Memory DCA Clock Training for Faster Eye Window Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor memory devices face challenges in achieving high signal integrity for data signals at high speeds due to deteriorated signal integrity characteristics, and the process of duty cycle adjuster (DCA) training can be time-consuming and may fail if the command even gap is not observed.
Innovation Solution
A method for DCA training that involves selecting DCA codes corresponding to internal clock signals with specific phase differences (180°, 90°, and 270°) to adjust the duty cycle of internal clock signals, allowing for precise measurement and adjustment of the eye window size of data signals, thereby improving signal integrity without inter-phase code mixing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If DCA training is performed to adjust the duty cycle of internal clock signals, then the read margin of data signals is improved, but the training time is excessively long
Solution Approach 1:
The training process is divided into two distinct operations: first training for determining DCA codes of internal clock signals with 180° phase difference, and second training for determining DCA codes of internal clock signals with 90° and 270° phase difference. This segmentation allows systematic adjustment of duty cycles while managing training time efficiently.
Solution Approach 2:
The patent changes the measurement unit of eye window size from one unit interval to two unit intervals during first training, and vice versa during second training. This parameter change optimizes the training process by adapting the measurement granularity to the specific training operation, improving both accuracy and efficiency.
2Reliability
If DCA training is performed to improve signal integrity, then the read margin is increased, but the training may fail when command even gap is not observed
Solution Approach 1:
By separating the training into two operations targeting different phase relationships (180° and 90°/270°), the patent makes the training process more adaptable to various operating conditions, including cases where command even gap is not observed, thereby improving ease of operation.
Solution Approach 2:
The patent employs different eye window measurement units (two unit intervals for first training, one unit interval for second training) to optimize the training process under different conditions, making it more robust and applicable to a wider range of operational scenarios.
3Reliability
If traditional DCA training is used, then duty cycle adjustment can be performed, but too much time delay occurs in setting optimal duty cycle
Solution Approach 1:
The training process is segmented into two operations with different measurement units, allowing each operation to be optimized for its specific purpose. This segmentation improves training efficiency by reducing redundant measurements while maintaining duty cycle accuracy.
Solution Approach 2:
By changing the eye window measurement unit parameter based on the training operation type (two unit intervals for first training, one unit interval for second training), the patent optimizes the balance between accuracy and efficiency, reducing time delay while maintaining duty cycle precision.
Data Source
AI summary
A training method of a memory device adjusting an eye window of a data signal in response to a duty cycle adjuster (DCA) includes performing a first training operation that selects a first DCA code corresponding to a first internal clock signal having a phase difference of 180° relative to a reference internal clock signal, and performing a second training operation that selects a second DCA code and a third DCA code respectively corresponding to a second internal clock signal and a third internal clock signal having a phase difference of 90° and 270° relative to reference internal clock signal. In the first training operation, the eye window size of the data signal is measured in units of two unit intervals, and in the second training operation, the eye window size of the data signal is measured in units of one unit interval.