Clock Phase Recovery with Dynamic Accumulator Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high-speed data transfer systems, accumulators used in training algorithms face challenges in balancing convergence time and accuracy due to the need for repeated evaluations to mitigate supply noise and signal jitter.
Innovation Solution
A clock phase recovery method that dynamically updates the accumulator threshold based on convergence points determined by analyzing sampled data from in-phase and quadrate-phase clocks, allowing for faster convergence without compromising accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high accumulator threshold is used, then noise filtering and accuracy are improved, but convergence time increases
Solution Approach 1:
The patent applies dynamics by making the accumulator threshold adjustable rather than fixed. The threshold is dynamically modified based on the training stage: a first threshold is used during an initial training stage to enable fast convergence, and a second (higher) threshold is used during a final training stage to ensure accurate locking. This dynamic adjustment resolves the contradiction by allowing the system to prioritize speed when far from convergence and accuracy when near convergence.
Solution Approach 2:
The patent changes the threshold parameter throughout the training process. Instead of using a single fixed threshold value, the system transitions from a lower threshold value during early training iterations to a higher threshold value during later iterations. This parameter change enables the system to achieve both fast initial convergence and accurate final locking, resolving the trade-off between convergence speed and accuracy.
2Productivity
If a low accumulator threshold is used, then convergence time is reduced, but noise filtering and accuracy are compromised
Solution Approach 1:
The system dynamically adjusts the threshold based on training progress. During the initial training stage, a lower threshold is applied to achieve fast convergence. As training progresses to the final stage, the threshold is increased to a second value that provides better noise filtering and accuracy. This dynamic behavior allows the system to achieve both fast convergence and high accuracy at different stages.
Solution Approach 2:
The patent implements preliminary action by using a lower threshold during the initial training stage to quickly bring the system close to the convergence point. This preliminary fast convergence phase prepares the system for the subsequent final training stage, where a higher threshold is used to achieve accurate locking. The preliminary action of fast convergence sets the stage for the final accurate convergence.
Data Source
AI summary
The present disclosure relates to a system and method for clock phase recovery. Embodiments may include sampling data using an in-phase clock and a quadrate phase clock. Embodiments may further include analyzing sampled data from the in-phase clock and the quadrate phase clock. Embodiments may also include determining a convergence point based upon, at least in part, the analyzed sampled data, wherein the convergence point corresponds to a point where a number of early sampled outcomes is approximately equal to a number of late sampled outcomes. Embodiments may also include dynamically updating an accumulator threshold based upon the convergence point.