PAM Timing Recovery Circuit Using Pattern-Based Phase Error Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing timing recovery circuits for serial data streams encoded using Pulse-Amplitude-Modulation (PAM) schemes, such as PAM-2 and PAM-3, face challenges in accurately recovering the sampling clock due to limitations in phase offset detection, power consumption, and area usage, particularly with methods like Mueller-Müller, Early-Late, Gardner, and Oversampling techniques.
Innovation Solution
A method and circuit that samples data once per unit-interval using a clock generation device, applies a digital filter pattern decoder to estimate phase errors from adjacent data samples, and adjusts the sampling clock phase using calculated errors, with weight factors for different symbol sequences to improve phase recovery, reducing power and area consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Mueller-Müller TED is used for PAM-2 signals, then phase offset detection is achieved, but the probability of wrong phase adjustment decisions increases when applied to PAM-3 signals
Solution Approach 1:
The patent changes the fundamental parameter of phase error calculation from the conventional two-symbol Mueller-Müller method to a five-symbol pattern-based method. By analyzing longer symbol sequences (ŷ(k-2), ŷ(k-1), ŷ(k), ŷ(k+1)) and applying different weight factors based on detected patterns, the system achieves reliable phase adjustment for PAM-3 signals while maintaining compatibility with PAM-2
Solution Approach 2:
The patent segments the phase error calculation into multiple distinct patterns based on symbol sequences. Different patterns (e.g., '10-1', '01+1', '-10+1') are identified and handled with specific weight factors, allowing the system to adapt to different signal conditions and reduce wrong decisions
2Measurement precision
If Early-Late Detector is used, then phase offset detection is achieved, but power consumption and area usage increase due to requiring two samples per UI
Solution Approach 1:
The patent extracts only the essential information needed for phase detection from a single sample per UI. By using the five-symbol sequence (ŷ(k-2) to ŷ(k+1)) from one sampling point, the system eliminates the need for a second sampler while maintaining phase detection accuracy
Solution Approach 2:
The single sampler is made multi-functional by utilizing extended symbol sequences. The same single sample provides multiple symbols (ŷ(k-2), ŷ(k-1), ŷ(k), ŷ(k+1)) through storage and pattern recognition, replacing the function of multiple samplers
3Measurement precision
If Gardner TED is used, then phase offset detection is achieved, but power consumption and area usage increase due to requiring two samples per UI
Solution Approach 1:
The patent extracts phase information from a single sample by analyzing extended symbol sequences. The five-symbol pattern (ŷ(k-2), ŷ(k-1), ŷ(k), ŷ(k+1)) contains sufficient phase information, eliminating the need for a second sampler and reducing circuit area
Solution Approach 2:
The system performs preliminary storage of symbol sequences in registers before phase error calculation. By pre-storing ŷ(k-2) through ŷ(k+1) and their corresponding ADC values, the system prepares all necessary data from a single sample, avoiding the need for additional sampling hardware
4Measurement precision
If Oversampling TED is used, then optimal sampling point detection is achieved, but power consumption and area usage increase due to multiple samplers
Solution Approach 1:
The patent extracts optimal sampling point information from a single sample by analyzing the phase relationships in extended symbol sequences. The five-symbol pattern analysis reveals timing information without requiring multiple oversampled points, reducing power consumption
Solution Approach 2:
Instead of oversampling multiple times, the system uses a single sample but processes an excessive number of symbols (five symbols) from that sample. This partial action approach achieves timing recovery with one sampler rather than multiple oversampled points
5Productivity
If conventional MM-TED is used, then phase offset detection is achieved, but reliability decreases due to using only current and last detected symbols
Solution Approach 1:
The system performs preliminary storage of extended symbol sequences (ŷ(k-2), ŷ(k-1), ŷ(k), ŷ(k+1)) and their ADC values before phase error calculation. This preparation allows reliable pattern recognition and reduces wrong decisions while maintaining fast convergence
Solution Approach 2:
The phase error calculation is segmented into multiple distinct patterns based on symbol sequences. By identifying specific patterns (e.g., '10-1', '01+1') and applying appropriate weight factors, the system makes more reliable phase adjustment decisions compared to the conventional two-symbol method
Data Source
Figure 1~3
Figure 4~6
Figure 7
AI summary
The invention relates to a method and timing recovery circuit for recovering a sampling clock from a serial data stream encoded using Pulse-Amplitude-Modulation, comprising: applying a filter pattern decoder to detected symbol sequence at more than two adjacent data symbols, particularly to the detected symbol patterns of four adjacent samples ŷ(k— 2),ŷ(k— 1),ŷ(k),ŷ(k+ 1), and calculating an estimated phase error e(k).