WHAT IS CLADDING MODE, MODE STRIPPING, AND LEAKY MODE?

1. What is Cladding Mode? As shown in the above figure, certain light rays in the incident radiation are not captured by the core of the fiber, but pass through the core-cladding interface into the cladding region. Because of the finite radius of curvature of the outer cladding surface, some of this light at this boundary will be reflected back into the cladding, where it can be trapped and propagated. This light forms the cladding modesof the fiber, and appreciable coupling can occur with the higher-order modes of the core, resulting in increased loss of the core power. 2. What is Mode Stripping? Cladding modes are suppressed by placing a high-loss material outside of the cladding surface that will absorb the light as it strikes the interface, by increasing the scattering at the cladding interface to extract the cladding modes, or by surrounding a portion of the fiber with a material whose index of refraction matches that of the cladding, causing the cladding light to transmit into the index-matching material (see the figure below). This latter technique is called mode stripping.

3. What is Leaky Mode? Another type of mode is the leaky mode, which is a nonpropagating mode with significant power shared between the core and the cladding. They are predicted by theory and occur near the cutoff conditions for propagating modes. They are attenuated in long fibers but can carry significant power in short fibers. These modes also play a role at connectors and splices, occurring in fibers where the connector or splice can cause conversion of energy from a propagating mode to a leaky mode. OPTICAL FIBER LOSS AND ATTENUATION The attenuation of an optical fiber measures the amount of light lost between input and output. Total attenuation is the sum of all losses. Optical losses of a fiber are usually expressed in decibels per kilometer (dB/km). The expression is called the fibers attenuation coefficient and the expression is

where P(z) is the optical power at a position z from the origin, P(0) is the power at the origin. For a given fiber, these losses are wavelength-dependent which is shown in the figure below. The value of the attenuation factor depends greatly on the fiber material and the manufacturing tolerances, but the figure below shows a typical optical fibers attenuation spectral distribution.

The typical fused silica glass fibers we use today has a minimum loss at 1550nm.

Optical Fiber Loss Mechanisms Absorption Absorption is uniform. The same amount of the same material always absorbs the same fraction of light at the same wavelength. If you have three blocks of the same type of glass, each 1centimeter thick, all three will absorb the same fraction of the light passing through them.

Absorption also is cumulative, so it depends on the total amount of material the light passes through. If the absorption is 1% per centimeter, it absorbs 1% of the light in the first centimeter, and 1% of the remaining light the next centimeter, and so on. Intrinsic Material Absorption Intrinsic absorption is caused by interaction of the propagating lightwave with one more more major components of glass that constitute the fibers material composition. These looses represent a fundamental minimum to the attainable loss and can be overcome only by changing the fiber material. An example of such an interaction is the infrared absorption band of SiO2shown in the above figure. However, in the wavelength regions of interest to optical communication (0.8-0.9um and 1.2-1.5um), infrared absorption tails make negligible contributions. Extrinsic Impurity Ions Absorption Extrinsic impurity ions absorption is caused by the presence of minute quantity of metallic ions (such as Fe2+, Cu2+, Cr3+) and the OH- ion from water dissolved in glass. The attenuation from these impurity ions is shown in the following table. Impurity Ion Fe2+ Fe2+ Cu2+ Cr3+ V4+ OHOHOHLoss due to 1ppm of impurity Absorption Peak Wavelength (dB/km) (um) 0.68 0.15 1.1 1.6 2.7 1.0 2.0 4.0 1.1 0.4 0.85 0.625 0.725 0.95 1.24 1.38

From the table above, we can see that 1 part per million (ppm) of Fe2+would lead to a loss of 0.68 dB/km at 1.1um. This shows the necessity of ultrapure fibers. Luckily, losses due to the metallic ions can be reduced to very low by refining the glass mixture to an impurity level below 1 par per billion (ppb). The OH- ion from water vapor in the glass leads to absorption peaks at 0.72um, 0.88um, 0.95um, 1.13um, 1.24um and 1.38um. The broad peaks at 1.24um and 1.38um in the first figure cure are

due to OH- ion. The good news is OH-ion absorption band is narrow enough that ultrapure fibers can achieve losses less than 0.2 dB/km at 1.55um. With new manufacturing techniques, we can reduce the OH-ion content to below 1 part per billion (ppb). The results are ultra-low-loss fibers which have a wider low-loss window in silica glass fibers shown in the following figure. This improvement enables the use of WDM technology in fiber optic networks, which dramatically increased the capacity of fiber optic systems.

Hydrogen Effects When fused silica glass fiber is exposed to hydrogen gas, attenuation of the fiber also increases. The hydrogen can interact with the glass to produce hydroxyl ions and their losses. Hydrogen can also infiltrate the fiber and produce its own losses near 1.2um and 1.6um. The fibers can come into contact with hydrogen which is produced by corrosion of steel-cable strength members or by certain bacteria. The way to solve this problem is to add a coating to the fiber that is impermeable to hydrogen.

Scattering Scattering losses occur when a wave interacts with a particle in a way that removes energy in the directional propagating wave and transfers it to other directions. The light isnt absorbed, just sent in another direction. However, the distinction between scattering and absorption doesnt matter much because the light is lost from the fiber in either case.

There are two main types of scattering: linear scattering and nonlinear scattering. For linear scattering, the amount of light power that is transferred from a wave is proportional to the power in the wave. It is characterized by having no change in frequency in the scattered wave. On the other hand, nonlinear scatteringis accompanied by a frequency shift of the scattered light. Nonlinear scattering is caused by high values of electric field within the fiber (modest to high amount of optical power). Nonlinear scattering causes significant power to be scattered in the forward, backward, or sideways directions. Rayleigh Scattering (Linear Scattering) Rayleigh scattering (named after the British physicist Lord Rayleigh) is the main type of linear scattering. It is caused by small-scale (small compared with the wavelength of the lightwave) inhomogeneities that are produced in the fiber fabrication process. Examples of inhomogeneities are glass composition fluctuations (which results in minute refractive index change) and density fluctuations (fundamental and not improvable). Rayleigh scattering accounts for about 96% of attenuation in optical fiber. As light travels in the core, it interacts with the silica molecules in the core. These elastic collisions between the light wave and the silica molecules result in Rayleigh scattering. If the scattered light maintains an angle that supports forward travel within the core, no attenuation occurs. If the light is scattered at an angle that does not support continued forward travel, the light is diverted out of the core and attenuation occurs. Depending on the incident angle, some portion of the light propagates forward and the other part deviates out of the propagation path and escapes from the fiber core. Some scattered light is reflected back toward the light source. This is a property that is used in an OTDR (Optical Time Domain Reflectometer) to test fibers.

Rayleigh scattering describes the elastic scattering of light by particles which are much smaller than the wavelength of light. The intensity of the scattered radiation is given by

where R is the distance between the particle and the observer, is the scattering angle, n is the refractive index of the particle, and d is the diameter of the particle. The size of a scattering particle is parameterized by the ratio x of its characteristic dimension r and wavelength :

Rayleigh scattering can be defined as scattering in the small size parameter regime x 1. Scattering from larger particles is explained by the Mie scattering for an arbitrary size parameter x. For small x the Mie theory reduces to the Rayleigh approximation. It can be seen from the above equation that Rayleigh scattering is strongly dependent upon the size of the particle and the wavelengths. The intensity of the Rayleigh scattered radiation increases rapidly as the ratio of particle size to wavelength increases. Furthermore, the intensity of Rayleigh scattered radiation is identical in the forward and reverse directions. The Rayleigh scattering model breaks down when the particle size becomes larger than around 10% of the wavelength of the incident radiation. In the case of particles with dimensions greater than this, Mies scattering model can be used to find the intensity of the scattered radiation. Rayleigh scattering depends not on the specific type of material but on the size of the particles relative to the wavelength of light. The loss due to Rayleigh scattering is proportional to -4and obviously decreases rapidly with increase in wavelength (see the first figure above Loss vs.. Wavelength). Short wavelengths are scattered more than longer wavelengths. Any wavelength that is below 800nm is unusable for optical communication because attenuation due to Rayleigh scattering is too high. The attenuation coefficient due to Rayleigh scattering in (pure) fused silica is given by the

following approximate formula where 0 = 1.7 dB/km at 0 = 0.85um

The above formula predicts the Rayleigh scattering loss to be 0.31 dB/km at 1.3um and 0.15 dB/km at 1.55um wavelengths.

Intrinsic Losses of Silica Fiber From the figure above (you can also refer to the first figure in this tutorial), we can see that the fundamental loss limits for a silica-based glass fibers are the Rayleigh scattering at short wavelengths and the material absorption (the infrared absorption) properties of silica (SiO2) at long wavelengths. A theoretical attenuation minimum for silica fibers can be predicted at a wavelength of 1550nm where the two curves cross. This has been one reason for laser sources and receivers that work in this portion of the spectrum. Mie Scattering (Linear Scattering) Mie scattering is named after German physicist Gustav Mie. This theory describes scattering of electromagnetic radiation by particles that are comparable in size to a wavelength (larger than 10% of wavelength). For particles much larger, and much smaller than the wavelength of scattered light there are simple and excellent approximations that suffice. For glass fibers, Mie scattering occurs in inhomogeneities such as core-cladding refractive index variations over the length of the fiber, impurities at the core-cladding interface, strains or bubbles in the fiber, or diameter fluctuations. Mie scattering can be reduced by carefully removing imperfections from the glass material, carefully controlling the quality and cleanliness of the manufacturing process. In commercial fibers, the effects of Mie scattering are insignificant. Optical fibers are manufactured with very few large defects. (larger than 10% of wavelength) Here is an interactive Mie Scattering calculator on the web developed by Scott Prahl. Brillouin Scattering (Nonlinear Scattering)

Brillouin scattering is caused by the nonlinearity of a medium. In glass fibers, Brillouin scattering shows as a modulation of the light by the thermal energy in the material.

An incident photon can be converted into a scattered photon of slightly lower energy, usually propagating in the backward direction, and a phonon (vibrational energy). This coupling of optical fields and acoustic waves occurs via electrostriction. The frequency of the reflected beam is slightly lower than that of the incident beam; the frequency difference vBcorresponds to the frequency of emitted phonons. This is called Brillouin Frequency Shift. This phenomenon has been used for fiber optic sensor applications.

Brillouin scattering can occur spontaneously even at low optical powers. This is different than Stimulated Brillouin Scattering which requires optical power to meet a threshold high enough to happen. Above a certain threshold power, stimulated Brillouin scattering can reflect most of the power of an incident beam. The optical power level at which stimulated Brillouin scattering becomes significant in a single mode fiber is given by the empirical formula below.

where PB= Stimulated Brillouin Scattering Optical Power Level Threshold (watts) a = Fiber radius (um)

 = Light source wavelength (um) = Fiber loss (dB/km) v = Light source linewidth (GHz) Stimulated Raman Scattering (Nonlinear Scattering) Stimulated Raman scattering is a nonlinear response of glass fibers to the optical intensity of light. This is caused by vibrations of the crystal (or glass) lattice. Stimulated Raman scattering produces a high-frequency optical phonon, as compared to Brillouin scattering, which produces a low-frequency acoustical phonon, and a scattered photon. When two laser beams with different wavelengths (and normally with the same polarization direction) propagate together through a Raman-active medium, the longer wavelength beam can experience optical amplification at the expense of the shorter wavelength beam. This phenomenon has been used for Raman amplifiers and Raman lasers.

In Stimulated Raman scattering, the scattering is predominately in the forward direction, hence the power is not lost to the receiver. Stimulated Raman Scattering also requires optical power to be higher than a threshold to happen. The formula below gives the threshold where PR= Stimulated Raman Scattering Optical Power Level Threshold (watts)

a = Fiber radius (um) = Light source wavelength (um) = Fiber loss (dB/km)

Macrobending Loss Macrobending happens when the fiber is bent into a large radius of curvature relative to the fiber diameter (large bends). These bends become a great source of power loss when the radius of curvature is less than several centimeters. Macrobend may be found in a splice tray or a fiber cable that has been bent. Macrobend wont cause significant radiation loss if it has large enough radius. However, when fibers are bent below a certain radius, radiation causes big light power loss as shown in the figure below.

Corning SMF-28e single mode fibers should not be bent below a radius of 3 inches. 50um graded-index multimode fibers, such as Corning Infinicor 600, should not be bent below a radius of 1.5 inches. 62.5um graded-index multimode fibers, such as Corning Infinicor 300, should be be bend below a radius of 1 inch. Microbending Loss Microbendings are the small-scale bends in the core-cladding interface. These are localized bends can develop during deployment of the fiber, or can be due to local mechanical stresses placed on the fiber, such as stresses induced by cabling the fiber or wrapping the fiber on a spool or bobbin.

Microbending can also happen in the fiber manufacturing process. It is sharp but microscopic curvatures that create local axial displacement of a few microns (um) and spatial wavelength displacement of a few millimeters. Microbends can cause 1 to 2 dB/km losses in fiber cabling process.

The following figure shows the the impact of a single microbend, at which, analogous to a splice, power can be coupled from the fundamental mode into higher order leaky modes.

Because external forces are transmitted to the glass fiber through the polymer coating material, the coating material properties and dimensions, as well as external factors, such as temperature and humidity, affect the microbending sensitivity of a fiber. Microbending sensitivity is also affected by coating irregularities such as variations in coating dimensions, the presence of particles such as those in the pigments of color coatings, and inhomogeneities in the properties of the coating materials that vary along the fiber axis. Interface Inhomogeneities Interface inhomogeneities can convert high-order modes into lossy modes extending into the cladding where they are removed by the jacket losses. Impurities trapped at the core-cladding interface or impurities in the fiber buffering can cause these inhomogeneities.

Single mode fibers are more susceptible to losses from geometric irregularities or defects in the jacket material. However, optical fiber manufacturing technology have improved so much that these interface inhomogeneities now play a insignificant role in fiber losses.

Optical Fiber Dispersion

1. Light Pulse Dispersion and the Reasons

In digital communication systems, information is encoded in the form of pulses and then these light pulses are transmitted from the transmitter to the receiver. The larger the number of pulses that can be sent per unit time and still be resolvable at the receiver end, the larger is the capacity of the system. However, when the light pulses travel down the fiber, the pulses spread out, and this phenomenon is called Pulse Dispersion. Pulse dispersion is shown in the following figure.

Pulse dispersion is one of the two most important factors that limit a fibers capacity (the other is fibers losses). Pulse dispersion happens because of four main reasons: A) Intermodal Dispersion (also called Modal Dispersion or Group Delay) Modal dispersion is only important in multimode fibers. Many different modes propagate in a multimode fiber. Each of these modes takes a different path and thus different length when traveling down the fiber.

Another factor is the differing propagation constants associated with each mode (each modes group velocity). Both factors contribute to modal dispersion. Single mode fiber doesnt have modal dispersion since there is only one mode propagating in the fiber. As we shall see, single mode fibers bandwidth is mainly limited by material dispersion, waveguide dispersion and PMD (polarization mode dispersion).

B) Material Dispersion (also called Chromatic Dispersion) Material dispersion is caused by the velocity of light (or its refractive index) being a function of wavelength. Each wavelength takes different amounts of time to propagate the same path. The following figure shows refractive index versus wavelength in silica.

Any given light source (even a laser) emits over a range of wavelengths as shown below.

C) Waveguide Dispersion Waveguide dispersion is a type of dispersion attributable to the relationship of the physical dimensions of the waveguide and the optical signal. In optical fibers, this shows as the propagation constant of a mode (and, hence, its velocity) being a function of a/. Where a is the core radius and is the wavelength. Unlike material dispersion, even if the refractive indexes of the core and cladding are independent of wavelength, we will still have waveguide dispersion.

C) Polarization Mode Dispersion (PMD) PMD is only important in single mode fibers. In single mode fiber, only one mode can propagate. This mode is actually composed of two distinct polarization modes. The electric fields of the two modes are perpendicular to each other. As shown below.

These two polarization modes are identical in a perfectly symmetrical fiber. However, in reality there is no perfect optical fiber. Stresses within the fiber, and forces applied to it from the outside world, causes the refractive index of glass (thus velocity) to differ very slightly for light in the two polarization modes. Thus these two polarization modes arrive at slightly different time at the end of the fiber. This is called Polarization Mode Dispersion (PMD) and is shown below.

Polarization Mode Dispersion is smaller in magnitude than material dispersion, so it hasnt been a problem until recently high speed long distance single mode fiber systems becomes popular. PMD is a serious problem when data rate exceeds 2.5 Gb/s.
2. Modal Dispersion

A. Step-Index Multimode Fiber Modal Dispersion Here we stress that the following analysis uses ray optics. Ray optics modal analysis is only valid when the V number is greater than 10. If the V number is lower, wave theory has to be used.

In a step-index multimode fiber, the center ray with a 0 incident angle travels fastest straight down the fiber. The ray which is incident at the critical angle travels slowest. The time delay between these two modes is given by

where n1 is refractive index of the core, n2is refractive index of the cladding, L is the length of the fiber, V is the V number given by

here a is the core radius, is the lights vacuum wavelength. Example: Consider a step-index multimode fiber with a 50um diameter core, 125um diameter cladding, n1 = 1.47, = 1.5%. Calculate the modal dispersion in units of ns*km-1for this fiber at 850nm wavelength. From above formula, we get

Finding (n1-n2), we have To find V, we use

Hence,

B. Graded-Index Multimode Fiber Modal Dispersion

In grade-index profile multimode fiber, the modal dispersion calculation becomes a little more complicated since we have to account for the inhomogeneous velocity of the light in the fiber as well as the sinusoidal paths as shown above. Although the higher-order modes need to travel longer path lengths, their average velocity is also higher than the light ray at the center (graded-index fiber has highest refractive index at the center and then gradually decreases toward the cladding). Here we just give out the conclusion of graded-index multimode fiber modal dispersion. The delay time incurred by the m-th mode of the fiber will be given by

where

The information on d/d and dn1/d is material-dependent and wavelength-dependent and would have to be provided from a study of the materials used in fabrication of the fibers. From the first equation, if the profile index g is set to an optimum value gopt, the linear term in can be eliminated. gopt is given by

From above discussion, we know that graded-index fiber with a optimum index profile gopt can minimize the time delay of the modes. The net delay from the lowest-order modes to the highest order modes is given by

From this equation, net modal delay can be positive or negative depending on the size of g relative to gopt. For negative modal delay, it means the higher-order modes arrive before the lower-order modes. From the comparison of step-index fiber and graded-index fiber modal delay, we know that graded-index fiber modal dispersion is smaller than that of step-index fiber. This is one of the primary advantages of the graded-index multimode fiber over a step-index multimode fiber of the same size. Current total dispersion values in graded-index multimode fiber is in the order of 0.2ns*km-1for laser light sources and 1.0 ns*km-1 for LED light sources.

3. Material Dispersion

The above figure shows the variation of refractive index n with wavelength for fused silica used in current glass fibers. Material dispersion is caused by the velocity of light (or refractive index) being a function of wavelength as shown above. All light sources (even a laser source) have some degree of spectral width. This means that even in a single mode fiber (since laser source also has some spectral linewidth), longer wavelengths travel faster and arrive earlier than shorter wavelengths at the receiver, and this causes pulse spreading. When a temporal pulse propagates through a homogeneous medium, its group velocity vg is given by

Thus, the time taken by a pulse to traverse length L of fiber is given by

which is dependent on the wavelength . The quantity

is also referred to as the group refractive index since c/N() gives the group velocity. If the source is characterized by spectral width , then each wavelength component will traverse with a different group velocity, resulting in temporal broadening of the pulse. This broadening is given by

The quantity (2d2n/d2) is a dimensionless quantity. The above pulse broadening is referred to as material dispersion and occurs when a pulse propagates through any dispersive medium. Material dispersion Dmis usually specified in units of picoseconds per kilometer (fiber length) per nanometer (spectral width of the source) as given by

where is in micrometers (um). The quantity (2d2n/d2) characterizes the material dispersion of the fiber and is grouped in the equation. Here is the plot of (2d2n/d2) for fused silica (the material for current glass fibers).

From the above figure, we can see that with 1.27um, d n/d 0. This means that at 1.27um the material dispersion of fused silica changes sign. For 1.27um the material dispersion is zero. This wavelength is referred to as Zero Material Dispersion Wavelength (ZMDW). However, we will see that the other sources of dispersion move the wavelength for zero total dispersion to 1300nm.
2 2

4. Waveguide Dispersion

As shown in Basic Optics For Optical Fiber, for a step-index fiber, when 0 < V < 2.405, the fiber becomes a single mode fiber which means that only one guided mode can propagate in the fiber.

In single mode fiber, there is no Modal Dispersion as discussed above. Pulse spreading (pulse dispersion) happens because of two mechanisms: (1) Material Dispersion as discussed above. (2) Waveguide Dispersion as will be discussed here. Waveguide dispersion results from the propagation constant of a mode (its group velocity) being a function of a/ (a is fibers core radius, is lights vacuum wavelength). Waveguide dispersion is negligible in multimode fibers and in single mode fibers operated at wavelengths below 1um, but it becomes important for single mode fibers operated in above 1.27um. The normalized propagation constant b of a mode is defined by

Since, for guided modes /k0 lies between n1 and n2, b lies between 0 and 1. For step-index fibers, b depends only on the V value of the fiber, the above equation can be rewritten as

In the above equation, the step is concluded based on that n1 is very close to n2which is true for all practical fibers. Thus we can conclude

The group velocity is given by

assumed to be independent of . Since

In the above equation, n1 and n2 are

we have

So

So the time taken by a pulse to traverse length L of the fiber is given by

where

So for a light source with a spectral width of , the waveguide dispersion is

This is our desired expression for the waveguide dispersion. For the lowest-order mode(the HE11 mode), our have b(V) as

The following figure shows the variations of b, d(bV)/dV, and V(d2(bV)/dV2) for the fundamental mode as a function of V for a step-index fiber.

The single mode values of interest are from V = 2.0 to 2.4, as shown below.

In the above figure, the value of V(d (bV)/dV ) decreases monotonically from 0.64 down to 0.25. Hence in this region, the waveguide dispersion is, in general, a small negative value (because of the minus sign in the waveguide dispersion expression above).
2 2

5. Total Dispersion for Single Mode Fiber

It can be seen that the waveguide dispersion is negative in the single mode region. Since the material dispersion is positive for greater than the zero material dispersion wavelength, there is a wavelength at which the negative waveguide dispersion will compensate the positive material dispersion. At this wavelength, the net dispersion of the single mode fiber is zero. This is shown below.

Single mode fiber with zero total dispersion around 1300nm is referred to as conventional single mode fiber (CSF) or nondispersion shifted fiber (NDSF). Most installed fiber optic systems today operate with such fibers.

6. Dispersion Shifted Single Mode Fibers

Zero Dispersion-Shifted Fiber (ZDSF) From the above discussion, we know the conventional single mode fiber has zero total dispersion around 1300nm. However, the minimum fiber loss is at 1550nm. If the zero dispersion wavelength could be shifted to the 1550nm region, we could have both minimum loss and very low dispersion. This would lead to very high bandwidth systems with very long (~100km) repeaterless transmission. The size of the waveguide dispersion has been found to be sensitive to the doping levels as well as the values of and a (core radius). Indeed, we have achieved this by changing the fiber parameters. The resulted fiber is called dispersion-shifted fiber. The first dispersion-shifted fibers had zero dispersion shifted to 1550nm to match their minimum absorption wavelength.

Fiber manufacturers achieved this by adapting the layered core design shown in the following figure.

Although this type of fiber (zero dispersion-shifted fiber) worked well for single channel systems, it proved to be unusable for WDM (wavelength-division multiplexing) systems. When multiple optical channels pass through the same fiber at wavelengths where dispersion is very close to zero, they suffer from a type of crosstalk called four-wave mixing. The degradation is so severe that zero dispersion-shifted fiber cannot be used for dense-WDM systems. ZDSF were installed in some systems but never came into wide use because of four-wave mixing problem.

Non-Zero Dispersion-Shifted Fiber (NZDSF)

In order to avoid four-wave mixing problem in WDM systems, non-zero dispersion-shifted fibers were invented. The idea is to move the zero-dispersion wavelength outside the band used for erbium-fiber doped amplifiers. The name Non-Zero Dispersion-Shifted comes from the fact that the dispersion is shifted to a value that is low but not zero in the 1550nm bad of erbium-fiber amplifiers. But you can shift the dispersion to a positive or negative value. This small dispersion is enough to keep signals at closely spaced wavelengths from staying in phase over long distances and thus preventing fourwave mixing. The following figure shows two NZDSF fibers, one with plus sign at 1550nm and one with negative sign at 1550nm (both in light blue color).

Positive dispersion is an advantage because it is easier to compensate than negative dispersion. This is done by using other layered core structures to adjust the waveguide dispersion differently. The following figure shows two refractive-index profile designs.

Dispersion-Flattened Fiber (DFF)

Another way to minimize dispersion effects for dense-WDM systems is to reduce the slope of the dispersion curve. This is called Dispersion-Flattened Fiber (DFF) as shown in the 3rd figure above in pink color. The reduced dispersion curve slope helps to simplify the task of dispersion compensation for systems with many optical channels, but there are tradeoffs. An important one is that dispersionflattened fiber designs tend to have smaller mode-field diameter which concentrate optical power in a smaller volume. The increased power density in fiber can cause non-linear effects.

Dispersion-Compensating Fiber Now, in many countries, tens of millions of miles of CSF (Conventional Single Mode Fiber with zero dispersion near 1300nm) already exist in the underground ducts operating at 1300nm. One could increase the transmission capacity by operating these fibers at 1550nm and using WDM techniques and optical amplifiers. But, then there will be significant residual (positive) dispersion. On the other hand, replacing these fibers by non-zero dispersion-shifted fibers would involve huge cost. In recent years, there has been considerable work in upgrading the installed 1300nm optimized fiber links for operation at 1550nm. This is achieved by developing fibers with very large negative dispersion coefficients, a few hundred meters to a kilometer, which can be used to compensate for dispersion over tens of kilometers of the fiber in the link. These fibers tend to have a high-index difference between core and cladding, and often have a small effective area. The following figure shows the refractive-index profile of one design.

How to Test a Fiber Optic System with an OTDR (Optical Time Domain Reflectomer)
>> The Optical Time Domain Reflectometer (OTDR)

OTDR is connected to one end of any fiber optic system up to 250km in length. Within a few seconds, we are able to measure the overall loss, or the loss of any part of a system, the overall length of the fiber and the distance between any points of interest. OTDR is a amazing test instrument for fiber optic systems.

1. A Use for Rayleigh Scatter

As light travels along the fiber, a small proportion of it is lost by Rayleigh scattering. As the light is scattered in all directions, some of it just happens to return back along the fiber towards the light source. This returned light is called backscatter as shown below.

The backscatter power is a fixed proportion of the incoming power and as the losses take their toll on the incoming power, the returned power also diminishes as shown in the following figure.

The OTDR can continuously measure the returned power level and hence deduce the losses encountered on the fiber. Any additional losses such as connectors and fusion splices have the effect of suddenly reducing the transmitted power on the fiber and hence causing a corresponding change in backscatter power. The position and degree of the losses can be ascertained.

2. Measuring Distances
The OTDR uses a system rather like a radar set. It sends out a pulse of light and listens for echoes from the fiber. If it knows the speed of light and can measure the time taken for the light to travel along the fiber, it is an easy job to calculate the length of the fiber.

3. To Find the Speed of the Light

Assuming the refractive index of the core is 1.5, the infrared light travels at a speed of

This means that it will take

This is a useful figure to remember, 5 nanoseconds per meter (5 nsm-1). If the OTDR measures a time delay of 1.4us, then the distance travelled by the light is

The 280 meters is the total distance traveled by the light and is the there and back distance. The length of the fiber is therefore only 140m. This adjustment is performed automatically by the OTDR it just displays the final result of 140m.

4. Inside the OTDR

A. Timer The timer produces a voltage pulse which is used to start the timing process in the display at the same moment as the laser is activated. B. Pulsed Laser The laser is switched on for a brief moment. The on time being between 1ns and 10us. We will look at the significance of the choice of on time or pulsewidth a little bit later. The wavelength of the laser can be switched to suit the system to be investigated. C. Directive Coupler The directive coupler allows the laser light to pass straight through into the fiber under test. The backscatter from the whole length of the fiber approaches the directive coupler from the opposite direction. In this case the mirror surface reflects the light into the avalanche photodiode (APD). The light has now been converted into an electrical signal. D. Amplifying and Averaging The electrical signal from the APD is very weak and requires amplification before it can be displayed. The averaging feature is quite interesting and we will look at it separately towards the end of this tutorial. E. Display

The amplified signals are passed on to the display. The display is either a CRT like an oscilloscope, or a LCD as in laptop computers. They display the returned signals on a simple XY plot with the range across the bottom and the power level in dB up the side. The following figure shows a typical display. The current parameter settings are shown over the grid. They can be changed to suit the measurements being undertaken. The range scale displayed shows a 50km length of fiber. In this case it is from 0 to 50km but it could be any other 50km slice, for example, from 20km to 70km. It can also be expanded to give a detailed view of a shorter length of fiber such as 0-5m, or 25-30m.

The range can be read from the horizontal scale but for more precision, a variable range marker is used. This is a movable line which can be switched on and positioned anywhere on the trace. Its range is shown on the screen together with the power level of the received signal at that point. To find the length of the fiber, the marker is simply positioned at the end of the fiber and the distance is read off the screen. It is usual to provide up to five markers so that several points can be measured simultaneously. F. Data Handling An internal memory or floppy disk can store the data for later analysis. The output is also available via RS232 link for downloading to a computer. In addition, many OTDRs have an onboard printer to provide hard copies of the information on the screen. This provides useful before and after images for fault repair as well as a record of the initial installation.

5. A Simple Measurement
If we were to connect a length of fiber, say 300m, to the OTDR the result would look as shown in the following figure.

Whenever the light passes through a cleaved end of a piece of fiber, a Fresnel reflection occurs. This is seen at the far end of the fiber and also at the launch connector. Indeed, it is quite usual to obtain a Fresnel reflection from the end of the fiber without actually cleaving it. Just breaking the fiber is usually enough. The Fresnel at the launch connector occurs at the front panel of the OTDR and, since the laser power is high at this point, the reflection is also high. The result of this is a relatively high pulse of energy passing through the receiver amplifier. The amplifier output voltage swings above and below the real level, in an effect called ringing. This is a normal amplifier response to a sudden change of input level. The receiver takes a few nanoseconds to recover from this sudden change of signal level.

6. Dead Zones

The Fresnel reflection and the subsequent amplifier recovery time results in a short period during which the amplifier cannot respond to any further input signals. This period of time is called a dead zone. It occurs to some extent whenever a sudden change of signal amplitude occurs. The one at the start of the fiber where the signal is being launched is called the launch dead zone and others are called event dead zones or just dead zones.

7. Overcoming the Launch Dead Zone

As the launch dead zone occupies a distance of up to 20 meters or so, this means that, given the job of checking a 300m fiber, we may only be able to check 280m of it. The customer would not be delighted. To overcome this problem, we add our own patch cord at the beginning of the system. If we make this patch cord about 100m in length, we can guarantee that all launch dead zone problems have finished before the customers fiber is reached.

The patch cord is joined to the main system by a connector which will show up on the OTDR readout as a small Fresnel reflection and a power loss. The power loss is indicated by the sudden drop in the power level on the OTDR trace.

8. Length and Attenuation

The end of the fiber appears to be at 400m on the horizontal scale but we must deduct 100m to account for our patch cord. This gives an actual length of 300m for the fiber being tested. Immediately after the patch cord Fresnel reflection the power level shown on the vertical scale is about 10.8dB and at the end of the 300m run, the power has fallen to about 11.3 dB. A reduction in power level of 0.5 dB in 300 meters indicates a fiber attenuation of:

Most OTDRs provide a loss measuring system using two markers. The two makers are switched on and positioned on a length of fiber which does not include any other events like connectors or whatever as shown in the following figure.

The OTDR then reads the difference in power level at the two positions and the distance between them, performs the above calculation for us and displays the loss per kilometer for the fiber. This provides a more accurate result than trying to read off the decibel and range values from the scales on the display and having to do our own calculations.

9. An OTDR Display of a Typical System

The OTDR can see Fresnel reflections and losses. With this information, we can deduce the appearance of various events on an OTDR trace as seen below.

A. Connectors A pair of connectors will give rise to a power loss and also a Fresnel reflection due to the polished end of the fiber. B. Fusion Splice Fusion splices do not cause any Fresnel reflections as the cleaved ends of the fiber are now fused into a single piece of fiber. They do, however, show a loss of power. A good quality fusion splice will actually be difficult to spot owing to the low losses. Any signs of a Fresnel reflection is a sure sign of a very poor fusion splice. C. Mechanical Splice Mechanical splices appear similar to a poor quality fusion splice. The fibers do have cleaved ends of course but the Fresnel reflection is avoided by the use of index marching gel within the splice. The losses to be expected are similar to the least acceptable fusion splices. D. Bend Loss

This is simply a loss of power in the area of the bend. If the loss is very localized, the result is indistinguishable from a fusion or mechanic splice.

10. Ghost Echoes (False Reflection)

In the following figure, some of the launched energy is reflected back from the connectors at the end of the patch cord at a range of 100m. This light returns and strikes the polished face of the launch fiber on the OTDR front panel. Some of this energy is again reflected to be re-launched along the fiber and will cause another indication from the end of the patch cord, giving a false, or ghost, Fresnel reflection at a range of 200m and a false end to the fiber at 500m.

As there is a polished end at both ends of the patch cord, it is theoretically possible for the light to bounce to and fro along this length of fiber giving rise to a whole series of ghost reflections. In the figure a second reflection is shown at a range of 300m. It is very rare for any further reflections to be seen. The maximum amplitude of the Fresnel reflection is 4% of the incoming signal, and is usually much less. Looking at the calculations, even assuming the worst reflection, the returned energy is 4% or 0.04 of the launched energy. The re-launched energy, as a result of another reflection is 4% of the 4% or 0.042 = 0.0016 x input energy. This shows that we need a lot of input energy to cause a ghost reflection.

A second ghost would require another two reflections giving rise to a signal of only 0.00000256 of the launched energy. Subsequent reflections die out very quickly as we could imagine. Ghost reflections can be recognized by their even spacing. If we have a reflection at 387m and another at 774 then we have either a strange coincidence or a ghost. Ghost reflections have a Fresnel reflection but do not show any loss. The loss signal is actually of too small an energy level to be seen on the display. If a reflection shows up after the end of the fiber, it has got to be a ghost.

11. Effects of Changing the Pulsewidth

The maximum range that can be measured is determined by the energy contained within the pulse of laser light. The light has to be able to travel the full length of the fiber, be reflected, and return to the OTDR and still be of larger amplitude than the background noise. Now, the energy contained in the pulse is proportional to the length of the pulse so to obtain the greatest range the longest possible pulsewidth should be used as illustrated in the following figure.

This cannot be the whole story, as OTDRs offer a wide range of pulsewidths. We have seen that light covers a distance of 1 meter every 5 nanoseconds so a pulsewidth of 100nm would extend for a distance of 20 meters along the fiber (see the following figure). When the light reaches an event, such as a connector, there is a reflection and a sudden fall in power level. The reflection occurs over the whole of the 20m of the outgoing pulse. Returning to the

OTDR is therefore a 20m reflection. Each event on the fiber system will also cause a 20m pulse to be reflected back towards the OTDR.

Now imagine two such events separated by a distance of 10m or less as in the following figure. The two reflections will overlap and join up on the return path to the OTDR. The OTDR will simply receive a single burst of light and will be quite unable to detect that two different events have occurred. The losses will add, so two fusion splices for example, each with a loss of 0.2dB will be shown as a single splice with a loss of 0.4dB. The minimum distance separating two events that can be displayed separately is called the range discrimination of the OTDR. The shortest pulsewidth on an OTDR may well be in the order of 10ns so at a rate of 5nsm-1 this will provide a pulse length in the fiber of 2m. The range discrimination is half this figure so that two events separated by a distance greater than 1m can be displayed as separate events. At the

other end of the scale, a maximum pulsewidth of 10us would result in a range discrimination of 1 km. Another effect of changing the pulsewidth is on dead zones. Increasing the energy in the pulse will cause a larger Fresnel reflection. This, in turn, means that the amplifier will take longer to recover and hence the event dead zones will become larger as shown in the next figure.

12. Which Pulsewidth to Choose?

Most OTDRs give a choice of at least five different pulse length from which to select. Low pulse widths mean good separation of events but the pulse has a low energy content so the maximum range is very poor. A pulse width of 10ns may well provide a maximum range of only a kilometer with a range discrimination of 1 meter. The wider the pulse, the longer the range but the worse the range discrimination. A 1us pulse width will have a range of 40 km but cannot separate events closer together than 100 m. As a general guide, use the shortest pulse that will provide the required range.

13. Averaging
The instantaneous value of the backscatter returning from the fiber is very weak and contains a high noise level which tends to mask the return signal. As the noise is random, its amplitude should average out to zero over a period of time. This is the idea behind the averaging circuit.

The incoming signals are stored and averaged before being displayed. The larger the number of signals averaged, the cleaner will be the final result but the slower will be the response to any changes that occur during the test. The mathematical process used to perform the effect is called least squares averaging or LSA. The following figure shows the enormous benefit of employing averaging to reduce the noise effect.

Occasionally it is useful to switch the averaging off to see a real time signal from the fiber to see the effects of making adjustments to connectors etc. This is an easy way to optimize connectors, mechanical splices, bends etc. Simply fiddle with it and watch the OTDR screen.

14. OTDR Dynamic Range

When a close range reflection, such as the launch Fresnel occurs, the reflected energy must not be too high otherwise it could damage the OTDR receiving circuit. The power levels decrease as the light travels along the fiber and eventually the reflections are similar in level to that of the noise and can no longer be used.

The difference between the highest safe value of the input power and the minimum detectable power is called the dynamic range of the OTDR and, along with the pulse width and the fiber losses, determines the useful range of the equipment. If an OTDR was quoted as having a dynamic range of 36 dB, it could measure an 18km run of fiber with a loss of 2 dB/km, or alternatively a 72 km length of fiber having a loss of 0.5 dB/km, or any other combination that multiplies out to 36 dB.