Lesson 9/1090%
MODULE 09

Eye Diagram Analysis

Learn to interpret eye diagrams, quantify timing/noise margins, understand jitter and ISI effects, and evaluate channel quality for high-speed serial links.

60-90 min
Advanced
6 Sections

Eye Diagram Formation

An eye diagram is created by overlaying many consecutive unit intervals (UI) of a digital waveform onto a single time window. The resulting pattern resembles a human eye. It provides instantaneous visualization of signal quality, timing margin, noise margin, jitter, and channel quality.

Formation process: Generate PRBS → Transmit through channel → Slice into UI segments → Overlay all segments on a single time axis aligned to clock edges.

A wide-open eye = good SI, high noise immunity, reliable sampling. A closed eye = bit errors, timing violations, poor channel performance.

Key Eye Measurements

Eye Height (Vertical Opening)
Voltage margin available for receiver decision. Reduced by noise, ISI, and reflections.
Eye Width (Horizontal Opening)
Timing margin for receiver sampling. Reduced by jitter, ISI, and data rate.
Jitter (Horizontal Spread)
Timing uncertainty of transitions. Random (Gaussian, unbounded) + Deterministic (bounded).
BER (Bit Error Rate)
Probability of sampling error. Related to eye opening via Q-factor: BER = erfc(Q/√2)/2.

Real-Time Eye Diagram Generator

LIVE SIMULATION

Adjust data rate, rise time, noise, and jitter to observe how each impairment affects eye quality. The simulator generates a PRBS-7 pattern, applies channel effects, and renders a real-time eye diagram with measurements.

Data Rate
Rate (Gbps)5.0
Rise Time (ps)50
Impairments
Noise (mV)20
Jitter (ps)10
Channel
Loss (dB)3
ISI (symbol)0.1
Measurements
Eye Height--
Eye Width--
Noise Margin--
Timing Margin--
Est. BER--
Eye Diagram (2 UI overlay)
Eye Quality Dashboard

Jitter Decomposition

Total Jitter (TJ) = Deterministic Jitter (DJ) + Random Jitter (RJ). DJ is bounded and includes duty-cycle distortion, ISI, and periodic jitter. RJ follows Gaussian distribution and is unbounded — at BER=10-12, peak-to-peak RJ ≈ 14×σRJ.

Total Jitter at BER
TJ(BER) = DJ + 2 × N(BER) × σRJ
N(10-12) = 7.03 | N(10-15) = 7.94

Noise Sources

Random Noise (Thermal)
Gaussian, reduces eye height symmetrically. Vn = √(4kTRB)
Crosstalk
Data-dependent noise from adjacent lanes. Bounded, deterministic.
Power Supply Noise
Periodic, correlated with switching activity. Affects both amplitude and timing.
ISI (Inter-Symbol Interference)
Channel memory creates pattern-dependent voltage at sampling instant.

Exercise 1: 5 Gbps Link Assessment

A 5 Gbps serial link shows: Eye Height = 180mV, Eye Width = 60ps, Jitter(pp) = 90ps.

Task 1: UI = 1/5G = 200ps. Timing margin = (Eye Width/UI)×100 = 30%
Task 2: Noise margin = Eye Height/(2×Vswing)×100 = 180/(2×400)×100 = 22.5%
Task 3: Jitter exceeds eye width! 90ps > 60ps. BER will be poor.
Task 4: Add equalization (CTLE/DFE), reduce channel loss, improve PDN.

Exercise 2: Channel Comparison

Compare two channels at 10 Gbps (UI=100ps, Vswing=800mV):

Channel A
Noise=20mV, Jitter=15ps
Eye H=760mV, Eye W=70ps
Channel B
Noise=100mV, Jitter=40ps
Eye H=600mV, Eye W=20ps
Result: Channel A wins — 70% timing margin vs 20%, and 95% voltage margin vs 75%. Channel B likely fails BER target.
Question 1 of 8 Score: 0/8
☐ Eye height exceeds receiver specification
☐ Eye width exceeds timing requirement
☐ BER target achieved (typically 10-12)
☐ Noise margin verified under worst-case
☐ Jitter budget validated (TJ < 0.3 UI)
☐ Rise/fall times within spec
☐ Channel loss acceptable at Nyquist
☐ ISI analyzed and equalization sufficient
☐ Compliance eye mask passed
☐ Receiver sampling point optimized

Ready to complete this module?

Mark complete after reviewing all sections and passing the quiz.

📚

Extra Study: Eye Diagram Deep Dive

Optional

Go beyond the basics with an 8-module interactive course covering jitter decomposition, noise analysis, ISI effects, bathtub curves, compliance testing, and real-time eye diagram simulation with PCB visualization.

Launch Eye Diagram Course 8 modules • Interactive simulations • No prerequisite