Jitter Jianmin Zhang

Jitter

• Jianmin Zhang, David J. Pommerenke

References

• Agilent Technologies, numerous application notes, see http://www.home.agilent.com/agilent/facet.jspx?to=80030.k.1&co=153297.i.1&cc=US&lc=eng&sm=g&k=jitter

• LeCroy, Tektronix, Bertscope

• J. Hancock, “Jitter—understanding it, measuring it, eliminating it Part 1-3”, From 2004 High Frequency Electronics

• A. Kuo etc. “Jitter models and measurement methods for high-speed serial interconnects”

• S. Tabatabaei etc, “Jitter generation and measurement for test of multi-GBPS serial IO”

• Altera Web Site, Le Croy Web site

Outline

Eye pattern

• A very effective method of measuring time distortion thru a data transmission system is based on the eye pattern, displayed on an oscilloscope.

• The eye pattern is simply the superposition - over one unit interval – of all the Zero–to–One and One–to-Zero transitions, each preceded and followed by various combinations of One and Zero, and also constant One and Zero levels.

• The data sequence can be generated by a pseudo-random sequence generator (PRSG), which is a digital shift register with feedback connected to produce a max length sequence.

What Is Jitter?

Why Jitter Is Important?

• Jitter causes transmission errors

• or in another wording: Jitter limits the transmission speed.

• Satisfy jitter budget  BER (bit error rate) target

• Identify jitter components  diminish/decrease deleterious effects on circuit performance from jitter

• Jitter limits the ability of A/D converters!

• Note:

• Bit errors can also be caused by voltage noise: If the momentary noise voltage exceeds the noise margin, a wrong value can be sampled even if the sampling takes place at the correct moment in time.

Jitter applications

• Signal jitter: Timing of a

• signal (topic of this presentation)

Signal jitter: Why Jitter Is Used?

• Jitter is not a main system property, but Bit Error Rate (BER) is.

• Why not measuring BER directly?

• BER measurement might take hours or days.
• BER gives little information about the mechanism that cause errors, but jitter does.

What Causes Jitter

How does the Spectra look like?

How Does the Eye-Diagram Look Like?

Effect of data length:

• Few hundred bits:

• Billion bits:

TJ Estimation

Real Zero Crossings vs. Ideal Ones

Time interval error and jitter trend (integral of time interval error)

Histogram in A Square Wave

Channel Characteristics

Eye of a loss dominated differential channel

What Can You Tell?

Sequence of Jitter analysis

• Step 1: Measure

• - Real time scope

• - Equivalent time scope

• - Spectrum Analyzer

• Step 2: Separate jitter components

• - Average

• - Change data pattern (PRBS to 01010101)

• - TIE

• - etc.

Instruments to measure jitter

• Real time oscilloscope:

• Usually 8 bit, absolute clock, clock re-generation possible, bandwidth up to 20 GHz

• Sampling oscilloscope:

• 11-14 bit, relative to clock, bandwidth up to 1000 GHz

• Spectrum Analyzer:

• No time resolution (“real time spectrum analyzers” are an exception), very good dynamic range. Bandwidth up to 1000 GHz.

Real Time Jitter Analysis

• Timing Measurements:

• Data: Time Interval Error (TIE), also called phase jitter.
• Clock: Period, Cycle-to-cycle.

• Views:

• Eye diagrams (repetitive volts vs. time)
• Trend (time error vs. time)
• Histograms (hits vs. time error)
• Spectrum (time error vs. frequency)
• Phase noise
• Bathtub curves (BER vs. eye opening)

Advanced Instruments for Jitter Analysis

• Advanced instruments for jitter analysis such as DCA-J (Digital Communication Analyzer-Jitter ) and BERT (Bit Error Ratio Tester) support:

• Decomposition of jitter into Total Jitter (TJ), Random Jitter (RJ), Deterministic Jitter (DJ), Periodic Jitter (PJ), Data Dependent Jitter (DDJ), Duty Cycle Distortion (DCD), and Jitter induced by Inter symbol Interference (ISI)
• Jitter frequency spectrum

• If always the same edge of a bit pattern is observed, then all DDJ is removed. Only non-correlated jitter, RJ and non correlated PJ remains.

• If averaging is performed over the pattern (not the bits), then all non-correlated jitter: RJ and PJ is removed.

Single Edge

• Single Edge

• Histogram for RJPJ
• RJ, PJ causes σ
• DDJ causes

• Separate RJ, PJ by de-convolution

An entire picture from a Jitter Analyzer (Agilent)

A View of Real-Time Jitter Measurement

Inter-Symbol Interference (ISI): Caused by loss

Inter-Symbol Interference (ISI): Caused by reflection

Spread Spectrum Clock Measurement

Uncorrelated Periodic Jitter Coupling

How about A Gaussian Noise Modulating the Phase?

Duty-Cycle Distortion (DCD)

Inter-Symbol Interference (ISI)

Inter-Symbol Interference (ISI)

Uncorrelated Periodic Jitter Coupling