Académique Documents
Professionnel Documents
Culture Documents
Technologies SONET, SDH, OTN. 100M, 1GbE (Copper), Infiniband, S-ATA, USB,
OC-48/192/768, 10GbE (Optical), PCI Express, FibreChannel,
STM-16/64/192, 3.125G XAUI electrical RapidIO, HyperTransport
OTU-1/2/3.
Measurements Unit Interval (UI) vs. Freq. Stressed Eye, Transmitter Eye Diagram, Bathtub,
"Sinusoidal" Test with Dispersion Penalty (TDP). Histogram, RJ/DJ. Jitter
Jitter Analyzers "Worst Case Statistical" specifications are evolving.
Test with BERTs and DCA Test with Oscilloscopes,
DCA and BERTs.
2 w w w. a g i l e n t . c o m / f i n d / j i t t e r
Data Rates <3.2 Gb/s Data Rates >3.2 Gb/s
Analysis Stimulus/Analysis Stimulus Analysis Stimulus/Analysis Stimulus
or
or
n
n
tion
tion
rat
rat
RT
T
tio
tio
ER
e
e
T
RT
ene
ene
BE
olu
olu
DC scop
cop
olu
olu
ER
lB
901 ial BE
niB tter S
niB tter S
ial
- 5 0 BERT
-50 BERT
eG
eG
rS
rS
lB
llos
ria
lo
itte
tte
ria
uls
uls
812 2 Se
812 2 Se
8 61 Oscil
sci
A
A
r
Ji
Ji
r
r
i
Se
Se
DC
ER
ER
4P
4P
Pa
Pa
0J
0J
O
00
00
/
/
33/
33/
906
901
906
8 50
8 50
00
00
50
50
-10
-10
8 61
Om
Om
811
811
N4
N4
N4
N4
JS
JS
JS
JS
54
54
Best Fit
Eye Diagram/Mask
Enterprise & High-Speed I/O
Good Fit
Bathtub Curve
Fair Fit
Extrapolated Bathtub
TIE Histogram
Parameter Histogram
Measurement vs. Time
FFT Spectrum
RJ/DJ Separation
DJ Views
Jitter Generation
SDH/OTN
SONET/
Jitter Transfer
Jitter Tolerance
Wander
3
SONET/SDH/OTN Jitter Measurements
4 w w w. a g i l e n t . c o m / f i n d / j i t t e r
Wander
Wander is the longer-term phase
variations ranging from 10 Hz down
to micro-Hertz and below. While
jitter is normally measured with
reference to a clock extracted from
the data signal, wander is measured
against an external reference clock.
The fundamental measurement is
Time Interval Error (TIE), the
instantaneous time deviation of the
clock signal under test relative to
the reference source. TIE samples
are used to calculate wander
Figure 4: Jitter tolerance plot on OmniBER OTN
results such as Maximum TIE (MTIE)
and Time Deviation (TDev). The
Tolerance requires a source of sinu- than the SONET/SDH mandated OmniBER OTN and OmniBER 718
soidal jitter and a method to assess 0.1dB of peaking. The Agilent JS-1000 make these wander measurements
BER. Because the OmniBER OTN and Performance Jitter Solution, in real-time (see Figure 6).
OmniBER 718 contain both, they are OmniBER OTN, and OmniBER 718 can
ideally suited for this measurement measure jitter transfer on electrical
(see Figure 4). Using a modulation and optical devices (see Figure 5).
source such as the Agilent 33250A
function/arbitrary waveform genera- Figure 5:
tor, the Agilent N4901/2A Serial BERTs Jitter transfer
and 81250 ParBERT can measure measurement plot
jitter tolerance on electrical devices. from JS-1000
Jitter transfer
Jitter transfer is typically used to
describe how a clock recovery module
or repeater locks and tracks data as
jitter is placed on it. Jitter transfer
requirements on clock recovery cir-
cuits specify a minimum amount of
jitter gain vs. frequency up to a given
cut-off frequency, beyond which the
jitter must be attenuated. The jitter
transfer specification is intended Figure 6:
to prevent the buildup of jitter in a TIE plot on
network consisting of cascaded OmniBER 718
regenerators. When measuring jitter
transfer, you need to be assured
your DUT does not transfer more
5
Enterprise and High-Speed I/O Jitter Measurements
BER Multiplier
10 --4 7.438
What is jitter? 10 --6 9.507 Deterministic jitter
Jitter is defined as the misalignment 10 --7 10.399 Deterministic jitter (DJ) is due to
of the significant edges in a sequence 10 --9 11.996 systematic events and is bounded in
of data bits from their ideal positions. 10 --11 13.412 amplitude. DJ is composed of periodic
Misalignments can result in data 10 --12 14.069 jitter (PJ), data dependent jitter
errors. Tracking these errors over an 10 --13 14.698 (DDJ), duty cycle distortion (DCD),
extended period of time determines 10 --15 15.883 and uncorrelated (to the data) bound-
the system stability. Jitter can be due ed jitter. PJ is typically sinusoidal in
Figure 7: Converting RMS to Peak-to-Peak nature, while uncorrelated bounded
to deterministic and random phenom-
for specific BER
ena, also referred to as systematic jitter is frequently caused by cross-
and nonsystematic respectively. talk. DDJ includes data smearing
value. This allows one to assess the components such as Inter Symbol
From a general standpoint, jitter probability that the jitter will fall both Interference (ISI). ISI is due to band-
characterization involves a statistical inside and outside a specified range. width limitations of the system. DCD
measurement of the relative position When an RMS value is measured, the is due to voltage offsets between
variance of clock or data edges. Serial peak-to-peak RJ value can be estimat- differential inputs and system rise/fall
data communication eliminates the ed through multiplication by a factor time differences. Deterministic jitter
physical and bandwidth limitations of dependent upon the desired bit error is bounded and therefore measured
clock and data bus transmission by ratio (BER) probability (see Figure 7). as a peak-to-peak value.
embedding the clock in the transmit- Note that these multipliers are valid
ted data. Since the data clock is not for purely random jitter. When DJ is Total jitter
transmitted separately, the problem present, the BER margin is reduced
of maintaining the clock and data Total jitter (TJ) is the sum of the
by the DJ value. A significant peak-to-peak values of deterministic
temporal alignment is eliminated in
amount of data is required to yield jitter (DJ) and random jitter (RJ).
the data stream. However, other
a statistically accurate, high confi- By defining random jitter as an
issues become important, such as the
dence characterization of RJ. As equivalent peak-to-peak value at a
minimization of jitter in data trans-
seemingly small amounts of RMS RJ given probability, total jitter can
mission, faithful clock extraction from
correspond to large peak-to-peak always be expressed as a sum.
the serial data, and network timing.
values, precisely assessing the full Figure 8 illustrates the components
These problems are manifest because
impact of the jitter is achieved by of total jitter, each of which presents
of random and systematic effects.
directly measuring the probability itself in a different way. It can be
of RJ at the desired low BERs.
Random jitter difficult to separate deterministic
and random jitter from the measure-
Random jitter (RJ) is Gaussian in
ment of total jitter and not all test
nature and typically results from
solutions offer this capability. When
thermal noise, shot noise, etc. In that
characterizing your design, the key
RJ is unbounded, it is often charac- Total Jitter to success is to understand what
terized through an RMS or mean
you need to test for and optimizing
a solution that works for you.
Random Deterministic
Jitter Jitter
Figure 8:
Inter Symbol The components of jitter
Interference
6 w w w. a g i l e n t . c o m / f i n d / j i t t e r
What jitter measurement viewpoints
should you use?
Given the various jitter components,
understanding what you need to
measure is critical to understanding
what is the best test solution for you.
The intent of a jitter compliance
measurement is to ensure meeting a
certain bit error ratio (BER) perform-
ance specification. For low BERs
(i.e. 10 -12 and lower), empirical tests
can be long in time duration to cap-
ture enough data to confirm the BER
specification conformance with suffi-
cient statistical confidence. For design
optimization and troubleshooting,
post-processed jitter calculations can Figure 9: Eye diagram with the 54850 Infiniium oscilloscope
quickly provide insight.
7
Figure 10: Eye diagram with the 86100 Infiniium DCA
33250A
The eye diagram is useful for certifi- To create a jittered test signal,
cation and validation, development pair the oscilloscope with an
and manufacturing. For development Agilent 81133/4A Pulse/Pattern
engineers, the eye diagram is the most Generator to add jitter to
intuitive measurement. It is easy to clock and data signals up
set up and provides an immediate to 3.35 GHz and control Device
view of amplitude versus time. Blatant the eye closing over the time Under
irregularities are easily identified for axis. Real world, or worst case, Test
further development, however it is signals can be simulated with
hard to precisely correlate the cause the 81133/4A using the 81134A
of DDJ, PJ and ISI with this view Agilent 33250A function/
alone. In manufacturing, the eye dia- arbitrary waveform generator
gram provides a quick measurement as a modulation source. The
for confidence, but it does not com- Agilent N4901/2A 13.5/7.0 Gb/s
pletely guarantee jitter performance. Serial BERTs and 81250 ParBERT
sources can also be jittered with
an external modulation source for
device characterization. 86100B Infiniium DCA
Figure 11:
Stressed eye generation using the
81134A pulse generator
8 w w w. a g i l e n t . c o m / f i n d / j i t t e r
However the speed of the BER meas- and 10- 9 ) by adjusting the sampling
urement is a function of the BER to point in time, and then extrapolates
be measured and the desired statisti- the jitter at much lower BERs
cal confidence factor for the result. (i.e. 10- 12 ). The 81250 ParBERT
Various extrapolation techniques can and N4901/2/6A Serial BERTs have
be used to speed up the BER meas- accelerated extrapolation techniques
urement, one of which is based on to measure the bathtub curve,
the bathtub curve called BERT scan. extract RJ and DJ components, and
10-5
Deterministic
Deterministic
10-7
at several quick BERs (i.e. 10- 5, 10- 7,
Random
Random
10-12 Calculated
DJ DJ
0 TL 0.5T B TR TB
Bathtub curve
Figure 13:
Plotting the points on a graph of BER Bathtub curve
versus sampling location in time from the N4906A
Serial BERT
results in a curve that resembles the
cross section of a bathtub, hence the
term bathtub curve (see Figure 12).
A bathtub curve can be created
either by software extrapolation from
voltage over time measurement done
by an oscilloscope, or by actually
sampling the bits as done by a BERT.
Any instrument capable of performing
a BER measurement can be used
to create a bathtub curve, as long as
the sampling point can be varied in
time. The Agilent N4901/2/6A
13.5/7.0/3.6 Gb/s Serial BERTs
(see Figure 13) and 81250 ParBERT
(see Figure 14) can be used to create
the bathtub curve.
9
Histogram
Figure 16:
A histogram is a portrayal of the TIE histogram from
54850 Infiniium
relative occurrence of measured
6 GHz oscilloscope
values. Depending on the actual
jitter distribution, the histogram
shows deterministic jitter compo-
nents (time interval error) and
Figure 15:
spread. This is a key measurement Histogram from
in meeting compliance specifications. 86100 Infiniium
digital communica-
Agilents 86100B Infiniium tions analyzer
10 w w w. a g i l e n t . c o m / f i n d / j i t t e r
FFT spectrum
Any measurement vs. time can also For development engineers, this per- result of the circuit and not resulting
be processed to the frequency domain spective provides the highest level of from the probes. The Agilent N4901/2A
through an FFT conversion. Viewing insight into deterministic, periodic, Serial BERTs and 81250 ParBERT
the jitter spectrum that results can and data dependent jitter mecha- also offer a spectral decomposition
yield swift conclusions of interfering nisms. Agilents 54850 oscilloscope capability (see Figure 19).
mechanisms than cannot be readily with the E2681A jitter analysis soft-
seen in the time domain. Effects that ware provides a FFT spectrum (see
RJ/DJ separation
have periodicity in the time domain Figure 18). The 54850 oscilloscope
result in easily discernable spectral when used with the InfiniiMax prob- Most standards require the decom-
components in the frequency domain. ing system assures that the jitter is a position of total jitter into the con-
stituent components random jitter
(RJ) and deterministic jitter (DJ).
This separation can be achieved in
two ways: extrapolation from a bath-
Clock Jitter tub curve, or by analysis of the FFT
Signal Spectrum
spectrum. Using the bathtub curve
jitter model, the graph can be curve-
fitted to perform a best fit in order
Deter-
ministic to report the RJ and DJ values. The
Jitter Averaged
Spectrum DJ forms the horizontal wall portion
of the bathtub curve, while the bath-
tub curves sloped portion is due to
RJ (see Figure 12, pg. 9).
Figure 19:
Spectral
decomposition
view on
81250 ParBERT
11
Figure 20:
Histogram plot
For the certification engineer, this
and RJ/DJ separation
view is crucial to proving compliance. values on 81250
For the development engineer, the ParBERT
measurement is a tool to quantify the
effects of environmental changes on
RJ and DJ to circuit performance.
Agilents 54850 Infiniium oscilloscope
with the E2681A jitter analysis soft-
ware can be used to identify DJ in
the frequency domain (see Figure 18,
pg. 11). The 81250 ParBERT (see Figure 21:
Figure 20) and N4901/2/6A Serial RJ/DJ separation
BERTs (see Figure 21) separate RJ view on N4906A
Serial BERT
and DJ via extrapolation from the
bathtub curve model.
DJ views
12 w w w. a g i l e n t . c o m / f i n d / j i t t e r
Jitter Measurement Solutions
View optical and electrical waveforms The Agilent 54850 Series Infiniium The 81133A (1 channel) and the
with bandwidths to 80 GHz electrical oscilloscopes and the InfiniiMax 1130 81134A (2 channels) are high per-
and 65 GHz optical. Built-in measure- Series probing system deliver the formance sources for square waves,
ments for high-speed digital commu- highest performance real-time end-to- pulses, data patterns and pseudo ran-
nications and module configurations end measurement system available. dom data sequences up to 3.35 GHz.
for a test system to match your spe- With its 20 GSa/s sample rate per The hardware PRBS generation from
cific needs. View the true jitter of channel, the Infiniium 54850 Series 2 5-1 to 231 -1 and 8 kB standard/12 MB
your device with an ultra-low, intrin- is an ideal tool for making signal extended pattern memory make the
sic trigger jitter of less than 200 fs. integrity and jitter measurements for 81133A/81134A an ideal stimulus for
the following standards: PCI Express, jitter tolerance, jitter transfer and
Typically used to view an eye dia- Infiniband, HyperTransport, RapidIO, stressed eye measurements. Timing
gram, the magnitude of the jitter can Serial ATA, AGP and other similar and levels can be individually pro-
be immediately viewed in one simple high speed signaling standards. grammed. The fast rise times of less
display. Estimate random and deter- than 60 ps and low jitter of 1.5 ps
ministic jitter components using the Key features/benefits typical ensure the highest signal
built-in histogram capabilities. integrity when timing is critical.
6 GHz, 4 GHz, and 2.5 GHz band-
width real-time oscilloscopes with For stress tests, stressed eye meas-
Key features/benefits
20 GSa/s sample rate on all four urements, jitter tolerance and jitter
The eye diagram provides the most channels simultaneously transfer tests the amount of jitter
common and easy-to-understand
Up to 1 Mpts MegaZoom deep (random, deterministic and combina-
display of a waveform tions of these types) on the data and
memory at all sample rates and
Ultra low instrumentation jitter 32 Mpts MegaZoom deep memory clock signals can be controlled by an
allows measurement of extremely at 2 GSa/s and slower sample rates external modulation source.
low jitter on signals, critical for
Trigger jitter as low as 1.0 ps RMS
high-performance applications Key features/benefits
and 1.4 ps RMS jitter noise floor
Comprehensive waveform for accurate jitter measurements Delay Control Input (Jitter Control)
characterization is possible Direct Mode (instrument follows
E2681A jitter analysis software
83495A 9.95-11.3 Gb/s Clock with jitter set up wizard external frequency exactly)
Recovery Module with selectable Clock, data and PRBS signals
InfiniiMax 7 GHz, 5 GHz, and
loop BW
3.5 GHz probing systems that For more information order the
Time Domain Reflectometer support differential and single- following literature:
modules available to further char- ended measurements for a more Product Overview 5988-5549EN
acterize high-speed data channels cost-effective solution
For more information order the For more information order the
following literature: following literature:
86100B Technical Specifications Data Sheet 5988-7976EN
5988-5311EN
83495A Technical Specifications
5988-9827EN
13
Jitter Measurement Solutions
The 81250 ParBERT provides parallel Operating over the data range of The N4901/2A Serial BERTs are the
BER testing up to 45 Gb/s for high- 50 Mb/s to 3.6 Gb/s, this integrated ideal solution for the research, devel-
speed digital components. The modu- pattern generator and error detector opment and manufacturing test of
lar design lets you mix and match is designed to quickly solve your digital components; devices and sub-
analyzer channels, generator chan- design and manufacturing problems. systems from 150 Mb/s to 13.5 Gb/s
nels and speed classes (333 kHz-675 The N4906A brings a new level of (N4901A) or 7.0 Gb/s (N4902A).
MHz, 333 Mb/s-1.65 Gb/s or 2.7 Gb/s, insight into your testing toolbox. Bathtub jitter analysis is built-in and
21 Mb/s-3.35 Gb/s, 620 Mb/s-7 Gb/s, Easy to set up and use, in a small, provides insight into the underlying
620 Mb/s-13.5 Gb/s and 38-45 Gb/s). expandable package makes digital causes behind bit errors.
The 81250 ParBERT provides the testing easy. The instrument includes
bathtub jitter analysis built-in that With this high performance serial
ability to generate jitter (3.35 Gb/s,
provides insight into the underlying pattern generator and error detector,
7 Gb/s and 13.5 Gb/s modules only)
causes behind bit errors. you can perform error analysis to
and analyze jitter. The low intrinsic
verify the operation and quality of
jitter of the 3.35G/7G/13.5 Gb/s
receivers enable jitter analysis of a Key features/benefits lightwave submarine cable systems,
SONET/SDH telecom and high-speed
DUTs output. The 3.35G/7G/13.5 Gb/s Up to 3.6 Gb/s measurements
generators are capable of stressing I/O transceivers, 10 Gigabit Ethernet,
Easy to learn, easy to use and high-speed I/O serial links.
the DUT by injecting RJ and DJ with
a per channel delay control input World class waveform
(driven by an external modulation Key features/benefits
signal). Q-factor measurement
Up to 7.0 or 13.5 Gb/s measurements
Up to 10 ns clock/data delay range
Key features/benefits Easy to learn, easy to use
Bathtub Jitter Analysis
Generate custom patterns, pseudo World class waveform
random word sequences (PRWS) For more information order the
Jitter analysis (extract and
and standard PRBS up to 2 31 1 following literature:
extrapolation) with bathtub curve
Jitter analysis (extract and extrap- Product Brochure and spectral decomposition
olation) with bathtub curve and 5988-8918EN
spectral decomposition External delay control input
Technical Specifications
(jitter modulation) up to 1 GHz
External delay control input 5988-9276EN
with 100 ps delay range
(jitter modulation) on the
3.35G/7G/13.5 Gb/s generators For more information order the
(up to 200 MHz with 250 ps delay following literature:
range for 3.35 Gb/s, up to 1 GHz N4901A Technical Specifications
with 100 ps delay range for 5988-9674EN
7G/13.5 Gb/s)
N4902A Technical Specifications
For more information order the 5988-9675EN
following literature:
Product Overview 5968-9188E
14 w w w. a g i l e n t . c o m / f i n d / j i t t e r
JS-1000 JS-500 Clock Jitter
Performance Jitter Solution Measurement System
15
Jitter Measurement Solutions
16 w w w. a g i l e n t . c o m / f i n d / j i t t e r
Figure 24: Agilent jitter solutions for enterprise and high-speed I/O applications
<3.2 Gb/s N4906A Serial BERT N4906A Serial BERT 81250 ParBERT
JS-500
JS-1000
Related literature
Jitter Measurement Solutions CD 5988-9350EN
Measuring Jitter in Digital Systems App. Note 1448-1 5988-9109EN
Finding Sources of Jitter with Real-Time Jitter Analysis App. Note 1448-2 5988-9740EN
Jitter Analysis Techniques for High Data Rates App. Note 1432 5988-8425EN
Loop Bandwidth and Clock Data Recovery in Oscilloscope Measurements App. Note 1304-6 5988-6682EN
Generating/Measuring Jitter with the Agilent 81134A Product Note 5988-9411EN
Pulse/Pattern Generator and 54855A Infiniium Scope
Understanding Jitter and Wander Measurements and Standards Technical Note 5988-6254EN
17
Agilent Technologies Test and Measurement Support,
Services, and Assistance
www.agilent.com/find/emailupdates
Agilent Technologies aims to maximize the value you receive, while
minimizing your risk and problems. We strive to ensure that you get Get the latest information on the products
the test and measurement capabilities you paid for and obtain the and applications you select.
support you need. Our extensive support resources and services can
help you choose the right Agilent products for your applications and Agilent T&M software and connectivity
apply them successfully. Every instrument and system we sell has a
global warranty. Support is available for at least five years beyond Agilent's Test and Measurement software and connectivity products,
the production life of the product. Two concepts underlie Agilent's solutions and developer network allows you to take time out of
overall support policy: "Our Promise" and "Your Advantage." connecting your instruments to your computer with tools based on PC
standards, so you can focus on your tasks, not on your connections.
Our Promise
Our Promise means your Agilent test and measurement equipment Visit www.agilent.com/find/connectivity
will meet its advertised performance and functionality. When you are for more information.
choosing new equipment, we will help you with product information,
including realistic performance specifications and practical recom-
mendations from experienced test engineers. When you use Agilent For more assistance with your test & measurement needs or to find
equipment, we can verify that it works properly, help with product your local Agilent office go to www.agilent.com/find/assist
operation, and provide basic measurement assistance for the use of
specified capabilities, at no extra cost upon request. Many self-help Product specifications and descriptions in this document
tools are available.
subject to change without notice.
Your Advantage
Your Advantage means that Agilent offers a wide range of additional
expert test and measurement services, which you can purchase
according to your unique technical and business needs. Solve problems
efficiently and gain a competitive edge by contracting with us for
calibration, extra-cost upgrades, out-of-warranty repairs, and on-site
education and training, as well as design, system integration, project
management, and other professional engineering services.
Experienced Agilent engineers and technicians worldwide can help Agilent Technologies, Inc. 2003
you maximize your productivity, optimize the return on investment Printed in USA, September 9, 2003
of your Agilent instruments and systems, and obtain dependable
measurement accuracy for the life of those products. 5988-9592EN
or
n
tion
n
tion
rat
rat
rBE BERT
T
tio
tio
ER
e
e
T
RT
ene
ene
olu
olu
DC scop
cop
olu
olu
ER
lB
901 ial BE
niB tter S
niB tter S
l
RT
-50 BERT
eG
eG
rS
rS
lB
los
ria
ria
lo
itte
itte
ria
uls
uls
812 2 Se
812 2 Se
8 61 Oscil
8 61 Oscil
A
A
r
Ji
Ji
r
Se
Se
DC
ER
ER
4P
4P
Pa
Pa
0J
0J
00
00
/
/
33/
33/
906
901
906
8 50
8 50
00
-50
00
50
50
-10
-10
Om
Om
811
811
N4
N4
N4
N4
JS
JS
JS
JS
54
54
Best Fit
Eye Diagram/Mask
Enterprise & High-Speed I/O
Good Fit
Bathtub Curve
Fair Fit
Extrapolated Bathtub
TIE Histogram
Parameter Histogram
Measurement vs. Time
FFT Spectrum
RJ/DJ Separation
DJ Views
Jitter Generation
SDH/OTN
SONET/
Jitter Transfer
Jitter Tolerance
Wander