Worst Case Eye Diagram Evaluation for Multi-Level PAM Links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining signal integrity in multi-level pulse amplitude modulated links are time and resource intensive, particularly when aiming for low bit error rates, as they require extensive simulations that are computationally costly and often fail to meet stringent reliability standards within practical timeframes.
Innovation Solution
A computing system that performs a fast worst-case eye evaluation by identifying multiple step responses of the channel, determining distribution boundaries at each value level, and using these boundaries to predict signal integrity, thereby generating worst-case input patterns for eye diagram analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If time-domain simulation is used to determine signal integrity, then measurement precision is improved, but productivity deteriorates due to excessive computation time
Solution Approach 1:
The patent segments the signal integrity analysis into multiple components: channel impulse response characterization, worst-case input pattern generation, and eye diagram construction. This allows the complex simulation to be broken down into manageable steps that can be performed more efficiently
Solution Approach 2:
The patent performs preliminary characterization of the channel impulse response before generating worst-case patterns. By pre-computing channel properties and storing them for reuse, the method avoids redundant calculations during the actual signal integrity assessment, significantly reducing computation time
2Reliability
If bit-by-bit simulation is performed to achieve low bit error rates, then reliability is improved, but loss of time increases due to the need to simulate billions of bits
Solution Approach 1:
The patent creates a simplified model (eye diagram) that copies the essential characteristics of long-bit-sequence simulation results. By analyzing the channel's impulse response and generating representative worst-case patterns, the method produces eye diagrams that predict bit error rates for 1e-15 to 1e-20 without actually transmitting or simulating that many bits
Solution Approach 2:
The patent changes the simulation parameters by focusing on critical transitions and worst-case patterns rather than uniformly simulating all possible bit sequences. This selective approach to parameter selection allows achieving the same reliability prediction with fraction of the computational effort
3Productivity
If multi-level signal encoding is implemented to increase data throughput, then productivity is improved, but device complexity increases due to additional voltage levels and processing requirements
Solution Approach 1:
The patent develops a universal eye diagram methodology that can be applied to both binary and multi-level signaling schemes. The same basic approach of characterizing channel response and constructing eye diagrams works for PAM-2, PAM-4, and other encoding formats, providing a multi-functional analysis tool that reduces the need for separate complex analysis methods for each signaling type
Data Source
AI summary
This application discloses a computing system to perform a fast evaluation of a worst case eye diagram for a channel capable of communicating signals encoding data in more than two value levels. The computing system can identify multiple step responses of the channel, each corresponding to a transition between a plurality of the value levels. The computing system can determine distribution boundaries of the signals at each of the value levels based, at least in part, on the step responses of the channel. The computing system can utilize the distribution boundaries at the value levels to determine boundaries of eye openings between adjacent value levels or to build worst case input patterns used to generate the worst case eye diagram for the channel. The computing system can predict a signal integrity of the channel based on the distribution boundaries at each of the value levels.