Fiber Characterization and Testing Long Haul, High Speed Fiber Optic Networks:
Chromatic Dispersion, Polarization Mode Dispersion and Spectral Attenuation
One
of the big advantages of fiber optics is its capability for long
distance high speed communications. Attenuation at long wavelengths is
low. Fibers can be fusion spliced with virtually no loss. High-powered
lasers and fiber amplifier regenerators mean long distances are easily
obtained.
However over very long distances, new factors in fiber
performance become important. Chromatic dispersion, the dispersion
caused by light of different wavelengths, and polarization mode
dispersion, caused by the polarization in fibers, become factors
limiting fiber links. Even the variation of fiber attenuation with
wavelength can become an issue. All 3 may need testing on long distance
networks to ensure proper link performance.
Chromatic Dispersion
Chromatic
dispersion (CD) is caused by the fact that singlemode glass fibers transmit
light of different wavelengths at different speeds. The ratio of the
speed of light in a medium to the speed in a vacuum defines the index
of refraction or refractive index of the material. For optical fiber,
the effective index of refraction is about 1.45, so the speed of light
in glass is about 2/3 the speed of light in a vacuum. But the index of
refraction, and thereby the speed of light in the fiber, is a function
of the wavelength of light, the principle we all know from seeing a prism break light into a spectrum.
Most
sources used in long distance
fiber optic links are lasers which have very little spectral width. And
fibers are optimized for the wavelength of use. Both these factors
minimize the effects of chromatic dispersion but cannot totally stop
it. As the pulse proceeds down the fiber, the light of longer
wavelength travels slightly faster and spreads the pulse out as
shown here.
What Causes Chromatic Dispersion
There are two factors that cause chromatic dispersion: material dispersion and waveguide dispersion.
Material
dispersion is caused by the variation of the index of refraction in a
given material, glass in this case, over wavelength. Looking at the
graph below, the variation of the index of refraction over the entire
spectrum covered by fiber optics may seem small, only a few percent,
but when you are dealing with very high speed pulses over very long
distances it can add up.
Waveguide Dispersion
Waveguide
dispersion is a bit more complex. In singlemode fiber, the wavelength
of the light is not that much bigger than the core of the fiber and
(we’ll leave out the complex physics) as a result the light traveling
down the fiber actually travels in an area that exceeds the diameter of
the core, which we call the “mode field diameter” of the fiber. The
mode field diameter is a function of the wavelength of the light, with
longer wavelengths traveling in a larger mode field diameter. Thus part
of the light is traveling in the geometric core of the fiber and part
is traveling in the cladding. Since the core is made of a higher index
of refraction glass than the cladding, the light in the cladding
travels faster than the light in the core. Longer wavelengths have
larger mode field diameters so they suffer more material dispersion.
Engineered Dispersion in Fibers
Material
and waveguide dispersion have opposite variations with wavelength, so
careful design of the fiber materials and index profiles allows the
fiber to have a “zero dispersion wavelength.” On either side of that
wavelength, dispersion increases. The
importance of chromatic dispersion is a function of the application for
the fiber. As a result, different SM fibers have been developed for the
requirements of specific applications. Here are a listing of the types
of fibers currently in use.
| Description |
ITU Spec. |
Application |
| Standard Singlemode Fiber |
G.652 |
Original SM fiber, optimized for 1310 nm, OK for use at 1550 |
| Low Water Peak Fiber |
G.652 |
Processed to reduce water absorption at 1400 nm for DWDM |
| Dispersion-Shifted Fiber |
G.653 |
Optimized for 1550 nm |
| Cutoff Shifted Fiber |
G.654 |
Optimized for low loss at 1500 to 1600 nm for long haul submarine cables |
| Non-Zero Dispersion-Shifted Fiber |
G.655 |
Optimized for 1550 nm, DWDM |
| Wideband Non-Zero Dispersion-Shifted Fiber |
G.656 |
Wideband, DWDM from 1460 to 1625 nm |
Dispersion Compensation
As
mentioned earlier, the dispersion characteristics of a fiber can be
manipulated by the materials and design of the fiber. In fact, fibers
can be made that have CD inverse to the typical fibers and of a greater
magnitude. So a length of dispersion compensation fiber can be added to
a link, usually at a repeater (optical amplifier) that reverses the CD
of the fiber span before it. Such fibers tend to have high loss and
bend sensitivity, so alternatively a dispersion compensator made from a
specialized component called a bragg grating can be used, but it has a
more limited use and higher cost.
Chromatic Dispersion in the Cable Plant
As
with any other component, optical fiber performance parameters can vary
from batch to batch, so a long concatenated cable plant with many
different fibers will have a end-to-end chromatic dispersion which is
an integration of the CD of all the individual fibers. Therefore fiber
in long distance links will probably be tested for CD after
installation or before upgrading a link to higher bit rate electronics.
Testing Chromatic Dispersion
There
are several methods used for testing CD in fibers. All involve testing
at a variety of wavelengths using several discrete sources of various
wavelengths, a tunable laser or a broadband source with a monochromator
in the receiver and measuring the relative speeds of the signals. The
data taken at discrete wavelengths is then analyzed to calculate the
dispersion in terms of ps/nm/km.
Test methods use phase delay or time
of flight and generally require access to both ends of the fiber as
well as a second fiber for synchronization of the two test instruments
at either end. However, an OTDR test method is also used where traces
are taken at several discrete wavelengths and CD can be calculated from
the data obtained from the traces, allowing testing in the field from
one end of the fiber.
All these methods have international standards for the test methods, instruments and data analysis.
| Standards |
Description |
| IEC 60793-1-42 |
Measurement methods and test procedures—chromatic dispersion |
| ITU-T G.650.1 |
Definitions and test methods for linear, deterministic attributes of
singlemode fiber and cable |
| TIA FOTP-175-B |
Chromatic dispersion measurement of single-mode optical fibers |
| GR-761-CORE |
Generic criteria for chromatic dispersion test sets |
Phase Shift Method
Pulse Delay Method
OTDR Test Method
Polarization Mode Dispersion
Polarization
mode dispersion (PMD) is a bit more complex. Polarization is a
phenomenon of light traveling in a medium as a wave with components at
right angles. Some materials, like a glass optical fiber, have a
different index of refraction for each of those components of the light
wave, which is called birefringence. And a different index of
refraction means light travels at a different speed, so in the simplest
visualization, PMD in fiber looks like the drawing below, where each
component of the polarized light travels at a different speed, causing
dispersion. The magnitude of PMD in a fiber is expressed as this
difference, which is known as the differential group delay (DGD) and
called Δτ (“delta Tau”).
PMD is caused by the birefringence of the fiber which can be influenced by two factors, material birefringence and waveguide birefringence, similar to CD, but more complex. Waveguide birefringence is caused by geometrical variations in the fiber such as concentricity or ellipticity. Material birefringence is mainly caused by stress on the fiber.
Waveguide Birefringence
Material Birefringence
PMD
is a complex issue in installed optical fiber. In a long concatenated
fiber, each fiber can have different waveguide and material birefringence
characteristics caused by the random characteristics of each fiber in
the link and variations of the stress on the fiber. Variations
are particularly noticeable in aerial fiber, where the PMD may vary
considerably according to temperature and wind speed buffeting the
fiber!
PMD causes pulse broadening and/or jitter in the received
electrical signal, potentially causing errors in the reception of the
signals. Since the PMD can vary over time, an extra margin of 1 to 3 dB
is often added to the power budget to accommodate variation in PMD.
PMD
is an important issue as data rates on long distance links increases to
40 Gb/s and 100 Gb/s. Unfortunately, there are no reliable compensation
schemes for PMD, so the only solution is to test links to be upgraded
for PMD using one or more of the standardized test methods.
Testing PMD
PMD
is generally tested for fibers during manufacture or when being cabled.
In the field, it is common to test PMD on newly installed fibers which
are intended for operation at high speeds, generally above 2.5 Gb/s or
when upgrading fibers installed some time in the past. Since PMD varies
over time, a single test becomes an average and tests at a later time
may be done for comparison.
There are a number of commonly used
test methods for PMD, some of which are limited to the manufacturing
environment, while others can be used in the field. Essentially, all
the test instruments have a source which can vary the polarization of
the test signal and a measurement unit that can analyze polarization
changes.
Here are descriptions of the methods and relevant standards.
| Description |
Test Method |
Standards |
| PMD for single-mode optical fibers by the Fixed Analyzer method |
Extrema Counting (EC)
Fourier Transform (FT) |
FOTP-113 |
| PMD measurement for single-mode optical fibers by Stokes parameter measurements |
Jones Matrix Eigen-analysis (JME)
Poincaré Sphere Analysis (PSA) |
FOTP-122 |
PMD measurement for single-mode optical fibers and cable assemblies by
Interferometry |
Traditional Interferometry(TINTY)
General Interferometry (GINTY) |
FOTP-124 |
| Guideline for PMD and DGD measurement in single-mode fiber optic components and devices |
|
FOTP-196 |
| Measurement methods and test procedures—polarization mode dispersion |
Fixed Analyzer measurement method (EC / FT)
Stokes evaluation method (JME / PSA)
Interferometry method (TINTY) |
IEC 60793-1-48 |
| Portable PMD test sets used for analyzing single-mode fiber |
|
GR-2947-CORE |
Definitions and test methods for statistical and nonlinear attributes of single-mode
fiber and cable |
Stokes parameter evaluation technique (JME & PSA)
State of Polarization method (SOP)
Interferometric methods ( TINTY & GINTY)
Fixed Analyzer technique (EC / FT / Cosine Fourier Analysis) |
ITU-T G.650.2 |
PMD
testing is not an easy, reproducible, accurate test. The measurement
uncertainty can be as high as 10-20%, as shown by testing done within
international standards committees. These committees have concluded
that all these measurement techniques are permissible, that there are
factors in making these measurements that are not well understood, and
the methods of data analysis are not without question.
All this
uncertainty of PMD measurements has the effect of making comparisons
between tests and test methods difficult. Variations are particularly
high on tests of older fiber links. Presentations of field have even
shown that variations in PMD can be correlated to wind speed for aerial
cables.
PMD testing needs to be done on long links but the data must be analyzed intelligently to be of real use.
Spectral Attenuation
With
the develoment of low water peak fibers, the possibility of
transmission from 1260 to 1675 nm has been considered. This results
from careful manufacturing of the fiber to reduce the water in the
fiber (in the form of OH- ions) that causes higher spectral attenuation
at around 1244 and 1383 nm.
Since one may want to use available fibers
of unknown spectral attenuation for CWDM which uses lasers from 1260 to 1670 nm in 20 nm windows, it becomes necessary to
test for spectral attenuation to verify the usability. At the water
peaks, legacy fibers may have attenuation coefficients around 2 dB/km
while low water peak fibers may be as low as 0.4 dB/km.
Testing
spectral attenuation is done per TIA/EIA-455-61 or IEC 61300-3-7 with broadband sources like LEDs and a
spectrum analyzer on the receiving end of the fiber. Calibration is
done with a short fiber length, the the instrument calculates the
spectral attenuation on a long length being tested. The measurement of
spectral attenuation uses instruments similar to those used for CD testing by the
phase shift method, so some instruments do both measurements at one
time.
The
FOA wishes to acknowledge the invaluable reference materials on these
subjects in the JDSU Reference Guides to Fiber Optic Testing. These highly recommended books can be downloaded from the JDSU website.
|
