Wavelet Noise Filtering for PHY Electrical Idle Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed serial technologies in information handling systems face issues with non-deterministic noise during idle conditions, leading to unwanted signal threshold violations and potential system crashes, such as Windows 'blue screen' errors, due to insufficient noise margin in PCI-E and other serial data buses.
Innovation Solution
The implementation of a wavelet transform analysis system that minimizes non-deterministic noise in the PHY layer by decomposing and analyzing signals in both time and frequency domains, enhancing the noise margin and maintaining the electrical idle state within specified limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If wavelet transform analysis is applied to minimize non-deterministic noise, then noise margin is improved and electrical idle state is maintained, but device complexity increases
Solution Approach 1:
The patent extracts and separates the noise component from the signal using wavelet transform decomposition. By decomposing the signal into different frequency bands and identifying the noise portion, the system can selectively filter out non-deterministic noise while preserving the legitimate signal, thereby improving electrical idle state stability without requiring complete signal rejection
Solution Approach 2:
The wavelet transform provides dynamic time-frequency analysis that adapts to changing signal characteristics. The decomposition level and filtering parameters can be dynamically adjusted based on the observed signal conditions, allowing the system to optimize noise reduction performance in real-time while maintaining responsiveness to actual data transmissions
2Reliability
If noise margin is increased to prevent false activations, then system reliability is improved, but the noise specification budget is consumed faster
Solution Approach 1:
The patent applies localized noise filtering by targeting specific frequency bands where non-deterministic noise occurs, rather than uniformly increasing noise margin across all frequencies. The wavelet transform identifies and treats only the problematic frequency components, preserving noise margin in frequency regions where it is not needed and preventing false activations only where necessary
Data Source
AI summary
A wavelet transform noise minimization circuit comprises a differential receiver, a voltage comparator, a wavelet transform circuit, an electrical idle (EI) detector circuit, a phase interpolator, a phase-locked-loop (PLL), and a reference clock buffer. The wavelet transform noise minimization circuit may be beneficially applied wherever there is non-deterministic (e.g., random) noise in the PHY layer during an electrical idle state. The wavelet transform noise minimization circuit may be used to improve noise margin during an electrical idle state, and/or reduce the occurrence of false activation of a PHY layer when in the electrical idle state.