Data Training Device for High-Speed Memory Signal Alignment

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High-speed semiconductor memory devices face challenges in maintaining precise data transfer due to shorter setup and hold times of chip select signals, leading to potential data skew and erroneous data input, especially as operation frequency increases, making stable data transfer difficult in high-speed systems.

Innovation Solution

A data training device that includes a training control block to activate driving signals for a bit line sense amplifier with a deactivated word line during write training operations, allowing the bit line sense amplifier to store training data and maintain data integrity by aligning DQ data with the DQS signal, thereby securing a precise data window.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If operation frequency is increased to achieve high-speed data transfer, then productivity is improved, but data transfer precision deteriorates due to shorter setup and hold times causing data skew

Engineering Contradiction:
Improveoperation frequencyVSAvoiddata transfer precision
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent applies preliminary action by performing write training operations before normal data transfer to pre-align the DQ data with the DQS signal. The training control block executes training writes that adjust timing parameters in advance, ensuring that when normal operation begins at high frequency, the data window is already optimized and data skew is minimized without compromising precision.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If setup and hold times are reduced to increase data throughput, then productivity is improved, but reliability deteriorates due to data falling outside the valid window

Engineering Contradiction:
Improvedata throughputVSAvoiddata integrity
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements feedback through the training control block that monitors data alignment between DQ and DQS signals during write training operations. Based on this feedback, the system adjusts timing parameters and setup/hold times to optimize the data window. This closed-loop approach ensures that even with reduced setup and hold times for high throughput, the data remains within the valid window, maintaining reliability through continuous timing optimization.

Inventive Principle:
Principle #23Feedback

3Manufacturing precision

If additional data storage circuits are added to maintain data window precision, then manufacturing precision is improved, but device complexity increases

Engineering Contradiction:
Improvedata window precisionVSAvoidcircuit complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies universality by making the bit line sense amplifier perform multiple functions: it serves both as the primary data storage element during normal operation and as a training data storage element during write training operations. The training control block activates the sense amplifier to store training data, allowing timing optimization without requiring separate dedicated training storage circuits. This multi-functional approach maintains data window precision while avoiding additional circuit complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS9019784B2Data training device
Publication Date: 2015.04.28 MIMIRIP LLC
  • US9019784B2 patent drawing
  • US9019784B2 patent drawing
  • US9019784B2 patent drawing

AI summary

A data training device includes a training control block configured to activate driving signals for driving a bit line sense amplifier, with a word line deactivated, when a write training operation is performed according to a mode register write command; and the bit line sense amplifier configured to store training data according to the driving signals from the training control block.