Storage Controller Signal Training Using Offset Window Edges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor memory systems face challenges in improving the reliability and speed of data transmission between storage controllers and nonvolatile memory devices due to misalignment of data and data strobe signal timings, leading to increased boot-up times and inefficient training processes.
Innovation Solution
A storage controller performs a two-stage training process, including a first boot-up operation for initial alignment followed by a second operation focused on the left and right edge sections of data signal windows, using offset information generated from the first training to reduce detection steps and accelerate boot-up times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a full training process is performed on all data signal windows during each boot-up operation, then the alignment precision of data signals is improved, but the boot-up time increases
Solution Approach 1:
The training process is segmented into two distinct phases: a first training operation that performs detection on all data signal windows to establish comprehensive offset information, and a second training operation that performs detection only on edge sections (first and last windows) to update timing alignment. This segmentation allows the system to achieve precise alignment while significantly reducing the time required for subsequent boot-up operations.
Solution Approach 2:
Offset information is generated during the first training operation as preliminary data that captures the timing characteristics of all data signals. This pre-computed offset information is then reused during the second training operation, eliminating the need to perform full detection on all windows during every boot-up and thereby reducing boot-up time while maintaining alignment precision.
2Reliability
If the detection operation is repeated on all windows of data signals during second training, then the reliability of data transmission is improved, but the training speed decreases
Solution Approach 1:
The second training operation extracts and focuses only on the critical edge sections (first and last windows) of the data signal windows for detection, rather than processing all windows. This extraction of essential detection targets maintains data transmission reliability by verifying timing alignment at the boundaries where errors are most likely to occur, while dramatically improving training speed by eliminating redundant detection on intermediate windows.
3Device complexity
If offset information is not stored and reused, then the training process remains simple, but the boot-up operation must be performed repeatedly on all data signals
Solution Approach 1:
Offset information generated during the first training operation is stored and copied for use during the second training operation. This copying of previously computed timing offset data allows the system to reuse valuable alignment information without regenerating it, thereby reducing the computational burden during boot-up operations and significantly decreasing boot-up time while maintaining reasonable process complexity.
Data Source
AI summary
An operation method of a storage device, which includes a storage controller and a nonvolatile memory device, includes performing first boot-up operation, performing first training on a plurality of data signals such that a detection operation of the first training is repeatedly performed on windows of the data signals, storing offset information generated based on a result of the first training, performing a normal operation based on the result of the first training, performing a second boot-up operation, performing second training on the plurality of data signals based on the offset information generated in the first training such that a detection operation of the second training is repeatedly performed on a left edge section and a right edge section of windows of the data signals, and performing the normal operation based on a result of the second training.