Phase Jitter & Bit Error Rate
The High-Speed Sampling Crisis
The Eye Diagram: Visualizing Signal Integrity
In digital communications, an Eye Diagram is formed by overlaying multiple sweeps of the signal on an oscilloscope. Jitter and noise manifest as the "closing" of the eye, both vertically (voltage) and horizontally (timing).
PACKET JITTER SIMULATOR (CV)
Variance of Delay vs. Time
High jitter causes "robotic" audio or dropped calls as the jitter buffer overflows.
Devices use buffers to re-order packets and smooth out inter-arrival times.
Eye Diagram Metrics
- Eye Width (): The time interval where the signal is stable. , where is the Unit Interval.
- Eye Height (): The vertical distance between the minimum '1' level and maximum '0' level. Noise reduces this height.
- Crossing Point: The point where transitions intersect. High transition density here indicates low jitter.
Clock and Data Recovery (CDR) & The PLL
Modern receivers do not receive a clock signal on a separate wire; they must extract it from the data stream itself using a Phase-Locked Loop (PLL).
The CDR Feedback Loop
Phase Detector $\rightarrow$ Loop Filter $\rightarrow$ VCO (Voltage Controlled Oscillator)
The PLL tracks the average transition time of the data. High-frequency jitter that exceeds the Loop Bandwidth of the PLL cannot be tracked, leading to sampling errors.
The Physics of Bit Error Rate (BER)
BER is the probability that a bit is sampled incorrectly. In a system dominated by Additive White Gaussian Noise (AWGN), the BER for NRZ (Non-Return-to-Zero) signaling is defined by the Q-function:
Where is the signal amplitude and is the RMS noise. For 10GbE fiber, we target a BER of . If the SNR drops or jitter increases, the "waterfall" curve shifts to the right, requiring significantly more power to maintain the same integrity.
QPSK Constellation & Phase Noise
Visualize how Phase Jitter (Clock drift) and AWGN (Amplitude Noise) smear symbols across decision boundaries, causing Bit Errors.
Bit Error Rate (BER)
Low SNR (e.g. 5dB) causes large, blurry "clouds" of points due to thermal noise spanning uniformly in all directions. High SNR (e.g. 30dB) results in tight clusters.
High Phase Noise causes points to "smear" in a circular arc along the rotation axis. This represents clock jitter where the receiver samples slightly early or late.
Jitter Decomposition: The Dual-Dirac Model
Total Jitter (Tj) is the sum of deterministic and random components. Because Random Jitter (Rj) follows a Gaussian distribution, it has no theoretical peak; we measure it at a specific probability.
Deterministic Jitter (Dj)
Bounded. Includes Pj (Periodic/Reflections) and DDJ (Data-Dependent/Inter-Symbol Interference). Measured as peak-to-peak.
Random Jitter (Rj)
Unbounded. Caused by thermal noise and shot noise. Measured as RMS (). Multiplied by a Q-factor (e.g., 14 for BER).
Forward Error Correction (FEC): The Safety Net
At PAM4 (56G/112G) speeds, the signal is so fragile that the eye is effectively closed. We accept a high Pre-FEC BER (e.g., $10^-4$) and use Reed-Solomon (RS) algorithms to mathematically correct errors, bringing the Post-FEC BER down to $10^-12$.