:
The patent pre-programs redundancy information into the memory device before shipping, including multiple redundancy block addresses and their corresponding mapping relationships. This preliminary preparation enables the memory device to perform self-repair after shipping without external equipment, resolving the contradiction between manufacturing yield improvement and post-shipment reliability maintenance.
Solution Approach 2:
The memory device performs automatic self-repair by using its internal redundancy information to locate and activate replacement blocks when bad blocks are detected during operation. This self-service capability eliminates the need for external repair equipment after shipping, allowing the device to maintain reliability autonomously.
2Reliability
If laser repair is used to repair bad blocks before shipping, then bad blocks are corrected, but users lack suitable machines to repair newly-found bad blocks after shipping
Solution Approach 1:
The memory device is equipped with self-diagnosis and self-repair capabilities through internal redundancy information storage. When a bad block is detected after shipping, the device automatically retrieves replacement block addresses from its internal redundancy information and performs block switching without requiring external laser repair equipment, enabling ease of repair in field conditions.
Solution Approach 2:
Redundancy information including multiple replacement block addresses is pre-programmed into the memory device before shipping. This preliminary action ensures that when bad blocks are found after shipping, the device already has the necessary repair information stored internally, eliminating the need for external repair machinery.
3Productivity
If redundancy blocks are reserved for pre-shipment repair, then manufacturing efficiency is improved, but the number of available redundancy blocks decreases for post-shipment repairs
Solution Approach 1:
The patent pre-programs multiple redundancy block addresses and their mapping relationships into the memory device before shipping. This preliminary action allows the device to adapt to various bad block scenarios after shipping by selecting from multiple pre-configured redundancy blocks, maintaining post-shipment adaptability while supporting manufacturing efficiency.
Solution Approach 2:
The redundancy information storage structure is designed to serve multiple purposes: it supports pre-shipment bad block repair, post-shipment self-repair, and provides multiple replacement options for different failure scenarios. This multi-functionality resolves the contradiction between manufacturing efficiency and post-shipment adaptability.
Data Source
AI summary
An on-the-fly repair method for a memory includes: performing a block erase operation on the memory; checking whether the block erase operation is passed or not; finding whether there is any available and healthy redundancy block in the memory if the block erase operation is not passed; programming an address of a failed block to be repaired, an enable bit and at least one error correction bit into both first and second redundancy information regions in a redundancy information set of the memory; checking whether error in the first and the second redundancy information regions is recoverable based on the error correction bit; and if the error is recoverable, then programming the redundancy information set as effective to replace the failed block by the redundancy block related to the effective redundancy information set.