FOA Logo


Fiber Characterization Testing For Long Haul, High Speed Fiber Optic Networks:

Chromatic Dispersion, Polarization Mode Dispersion and Spectral Attenuation

Objectives: From this page you should learn:
What Fiber Characterization involves
What are Chromatic Dispersion (CD) and Polarization Mode Dispersion (PMD)
How CD and PMD affect high speed long distance transmission
What is Spectral Attenuation (SA)
How SA affects wavelength-division multiplexing
How one tests CD, PMD and SA


Note: It is recommended that techs learning about fiber characterization for field operations have an extensive knowledge of fiber optics and especially fiber optic testing. Managers should have a knowledge of basic fiber optics and testing.

Introduction
One of the big advantages of fiber optics is its capability for long distance high-speed communications. Singlemode fiber attenuation at long wavelengths (~1550 nm) is extremely low. Fibers can be fusion spliced with virtually no loss. High-powered lasers, sophisticated transmission protocols and fiber amplifier regenerators mean long distances are easily obtained. Dense wavelength division multiplexing (DWDM) allows up to 128 channels of signals on a single fiber.

However, for high-speed networks operating over very long distances, new factors limiting fiber performance become important. Chromatic dispersion, the dispersion caused by light of different wavelengths, and polarization mode dispersion, caused by the polarization of the light in the fiber, become factors limiting the bandwidth capacity of fiber links. Pulse broadening due to chromatic dispersion and the variation of fiber attenuation with wavelength can become issues with DWDM.

All these factors need testing on long distance networks to ensure proper link performance. Tests are performed on new installations to ensure the fiber being installed is capable of future upgrades. Older cable plants are tested to evaluate fibers for upgrades of legacy communications systems at slower speeds.

A suite of tests for these factors has been developed to test fibers for long distance high-speed networks. These tests are normally called “fiber characterization,” but technically they are “fiber optic cable plant characterization” since it must include the complete end-to-end cable plant.

Tests For Fiber Characterization

Test
Reason For Test
Connector Inspection.  (Review)  Verify quality and condition of connectors
Insertion Loss.  (Review)  Compare cable plant loss to loss budget and network power budget
Reflectance/ORL. (Review)
Look for locations of reflectance problems
OTDR. (Review) Check splices, connectors, fiber attenuation, look for stress induced losses from installation
Spectral Attenuation (SA)
Wavelength division multiplexing uses fiber over a large range of wavelengths
Chromatic Dispersion (CD)
Long distances at high speeds (>2.5Gb/s) may suffer dispersion
Polarization Mode Dispersion (PMD)
Long distances at high speeds (>2.5Gb/s) may suffer dispersion

These are NOT the only important tests for long links, they are in addition to the traditional cable plant tests: careful inspection of connectors and the installed cable plant (neatness and lack of stress in cables and patchcords), insertion loss testing with a test source and power meter or optical loss test set (OLTS) and optical time domain reflectometer (OTDR) testing. You can review those tests in the links above or the testing pages in the FOA Guide.

Note: Many long haul high speed (>100 Gb/s) fiber optic networks now use coherent transmission. Coherent network transceivers are very complex and can compensate for some of the problems tested by fiber characterization. If the networks you are testing are inteneded for coherent transmission, refer to manufacturer's specifications for required testing of the cable plant.


Background Information
There is some background information you should have before trying to understand the material in this page. Here are links to several other pages that you may find helpful.

Optical Fiber.

Singlemode Fiber Types

Wavelength Bands.  

Wavelength Division Multiplexing

Dense wavelength division multiplexing (DWDM) originally used optical signals multiplexed within the 1550 nm band compatible with erbium doped fiber amplifiers (EDFAs), which are effective for wavelengths between approximately 1525–1565 nm (C band), or 1570–1610 nm (L band). Dense wavelength division multiplexing (DWDM) channel plans vary, but a typical system might use 40 channels at 100 GHz spacing or 80 channels with 50 GHz spacing.


WDM bands
Wavelength division multiplexing in the transmission bands

 

Coarse wavelength division multiplexing (CWDM), a lower cost version of WDM, uses lasers from 1260 to 1670 nm in 20 nm windows. This allows less expensive lasers but does require verification of the spectral attenuation of the fiber over a longer range of wavelengths.

 


Spectral Attenuation

With the development of low water peak fibers, the possibility of transmission from 1260 to 1675 nm has been realized. 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 DWDM or CWDM that use spectrum all the way from 1260 to 1670 nm, it becomes necessary to test for spectral attenuation to verify the usability. At the water peaks (1244 and 1383 nm – there is also a peak out of band at 950 nm), legacy fibers may have attenuation coefficients around 2 dB/km while low water peak fibers may be as low as 0.4 dB/km.

low water peak fiber

Spectral attenuation of regular and low water peak singlemode fibers


Typical Fiber Attenuation Over Wavelength Range (Corning SMF 28)

Wavelength (nm)

Maximum Loss (dB/km)

1310

0.33 – 0.35

1383

0.31 – 0.35

1490

0.21 – 0.24

1550

0.19 – 0.20

1625

0.20 – 0.23


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.

Spectral attenuation should be tested in the wavelength ranges (bands) of interest. DWDM systems focus mainly on the C-band, with some also operating the S- and L-bands. CWDM systems operate in the entire singlemode wavelength range, 1260 to 1625 nm.

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.

Spectral attenuation test

Measuring spectral attenuation with wideband spectral source


Spectral attenuation test

Actual data from spectral attenuation test




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.


Relative spectral width of F-P and DFB lasers

lasers


Most sources used in long distance fiber optic links are lasers that 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. Singlemode fiber optic links use lasers for transmitters. Slower systems use Fabry-Perot (F-P) lasers that have a relatively narrow spectral width (range of wavelengths in the source). Faster systems and DWDM systems use distributed feedback (DFB) lasers that have spectral widths less than 1/10th as large as F-P lasers.




Chromatic Dispersion

Chromatic dispersion in fiber caused by longer wavelengths traveling faster


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.


Chromatic Dispersion - material dispersion

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.

Chromatic Dispersion  - waveguide 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.

engineered dispersion in optical fiber
Engineered dispersion in optical fiber. The fiber dispersion (blue) is a combination of material dispersion (red) and waveguide dispersion (yellow)


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


CD by fiber type

A graph of their typical CD as a function of wavelength

For each fiber, the specifications for CD are the absolute dispersion and for some fibers like G.652, the slope of the dispersion curve at the zero dispersion wavelength (lo). G.652 fiber is widely used for short and medium length networks. This fiber can be used for transmission at both 1310 nm and 1550 nm, so it is specified for CD at both wavelengths. In fact, the zero dispersion wavelength is around 1310 nm so chromatic dispersion is low, but at 1550 nm, the CD is higher, but acceptable for many applications.
 
The units for specifying chromatic dispersion are the actual dispersion, in picoseconds (ps) per nanometer of spectral width (nm) per kilometer of fiber, or
 
CD = ps/nm-km)
 
In addition, fibers that may be used over a broad range of wavelengths like G.652 will have a specification for the slope of the CD curve over wavelength at the zero dispersion point (λ0). This slope is the derivative of the curve at λ0, and the specification is expressed in picoseconds per nm2 per kilometer:

 Slopeλ0  = ps/nm2-km

slope at l0
 

CD specifications for typical fibers

Fiber

λ0 (nm)

Slopeλ0

(ps/nm2-km)

Dλ (ps/nm-km)

G.652

1300-1324

0.092

17 @ 1550

G.653.A

1500-1600

0.085

3.5

G.655.A-C

1530-1565

NS

 >1, <10*

G.655.D

1530-1565

NS

 >1.2, <7.2*

G.655.E

1530-1565

NS

 >4.8, <10.1*

G.656

1460-1550 1550-1625

NS

 >2, <14*

*Approximate, calculated function of wavelength (λ), must be >1 but <10 or as noted


CD "Budgets"

You can calculate an estimate of te CD of a cable plant, call it a "CD Budget." Simply multiple the chromatic dispersion Dλ of the fiber by the length just like calculating a loss budget. This will give you an idea of what to expect when testing, as well as compare to the requirements of the potential usage for higher speed networks. All high speed networks will have a specification for maximum tolerable CD.  

Calculate based on “Dλ” and fiber length, e.g.

50km G.652 fiber = 50km X 17ps/nm-km = 850ps/nm

100km G.653.A fiber = 100km X 3.5ps/nm-km = 350ps/nm

 If testing is being done for a system upgrade, there may be a CD specification for the intended system type and speed. Below are some maximum CD specifications for some typical networks.

Typical System Specifications For Maximum Chromatic Dispersion

Bit Rate (Gb/s)

System

Max CD (ps/nm)@1550nm

Maximum

(km) G.652

Distance

(km)G.655

2.5

SDH STM-16

SONET OC-48

18817

1100

4700

10

SDH STM-64

SONET OC-192

1176

70

290

10

Ethernet

738

44

180

40

SDH STM-256

SONET OC-768

73.5

5

20

Note the table stops at 40Gb/s. There are 100Gb/s systems and even faster being used or in field trials today. Some of those are WDM, e.g. 100 Gb/s with 10X10 Gb/s or 4X25Gb/s and some long distance ones that use coherent transmission. The best way to determine system limits is to ask the manufacturer of any specific system of interest.

The Telcordia/Bellcore standard GR-253-CORE provides another method of estimating CD effects. The standard specifies that for a 1 dB performance penalty, the total dispersion should be less than 0.306 times one bit period. This can be expressed as:

Dλ  (fiber dispersion) X L(length) X B (bit rate) X Dl(source spectral width) < 0.306

Thus for a G.652 fiber with a Dλ =17 ps/nm-km, B=10Gb/s and  Dl=0.1nm, the maximum length would be:

(17 X 10-12) L (10 x 10+9) (0.1) <0.306

17 x 10-3 L <0.306

.017 L <0.306

L < 18 km

Likewise you could insert a given length (for example on a cable plant being tested) and projected bit rate and calculate a maximum Dλ for a given fiber. Or you can solve for a given bitrate and calculate a maximum CD:  Dλ x L.




Chromatic 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.

dispersion compensation

Dispersion compensating fibers reverse normal CD at a high rate to undo dispersion with a short length of fiber


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


Chromatic Dispersion  - phase test

Phase Shift Method



Chromatic Dispersion - pulse delay test

 Pulse Delay Method




Chromatic Dispersion  - OTDR Test
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”).


Polarization Mode Dispersion


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.



PMD - Waveguide Birefringence

Waveguide Birefringence



PMD - Material 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.

 

Fiber Specifications For PMD

Unlike most fiber specifications, PMD is not a concrete specification, but is tested and specified on a statistical basis for cabled fiber. Cabled fiber, of course, has the stress induced due to the cabling process. Manufacturers of transmission systems also specify PMD for a link on a statistical basis.

 

From ITU G.652 Standard: “Cabled fibre polarization mode dispersion shall be specified on a statistical basis, not on an individual fibre basis. The requirements pertain only to the aspect of the link calculated from cable information.”

“The manufacturer (of transmission equipment) shall supply a PMD link design value, PMDQ, that serves as a statistical upper bound for the PMD coefficient of the concatenated optical fibre cables within a defined possible link of M cable sections. The upper bound is defined in terms of a small probability level, Q, which is the probability that a concatenated PMD coefficient value exceeds PMDQ. “

 

Typical Fiber Specifications For PMD

Fiber Type

PMDQ 

G.652A/C

0.50 ps/√km

G.652B/D

0.20 ps/√km

G.653A

0.50 ps/√km

G.653B

0.20 ps/√km

G.655C/D/E

0.20 ps/√km

G.656

0.20 ps/√km

G.657A

0.20 ps/√km

G.657B

0.50 ps/√km

 

From TIA FOTP-122 (an adaptation of IEC 61282-9):

“In long fibre spans, DGD (differential group delay caused by birefringence) is random in both time and wavelength since it depends on the details of the birefringence along the entire fibre length. It is also sensitive to time-dependent temperature and mechanical perturbations on the fibre. For this reason, a useful way to characterize PMD in long fibres is in terms of the expected value, <∆τ> (ps/√km), or the mean DGD over wavelength. In principle, the expected value <∆τ> does not undergo large changes for a given fibre from day to day or from source to source, unlike the parameters δτ or ∆τ. In addition, <∆τ> is a useful predictor of lightwave system performance.

 

Table from the ITU G.652/G.653 fiber specs for PMD limits

Max PMD

Link Length

Max Fiber DGD

Bit Rate

Not Specified

 

 

Up to 2.5Gb/s

0.5 ps/√km 

400

25 ps

10Gb/s

 

40

19 ps

10Gb/s*

 

2

7.5 ps

40Gb/s

0.20 ps/√km

3000

19 ps

10Gb/s

 

80

7.0

40Gb/s

0.10 ps/√km

>4000

12 ps

10Gb/s

 

400

5 ps

40Gb/s

* Also applies to 10Gb/s Ethernet systems

 

The table below shows examples of the distance limitations for some typical systems. Here is how it is calculated.

The bit period is calculated from the bit rate – it is the period from bit to bit, so 1Gb//s means 1 bit per nanosecond, 10 Gb/s means 1 bit per 100nanoseconds, etc.

The Max Mean DGD (Δτ) is calculated by a simple estimate – it should be no larger than 1/10 of the bit rate to not cause problems.

The DGD varies by the square root of the length so, for 2.5Gb/s, the bit period is 400ps, the max DGD is 40ps (1/10 of 400ns), and for 400km, the √400=20 so we calculate 40ps/20=2ps/√km.

Note the sensitivity of networks to PMD rises linearly with bit rate and with the square root of length. Ethernet has a lower tolerance to PMD than SDH/SONET because it has a different error correction scheme and a BER requirement that is more stringent than SDH/SONET. The numbers in the table below are estimated as the requirements are still under study.

  Estimated System Specifications For PMD

Bit Rate (Gb/s)

System

Bit Period (ps)

Max Mean DGD (Δτmax) (ps)

PMD Coeff (ps√km) @ 400km

2.5

SDH STM-16 SONET OC-48

400

40

<2

10

SDH STM-64 SONET OC-192

100

10

<0.5

10

10G Ethernet*

100

5

---

40

SDH STM-256 SONET OC-768

25

2.5

<0.125

Here is a graphical way of looking at the data on bitrate, PMD and distance. This helps visualize the approximate PMD coefficient for each network at various distances. Engineers designing these systems will look at each network separately, including other factors like the data encoding formats and fiber specs. Consider this and the previous table approximations only.

 PMD distance



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, 100 Gb/s and above. Unfortunately, there appear to be 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.
 
Expected Test Results (PMD Budget)

Here is a table based on the earlier table from the ITU G.652/G.653 fiber specs for PMD limits. Instead we show how to calculate the DGD from the average PMD specification for a given fiber length. This provides a reference value for each link when testing in the field.

PMD (ps/√km) X √length (km)  = DGD (ps)

Calculating DGD From Average PMD

Max PMD

Link Length (km)

√km

DGD (ps)

 

0.5 ps/√km 

400

20

10ps

40

6.3

3.2ps

2

1.4

0.7ps

0.20 ps/√km

3000

54.8

11ps

80

8.9

1.8ps

0.10 ps/√km

4000

63.2

6.3ps

400

20

2ps




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.

PMD testing


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

Whatever method the test equipment uses, the data will probably be presented as DGD in ps with its wavelength dependence, perhaps as a graphical display like a spectrum.


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 and vehicles or trains passing near underground cables. Tests are usually repeated over longer periods and the results analyzed.

PMD testing needs to be done on long links but the data must be analyzed intelligently to be of real use.


PMD Compensation

Compensation for PMD has been studied and discussed for years but no simple, reliable compensation method has been offered.


Fiber Characterization Test Equipment

There are a number of test sets available for lab and field testing of the parameters we combine in fiber characterization. For field testing, test sets may combine several of these tests into one instrument. Some require remote instruments for operation and some, primarily those based on OTDR test techniques, offer single-ended testing.

Due to the complexity of these tests and the instruments involved, as well as the evolution of the market to provide tests for increasingly higher bitrate systems, it is beyond the scope of this chapter to cover these instruments in the depth of the coverage of OLTS and OTDRs, for example.

As with all test equipment, it is important to first understand the tests involved and then to choose appropriate test equipment. Once equipment is chosen the operators should get trained on using that instrument so they know how to use it to perform the tests correctly and how to interpret the data it provides in the context of those tests.

 

Documentation

 


Like every step of the fiber optic design, installation and operation processes, it is important to fully document the test and record all relevant data. All tests should be recorded with the following data:

 

For all tests performed at one time:

-   Date of the test

-   Location of the test

-   Environmental conditions (temp/humidity/local conditions)

-   Cable plant identification (cable type/fiber type/connector type/length)

-   Type of test (visual inspection, insertion loss, OTDR, CD, PMD, SA)

-   Test equipment used (type, brand, model, serial number, date of last calibration)

-   Wavelength

-   Insertion loss: reference method (1/2/3 cable reference methods)

-   OTDR: manual or auto test

 

For each individual test

-   Identification of component under test (e.g. fiber #)

-   Test results

-   Note if results are filed electronically

 

 
The FOA wishes to acknowledge the invaluable reference materials on these subjects in the VIAVI Reference Guides to Fiber Optic Testing. These highly recommended books can be downloaded from the VIAVI website.




 


(C)2022, The Fiber Optic Association, Inc.