Fundamentals of Optical Fibre

2. Optical Fibers

2.1 Fundamentals of Optical Fibers

The term optical fibers indicates special forms of optical waveguides, the most im-portant special features of which are:

rotationally symmetrical cross-section flexiblecan be produced in great lengths

The characteristics of optical fibers are determined by a multitude of possible constructive details. For example, the material selected primarily determines the attenuation and the thermal stability. On the other hand, the optical bandwidth, in essence the transmission capacity, is determined by the refractive index profile. This is most likely the reason why most optical fibers are named after their index profile. All current variations will be presented in the following sections.

2.1.1 Refractive Index Profiles

The properties of wave guiding through a fiber are governed largely by the profile of the refractive index of the core and cladding. In a step index profile fiber the refractive index is constant across the entire cross section of the core and cladding (Fig. 2.1) while the light rays propagate along straight lines in the core and are completely reflected at the core/cladding interface.

refractive index n(r)

r

a

-a

Fig. 2.1: Refractive index profile in a step index profile fiber

38 2.1 Fundamentals of Optical Fibers

The profile of the refractive index in the core and in the cladding is expressed as follows:

arforn)r(n

arforn)r(n

cladding

core

where a is the core radius. The individual rays cover different distances, so that there are considerable dif-

ferences in their respective transit times. Choosing a fiber with a graded-index profile can minimize these differences. Fibers with a graded-index profile are made up of a core having a radius-dependent refractive index and a cladding with a constant refractive index (Fig. 2.2):

arforn)r(n

arfora

r1²n)²r(n

cladding

g

max,core

where g is the profile exponent and is the relative refractive index difference:

2core

2cladding

2core

n2

nn

refractive index n(r)

ra

-a

Fig. 2.2: Principle of a fiber with a graded-index profile

Those rays propagating in the center travel a shorter distance, but because of the higher refractive index there, they travel at a lower speed. On the other hand, the smaller refractive index near the cladding causes the rays traveling there to have a higher velocity, but they have a longer distance to travel. By choosing a suitable profile exponent it is possible to compensate for these differences in tran-sit time. For negligible chromatic dispersion the ideal profile exponent is 2. One then speaks of a parabolic index profile.

2.1 Fundamentals of Optical Fibers 39

2.1.2 Numerical Aperture

When light enters the fiber's input face at an angle max, it is refracted at an angle

max (Fig. 2.3). Applying the law of refraction we have:

2corecladdingcoremax0

max22

corecladdingmax2

coremax0

maxcoremax0

maxcoremaxcoremax0

nn-1nsinn

sinnnithw,sin-1nsinn

cosnsinn

)-sin(90nsinnsinn

2claddimg

2coremax

02

cladding2

coremax0

nnsin

follows1nfor,nnsinn

The sine of the maximum incident-ray angle max is defined as the numerical aperture AN (Fig. 2.3). The angle max is referred to as the acceptance angle, and twice the acceptance angle is referred to as the aperture angle. Using the relative refractive index difference , the value for AN is obtained as:

2nsinA coremaxN

max

n0

max

max

Fig. 2.3: Definition of the acceptance angle

Thus, the value of the numerical aperture (NA) is solely dependent on the diffe-rence in the refractive indices of the core/cladding material.

Example: The refractive indices of a standard PMMA fiber are ncore = 1.49 and ncladding = 1.40; we thus obtain AN = 0.50 and max = 30 .

Whereas the numerical aperture of the step-index profile fiber remains constant over the entire core, the graded-index profile fiber exhibits a decreasing accep-tance angle from the center of the core to the cladding (Fig. 2.4).


Fig. 2.4: Acceptance angle of a graded-index profile fiber

Compared with other fiber types (Fig. 2.5), POF has the largest numerical aperture and the largest core diameter. This is one of the most important advan-tages of POF, since the connection technology that can be used for POF is more economical to apply than that used for glass fibers.

10/ 125 μm 50/ 125 μm

100/ 140 μm

980/ 1000 μm

0 mm 0.5 mm 1.0 mm

singlemode glass fiber multimode glass fiber polymer fiber

200/ 230 μm

multimode glass fiber(plastic clad)

62.5/ 125 μm

Fig. 2.5: Aperture angle and core diameter of glass fibers and polymer fibers

2.1.3 Ray Trajectory in Optical Fibers

In the step index profile fiber, light propagates along a zigzag path, being totally reflected at the core/cladding interface; in the graded-index profile fiber, light pro-pagates on a sinusoidal trajectory that is created within the graded-index profile through refraction. If the incident light rays lie within one and the same plane through which the fiber axis runs, meridional rays are formed. In all other cases, skew rays are formed. Figure 2.6 shows the projection onto the fiber's incident face. Step and graded-index profile fibers show the same behavior. The speci-fication of the numerical aperture always refers to the meridional rays.


’’

’’

Fig. 2.6: Meridional rays

Skew rays form an angle of < 90 with the tangential plane at the core/cladding interface (Fig. 2.7). They never cross the fiber axis and propagate along screw-like paths. For step index profile fibers, the projection onto the cross-sectional area resembles a polygonal line so that these rays do not cross a circle-shaped area having a radius rk around of the axis.

rk

Fig. 2.7: Skew rays in step index profile fibers

In graded-index fibers with a parabolic profile, ellipses are formed in the pro-jection (Fig. 2.8 left) that may under certain circumstances form circles; these rays are called helical rays (Fig. 2.8 right). Their distance from the fiber axis is always constant.

Fig. 2.8: Helical rays (left) and skew rays (right) in graded-index profile fibers


2.1.4 Modes in Optical Fibers

2.1.4.1 The Mode Concept

The phenomena of refraction and reflection discussed so far can be graphically explained with the help of geometrical optics, whereby the size of the wavelength and the diameter of the finite ray are not considered ( and dray= 0). However, to obtain a complete description of the wave guiding phenomenon, the wave pro-perties of light must also be considered. The goal is to calculate the electric field and intensity distribution of the light in the optical fiber. The Eigenvalue equation is derived and solved on the basis of the Maxwell equations. Ref. [Blu98] provides a detailed description. The solutions to the Eigenvalue equations are a finite number of field distributions within the light waveguide. These field distributions are referred to as modes of the waveguide. If we apply this concept to the ray model, this means that apparently not all incidental rays for which < max is true can propagate, but rather only those rays that have a particular angle. Figure 2.9 illustrates this situation: in order for light to propagate in a particular direction, a wave must constructively overlap itself with its own reflecting wave in such a way that the phase position is repeated after double reflection. The black lines perpendicular to the direction of propagation identify the planes with the same phase angle. The spacing is /ncore.

electrical field

transverse direction

d

/ncore

Fig. 2.9: Formation of the mode structure within the waveguide

Whereas the zigzag paths would lead to intensity distributions within the ray-optical model that would change depending on the length of the fibers, the wave model provides a constant light-dark distribution that is independent of the length across the waveguide's cross-section.

The number N of the guided modes is approximately described by:

2V2g

g

2

1N

where V = 2 a AN/ , a is the radius and g is the profile exponent (see also Sec-tion 1.1.5).

For step-index profiles g . This results in a value of N ½ · V² for the number of modes. For parabolic profiles g = 2 and thus N ¼ · V². A polymer op-


tical fiber with AN = 0.5, a core radius of 0.5 mm and a wavelength of = 650 nm can carry 2.9 million modes. If the angle of total reflection is exceeded, radiation modes are created and the light is radiated into the cladding. If the refractive index of the cladding is higher than the surrounding medium (air, for example), cladding modes may be formed. In the POF, the optical cladding is encased in an absorbing jacket so that no cladding modes can form. In contrast to guided modes, it is not possible to count radiation modes. They do not take part in signal transmission. (Fig. 2.10 special conditions for POF are explained below). Higher modes propa-gate under a larger angle, lower modes under a smaller one. Under certain circum-stances skew rays may turn into so-called leaky waves, which, on the one hand, are guided in the Z-direction and, on the other hand, transfer energy to the cladding. Under certain conditions they can still be detected in POF even after several 10s of meters. Hence, they can influence both the transmission process as well as the measuring techniques used.

radiation mode

higher modelower mode

cladding mode

Fig. 2.10: Guided, cladding and radiation modes

The following equation describes the relationship between the angles , and in Fig. 2.11 ([Sny83]):

sinsincos

is the angle of the incident and reflected ray relative to the surface normal of the tangential plane in P. describes the angle between the reflection plane and the tangent plane, and is the angle between the projection of the skew ray on the cross-sectional plane and the direction of propagation (parallel to the fiber axis).

Figure 2.12 summarizes the various ray types according to the respective angles derived from the above equation ([Bun99a]). For guided rays holds < max and > max. The leaky waves are shown in the subsequent rectangle while the ray modes are shown above the line = max. For meridional rays = 90 - because

= 90 , i.e. they lie on the blue line.


P

Fig. 2.11: Designation of the angles of a skew ray; the right diagram shows the angle ,which is obtained by projecting the skew ray on to the cross-sectional plane

0 10 20 30 40 50 60 70 90 80

0

10

20

30

40

50

60

70

90

80max

max [°]

inside this triangle there are the guided rays

inside this rectangle there are the leaky modes

inside this triangle there are the radiation rays

meridional rays

[°]

Fig. 2.12: The different types of rays

2.1.4.2 Mode Propagation in Real Fibers

Several chapters of this book discuss the special characteristics of light propa-gation in POF. Here now, the processes that need to be considered will be looked at as a whole. The function of fibers as a waveguide for passing on light by means of total reflection at the core/cladding interface has already been discussed.

If the ray model were applied consistently, then a light ray launched into an ideal fiber would always propagate at the same angle relative to the fiber axis. With a divergent light source, the far field would always remain constant along the length of the fiber. This would not be true for the near field, as Fig. 2.13 illus-trates: depending on the course of the ray, different locations along the fiber would


generate different near fields in the form of point structures. However, this contra-dicts the results obtained through experiments: there a continuous distribution of intensity is obtained, and from a certain length onwards the intensity does not change at all. Although the ray model is very illustrative, its practical application is limited as the example above shows. In order to be able to describe experi-mental results it is thus necessary to move on to the mode concept. In this respect it is important to keep in mind that many optical simulation programs work on the basis of discrete light rays. In order to obtain truly realistic results, a sufficient number of rays has to be simulated.

nearfield (very schematically)

only a few launched discrete modes

Fig. 2.13: Near fields under conditions of the ray model with only a few discrete light paths (in practice very difficult to measure and visible only on very short lengths)

2.1.5 Parameters for Describing Real Fibers and Waveguides

In order to describe the characteristics of real fibers and waveguides different parameters are defined which vary in importance depending on their respective application. All of these parameters are influenced by the propagation conditions of the different modes. In the case of multimode fibers most characteristics depend typically on the mode distribution. This means that a fiber initially allows the propagation of light in different paths (modes). Depending on the light sources at the front end of the fiber, not all these modes are launched, at least not with a uni-form power distribution. Since each mode has different characteristics, an altered behavior of the fiber is on average the result. In addition, the problem becomes more complicated since an exchange of energy between the modes can occur over the length.

Typical fiber characteristics will be defined and explained in the following sections. Uniform Mode Distribution (UMD) and Equilibrium Mode Distribution (EMD) are the usual standard measuring conditions.


2.1.5.1 Attenuation

The most important process encountered by light as it passes through a fiber is at-tenuation. When passing through an optical fiber of the length L, the power of the light decreases (Fig. 2.14). The following equation applies to the optical power:

L0L ePP

where PL and P0 are the power of the light after passage through a fiber of length L in km and at the front end of the fiber, respectively; ´ is the value of the attenuation coefficient in km-1.

P0 PL

L

Fig. 2.14: Definition of attenuation

To make it easier to work with the numbers involved here, it is usual to express attenuation logarithmically. Thus, the attenuation coefficient is expressed as in dB/km.

343,4P

Plog

L

10

L

0

Attenuation value a is the non-dimensional variable (given as a number or in dB) obtained from the product · L. Figure 1.19 illustrates the relationship bet-ween the attenuation value and the change in power as a percentage.

05101520 25 30

0.1 1 10 100power ratio PL/P0 [%]

attenuation factor a [dB]

Fig. 2.15: Conversion of the power ratio PL/P0 in % into the dB value

Very often there is not a clear differentiation in the technical literature between attenuation per unit length and attenuation factor a. One often speaks simply of the attenuation of the fiber. The addition “spectral” refers to the wavelength dependence. A mistake is avoided, however, when the unit is indicated. We still have to mention that attenuation and attenuation per unit length are practically always indicated as positive numbers.


Quantity Symbol Unit Formula

attenuation coefficient, lin. ´ km-1 {ln (P0/PL)}/L

attenuation coefficient, log. dB/km {10 log (P0/PL)}/L

attenuation a dB 10 log (P0/PL)

Especially in the area of optical short-range communication, indicating the fiber attenuations in dB is much more practical than, for example, representing the ab-solute transmission. POFs are being used more and more in the near infrared range for quite short transmission lengths. Finally, PMMA can also be used for wave-guide structures in the mm range. Fig. 2.16 shows the attenuation curve of a PMMA-POF according to [Hess04].

500 600 700 800 900 1000

attenuation [dB/km]

wavelength [nm]10

100

1,000

10,000

30

300

3,000

100,000

30,000

theory measured

Fig. 2.16: Attenuation spectrum of the PMMA-POF (theory and measured by [Hess04])

Nevertheless, the representation comprises approximately 3 decades, i.e. a fac-tor of 1,000 which cannot be overlooked on a linear scale.

2.1.5.2 Mode-Dependent Attenuation

When talking about glass fibers, it is often assumed that the attenuation of all light rays is identical. For practical purposes, this assumption is sufficiently accurate. With POF, the path difference between the rays parallel to the axis and the propa-gation directions close to the critical angle of total reflection can become quite large. For the standard NA-POF with AN = 0.50 this difference is about 6%. For polycarbonate fibers with AN = 0.90, the difference is even 21%. For this reason alone, there is a considerably greater level of attenuation where large propagation


angles are involved. In 100 m of POF, a light ray of this type will travel 6 m farther which results in an additional loss of more than 1 dB when the attenuation level is 200 dB/km. At 1,000 dB/km for polycarbonate fiber, this would result in an additional loss of 4 dB after 20 m of travel (less than 50% of the launched power reach the fiber output).

The second, more significant cause for mode-dependent attenuation is the attenuation resulting from the cladding material. Fluorinated polymers are used as optical cladding for PMMA fibers; these claddings may have an attenuation of several 10,000 dB/km [Paar92]. Locking more exactly on the propagation of a plane wave at the interface, we find that, even if total reflection results, the electri-cal field escapes into the optically thinner medium by a distance in the order of magnitude of the wavelength. This process is also known as the Goos-Hänchen Shift ([Bun99a]) and the model explains this as resulting from a shift of the reflec-tion plane into the optically thinner medium. The reflected ray is hence slightly displaced on the interface surface, as can be seen in Fig. 2.17. In this model, the additional light path would be subjected to the higher attenuation of the cladding material.

core

cladding

area of higher attenuation

Fig. 2.17: Goos-Hänchen shift

Although the light path in the cladding is only in the m range for each reflec-tion, it still plays a significant role because of the much higher attenuation encoun-tered there. This effect is particularly striking when the core diameters are reduced in size. Theoretically speaking, attenuation and bandwidth should not be depen-dent on the core diameter. Nevertheless, thin cores such as those used in multi-core fibers have indeed considerably larger bandwidths [Tesh98], a slightly increased attenuation and narrower far-field widths. These effects are explained quite well in [Bun99b] and [Ziem99c].

This effect also occurs in glass fibers. Silica glass fibers with a polymer cladding (PCS) have losses in the core below 10 dB/km (wavelength range from 650 nm to 1,300 nm), whereas the polymer cladding has an attenuation of several 100 to 1,000 dB/km.

Attenuation values of 180 dB/km for the core and 9,000 dB/km for the cladding are indicated in [Ebb03] for step index profile glass-glass fibers (used in fiber bundles). Reasonably priced conventional glasses - albeit much purer than in win-dow glass - are used in these fibers and not silica glass.

In singlemode and graded-index profile silica fibers there are no mentionable differences in attenuation between the core and the cladding since both consist of


Si02. The germanium dopant in the core does not have any great influence. An important consequence of the mode-dependent attenuation is, as will be discussed later on, a significantly narrower far field after greater fiber lengths than one would expect from the fiber NA.

2.1.5.3 Mode Coupling

The term mode coupling refers to the process by which energy from one direction of propagation is transferred to several others. This can happen, for example at scattering centers. Since the light scattering in a PMMA-POF makes up a con-siderable part of the attenuation, this process is always present. Figure 2.18 clarifies the procedure (still in the ray model).

scattering center

Fig. 2.18: Mode coupling at a scattering center

Many experimental results clearly indicate that mode coupling occurs predomi-nately at the core/cladding interface (Fig. 2.19). This can be explained by the fact that is it not possible to create an ideal surface in the sub-nanometer range when very large polymer molecules are involved. Thus, mode coupling is also depen-dent on the angle of propagation.

scattering center

cladding

core

Fig. 2.19: Mode coupling at the core/cladding interface

Mode coupling alters the bandwidth of a fiber. When collimated light is launched, energy is gradually transferred to the higher angle ranges so that mode dispersion increases and bandwidth decreases. If light is introduced in all angle ranges, so that maximum differential delays occur, energy is exchanged between the angles so that the initially slower rays become “faster” and vice versa. Accor-


ding to the laws of statistics, the differential delay (or more precisely, the standard deviation) does not increase in a linear relationship to the length but approxi-mately only proportional to the square root of the length. This applies to lengths in excess of a characteristic coupling length, which for PMMA-POF is generally several 10 m.

Mode coupling always results in additional attenuation. Whenever there are changes in the light propagation, energy is coupled into those angle ranges in which there is no longer any light guiding. The shorter the coupling length, the larger the additional attenuation will be. If the observed behavior of the POF, namely the filling up of the near field after a few 10 cm of fiber, could be ex-plained exclusively because of the mode coupling, then additional attenuations in the range of 1000 dB/km would result - which indeed does not occur.

Figure 2.20 shows an electron microscope picture of the core-cladding interface layer (photo ZWL, 2003). The marked smooth part running from the top left to the bottom right is the surface of the core with the cladding removed. At the top right you can see the cracked core. The step is the 10 μm thick optical cladding. Further theoretical considerations on the problems of scattering can be found in [Kru06a] and [Kru06b].

Fig. 2.20: Photo of the core-cladding interface of SI-POF taken by electron microscope (ZWL Lauf)

2.1.5.4 Mode Conversion

The definition of propagation angles or of modes actually applies only to wave-guides that are straight. It takes just one bend to make a different approach neces-sary. The most precise method would be to recalculate the modes for the system of the now bent fiber; however, this is theoretically and practically much too com-plex a process. It is more appropriate to consider the zone before and after the bend as a straight waveguide and, at the bend, to perform a transformation onto the new reference axis. Formally, light is thus transmitted from one propagation direction to another, as Fig. 2.21 demonstrates.


fiber axis in front of a bend fiber axis behind a bend

new propagation angle

Fig. 2.21: Mode conversion at a bend

Strictly speaking, mode conversion can be described as a special case of mode coupling. The difference is that the number of modes or the propagation directions is not increased. In the POF mode conversion most likely occurs at the core/clad-ding interface surface, for example at micro bends or at fluctuations in the re-fractive index difference. The question of the influence of mode conversion and coupling on the additional attenuation depends essentially on the angle depen-dency of the processes. The more the direction of the light is altered, the more los-ses occur. A quantitative analysis of these processes for POF is extremely difficult and is yet to be carried out. However, for the physical processes assumed, mode coupling should have a larger angle-independent contribution (scattering on larger inhomogenities).

Fig. 2.22: Far fields of different POF (product A/B at the top/bottom); left/right after 20 m/50 m of fiber, launch with collimated light (AN Launch < 0.016)


An impressive experiment that confirmed this statement was shown in [Poi00]. If collimated light is launched into a SI-POF, a ring-shaped far field can be gene-rated at the output even after 50 m of fiber, for which purpose the fiber might be properly bent. This experiment can only be explained under the assumption that mode conversion predominates. However, the different fibers made by different manufacturers show considerable differences in their behavior which do not necessarily have an effect on attenuation.

It is easy to see here that the mode field is not completely filled even after 20 m to 50 m (Fig. 2.22).

2.1.5.5 Mode Coupling Lengths

The length of a fiber in which a state of equilibrium arises through mode conver-sion and coupling is described as coupling length whereby different definitions exist. The best known is the description with the aid of a length-dependent band-width. Here the coupling length is the point at which the linear decrease in the bandwidth turns to a root dependency (see Fig. 2.36). In practice this point is diffi-cult to measure, but other parameters such as far field width and attenuation, change with fiber length. For example, values for the kilometric attenuation with different launch conditions are shown In Figures 2.23 and 2.24.

50

100

150

200

250

300

350

400

lPOF [m]

[dB/km]

1 10 1002 205 50

fiber “A”

source “I”source “II”source “III”source “IV”

Fig. 2.23: Attenuation of a SI fiber under different excitation (acc. to [Lub02b])

Both diagrams show very clearly that the different launch conditions (source I emits very widely, source IV nearly collimated) lead to extremely different attenu-ation values. After some ten meters, however, the differences disappear for the most part through mode coupling. Evidently, there are great differences among the fiber types.


50

100

150

200

250

300

350

400

lPOF [m]

[dB/km]

1 10 1002 205 50

fiber “B”

source “I”source “II”source “III”source “IV”

Fig. 2.24: Attenuation of another SI-fiber at different launch conditions

The next two figures 2.25 and 2.26 show measurements of far field widths for a POF and a PCS each with altered launch conditions. Once again it can clearly be seen how the differences caused by the different coupling conditions are evened out after some 10 to 100 m.

10

12

14

16

18

20

22

24

2628

30

32far field width [°]

POF length [m]

1005 5020

PMMA SI-POF

0.64

0.48

0.330.19

0.09

0.05

NALaunch:

Fig. 2.25: Launch dependent far field widths of a PMMA SI-POF

In the 200 μm thick PCS it takes considerably longer to establish the equili-brium mode distribution especially when the length is related to the fiber diameter. The values of the NA (calculated from the 5% far field width) are represented for lengths up to 500 m.


0.10

0.15

0.20

0.25

0.30

0.35

0.40

1 10 100 1000

measured NA

fiber length [m]

AN = 0.02AN = 0.09AN = 0.17AN = 0.26AN = 0.34AN = 0.48

Fig. 2.26: Excitation dependent far field width of a 200 μm-PCS

In general, the mode coupling length is characterized as the distance in which a parameter has come closer by 1/e to the state of equilibrium. For example, this corresponds to the charging time constants of a capacitor. One cannot therefore say that EMD conditions exist after one coupling length. Depending on how large the tolerated deviations are, several coupling lengths have to be considered. Figure 2.27 shows the theoretical curve of a parameter.

100

200

300

400

500

600mode dependent fiber parameter [a.u.]

characteristic at Lc = 100 m

equilibrium mode value

10 100 100050 20020 500

fiber lenght [m]

parameter deviation for short fibers

parameter deviation for 1, 2 und 3 Lc

Fig. 2.27: Approximation of an optical parameter to the equilibrium value by mode coup-ling (schematically)


2.1.5.6 Leaky Modes

The significance of leaky modes has already been touched upon earlier. For the sake of completeness, it should be noted here again that light rays that lie above the critical angle of the total reflection do not entirely vanish but still contribute significantly to light propagation even after several 10s of meters.

Not until we examine the interaction of attenuation, mode-dependent attenu-ation, mode coupling and mode conversion and take leak modes into account, can we establish a model for the light propagation of SI polymer fibers that can at least qualitatively describe the experimentally observed behavior. In principle, the same processes take place in GI-POF; however there are basic differences:

With GI-POF, there is no core/cladding transition to serve as an essential cause for mode coupling, mode conversion, and mode-dependent attenuation.Fluorinated GI-POF are used in wavelength ranges in which Rayleigh scattering is less significant.To form the index profile, various zones of the fiber, as seen from the axis, are provided with varying concentrations of a dopant or a copolymer so that the attenuation usually gets a gradient. This is probably the most significant cause of mode-dependent attenuation in GI-POF.

Yabre and Zubia made comprehensive observations on mode propagation in GI-POF [Yab00a], [Yab00b], [Arr99], [Arr00].

The problem of mode coupling and mode conversion is sure to be very inte-resting for multi step index fibers. Bandwidths could result that are larger than what is theoretically expected. Some different theoretical investigations were made in cooperation between the POF-AC and the University of Bilbao (Spain). More details will be given in the fiber simulation chapter.

As the example of the multi-core fibers shows, mode-dependent attenuation can be used to exchange attenuation for bandwidth. Less attenuating cladding would reduce the overall attenuation of the POF, but more than likely also reduce the bandwidth (always assuming equilibrium mode distribution). The future will decide which parameter is of greater significance for users. If the transmission budget is sufficiently large, it would be possible to increase the bit rate though multi-level coding or by electrically compensating the dispersion so that a reduc-tion in attenuation is the minimum goal to be targeted in this field.

2.1.5.7 Dispersion in Optical Fibers

Dispersion refers initially to all processes that result in a difference in the transit times of various modes. One mode is thereby always a propagation condition of the light that is uniquely defined by the wavelength, polarization, and propagation path.

Differential delays between the various light components lead to a reduction in the modulation amplitude of higher frequencies. This makes the fiber a low-pass filter.


The bandwidth of a fiber communication transmission system is usually con-sidered to be the frequency for which the optical level of a sine-modulated signal has dropped by 3 dB. Strictly speaking this approach only applies to a Gaussian low-pass filter. This means that a pulse of insignificant width will correspond to the Gauss function after it has traveled the length of the fiber:

ff-0

20

2

e)f(P)f(P

where P(f) is the power of a random frequency f at the end of the measuring path, P0(f) is the launched power and f0 is a constant that describes the bandwidth. Figure 2.28 illustrates the process schematically.

b)

c)

a)P0(f)

e)P(f)

d)

time t

pulse response

Fig. 2.28: Effect of dispersion on a sine-wave signal

Curve 'a' shows the sine-modulated source optical signal (it must be noted that optical power can only take positive values). Figure 'b' shows how a single pulse approaching a Gaussian function after traveling through the fiber. This is a theore-tical borderline case because the Gaussian function extends from - to + , but the output pulse cannot begin before the input pulse has started. To measure the shape of the complete output signal, the input signal can be split into a series of pulses, as shown in Fig. 'c'. After traveling through the fiber, every pulse forms a Gaus-sian function of the respective height (Fig. 'd'). These have to be brought together again to achieve the result in curve 'e' (mathematically speaking, this is a convolu-tion of the input pulse with the so-called pulse response of the transmission link).


It is easy to see that the amplitude of the signal has decreased. Attenuation of the light has not been taken into consideration.

A short light pulse is briefly broadened when it travels the length of a fiber (Fig. 2.29) and this in turn reduces the transmission bandwidth.

optical inputpower

optical fiber

time time

optical output power

100 %

50 %

tin

100 %

50 %

tout

Fig. 2.29: Pulse broadening by passing an optical fiber

If Gaussian-shaped pulses are assumed, the result of the pulse broadening t is the square root of the difference of the squares of the input and output pulse width (FWHM full width at half maximum):

2in

2out ttt

The consequence of this broadening is that the time gap between the bits becomes smaller, that the pulses finally overlap and that the receiver can no longer differentiate between the two. The transmission bandwidth is limited as the light waveguide functions as a low-pass filter. The product of bandwidth and length characterizes the transmission capacity of a fiber. [Gla97] applies to Gaussian-shaped pulses:

Lt

44.0LB

Pulse broadening is caused by mode dispersion and chromatic dispersion. For multimode fibers it is necessary to consider the factors of material, modes and pro-file dispersion (in graded index fibers). Waveguide dispersion additionally occurs in singlemode fibers, whereas profile dispersion and mode dispersion do not.

All the kinds of dispersion appearing in optical fibers are summarized in Fig. 2.30. The mechanisms dependent on the propagation paths are marked in yellow, whereas the wavelength-dependent processes are marked in green.


dispersion

modal dispersion(multimode fibers)

chromatic dispersion(multimode and singlemode fibers)

profile dispersion

(multimode fibers)

material dispersion(multimode and

singlemode fibers)

waveguide dispersion (singlemode fibers)

polarization mode dispersion(singlemode fibers)

Fig. 2.30: Dispersion mechanisms in optical fibers

In regard to the fibers and applications dealt with in this book only mode and chromatic (material) dispersion play a role so that the following sections deal solely with these two effects.

2.1.5.8 Mode Dispersion

Since the light paths have different lengths, the pulses that have started simultane-ously arrive at different times at the fiber's output, a fact that leads to pulse broa-dening. Figure 1.29 shows the 'fastest' ( = 0) and the 'slowest' ( = max) rays.

max

L1

a

ncladding

ncore

max

1

2L2

Fig. 2.31: Deriving the difference in the transit time


The propagation times of the two different propagation paths are determined purely geometrically for:

c

nLA

nc2

L

n

nn

c

nLttt

n

n

c

L

sin

1

c

nL

c

nLt

c

nLt

core12N

cladding

1

cladding

claddingcorecore112mod

cladding

2core1

max

core1core22

core11

Figure 2.32 shows the dependence of the bandwidth on the numerical aperturewith which the light is launched. The assumption is that the far field, i.e. the angular distribution of the light in the fiber, will remain constant over the entire length of the sample (no modal coupling or conversion). For a PMMA standard fiber with an AN = 0.50, a differential delay of t 25 ns for 100 m is produced. The transit time is proportional to the square of the NA. From the above-men-tioned expression B 0.44/ tmod, a value of 15 MHz results for the bandwidth.

0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60numerical aperture

theoretical bandwidth [MHz]

100 m75 m

50 m

25 m

10 m

10

100

1,000

20

50

200

500

fiber-length:

Fig. 2.32: Bandwidth calculated as a function of the launch NA


The critical angle max of total reflection is determined by the ratio of both refractive indices (example, 1.492 for the core and 1.456 for the cladding):

)12.6:axistoangle(max.

77.4=0.976arcsin=1.492

1.456arcsin=

max

Thus, the relationship between both paths y and z is:

z = y/sin ( ) = y 1.0247

The NA of this fiber is determined by:

AN = (ncore2 - ncladding

2)0.5

= (1.4922 -1.4562)0.5

= 0.32

The pulse broadening for a fiber length L is derived as follows:

Transit time of the parallel-axis modes: t1 = L · n/c0

Transit time of the modes with max. angle: t2 = L · n/c0 · 1.0247 Differential delay: t = L · n/c0 · 0.0247 For example for 100 m, n = 1.492: t = 12.3 ns With the approximation B · t = 0.44 B = 33 MHz

Different NA lead to different bandwidths, whereby a doubling of the NA reduces the bandwidth to a quarter:

Theoretical bandwidth: AN = 0.60: 10 MHz 100 m AN = 0.50: 14 MHz 100 m AN = 0.40: 22 MHz 100 m AN = 0.30: 40 MHz 100 m

AN = 0.25: 57 MHz 100 m AN = 0.19: 97 MHz 100 m

To correctly calculate the theoretical bandwidth, it is just not sufficient to con-sider the two possible ray paths selected here. A very comprehensive description of mode propagation in POF is provided in [Bun99a]. In the ray model, each possible propagation direction is described by the two angles and (for an explanation of these angles, please refer to Fig. 2.11).

As far as transit time is concerned, only angle is of relevance. Figure 2.33 is an illustration from [Bun99a] of the zone of the guided rays and the leaky rays which themselves are again subdivided. Regardless of the size of , angle cannot exceed a particular maximum value so that a maximum possible differen-tial delay is the consequence.

Only the marked triangle contains not attenuated rays that are capable of propa-gation. If one assumes that all possible propagation paths have the same energy (UMD - uniform mode distribution), it can be seen that paths having a larger propagation angle are more probable than rays traveling parallel to the axis.


0 5 10 15 20

70

75

80

85

90

max

meridional rays

guided rays

Fig. 2.33: Possible rays in an optical fiber

As measurements of the far field (that is the power as a function of the angle to the fiber axis, measured in a sufficient large distance) of a POF shows, this is also reflected in the greater power obtained with larger angles. If the power is ex-pressed in relation to the solid angle element, a constant power density is found because larger angles cover a correspondingly larger arc. This is shown schemati-cally in Fig. 2.34.

angle to the fiber axis [°]

rel. power/solid angle

0.0

0.4

0.8

0.2

0.6

1.0

-30 -20 -10 0 10 20 30

UUMMDD

Fig. 2.34: Power distribution with UMD

The differential delay increases approximately by the square of the angle rela-tive to the fiber axis. If a short pulse having a mode distribution that correspon-ding to UMD is launched into the fiber input, an approximately rectangular pulse is generated at the output of the length of which corresponds to the approximate values shown above for the maximum differential delay. Figure 2.35 demonstrates the precise results for an assumed attenuation-free standard NA POF for the pulse form obtained after 10 m, 20 m, 50 m, and 100 m of ideal POF (from [Bun99a]).


0%

20%

40%

60%

80%

100%

0 5 10 15 20 25 30 35

norm. signal

100 m50 m20 m10 m

time [ns]

Fig. 2.35: Output pulses of a POF under UMD conditions ([Bun99a])

Real SI-POF provide considerably higher bandwidths. The main reason for this is the presence of mode-dependent attenuation in conjunction with mode mixing, as will be shown in the next chapter.

The differential delay t increases proportionally to a particular length Lc

(coupling length); for longer lengths, the increase is sub-linear (Fig. 2.36). The following holds true:

1 withLLforLt

LLforLt

c

c

whereby the exponent must be determined for each fiber. It is typically bet-ween 0.5 and 0.7. The coupling length Lc ranges between 30 m and 40 m for stan-dard SI-POF.

0.0

0.5

1.0

1.5

2.0

2.5

0 20 40 60 80 100 120 140 160 180 200

length [m]Lc

pulse broadening [a.U.]

l

l

in reality

Fig. 2.36: Schematically representation of the pulse broadening reflecting mode coupling effects


The impulse response of a 50 m long standard POF can be seen in Fig. 2.37. The half-value width of the impulse amounts to about 50 ns, i.e. only about 30% of the expected value. Furthermore, it is noticeable that the rear pulse edge drops more slowly. It is in this range that the higher modes lie which are attenuated very greatly by the mode-dependent losses. The dropping off of the rising edge can be explained by the effect of modal mixing.

5 20 3510 25 40 15 300

0.7

0.6

0.5

0.4

0.3

0.0

0.2

0.1

U [V] theoretical pulse shape

= 16 ns

t [ns]

== 55 nnss

Fig. 2.37: Real pulse shape for 50 m St.-NA-POF

Calculating the bandwidth of graded-index fibers is clearly more complex. Current studies in this field can be found in [Yab00a], [Yab00b] and [Arr99].

Profile dispersion occurs in graded-index profile fibers. It is the remainder of the mode dispersion that can no longer be compensated for and it depends on the relative refractive index difference , which in turn is wavelength-dependent. An optimization of the profile exponent can be accomplished for a certain wavelength for which d /d = 0. A profile exponent of g 2 causes a temporal broadening of:

2c

nLt

2core1

prof ,

in other words, a factor /2-reduced broadening of the pulse as compared with step index POF; for a typical graded-index POF this means a reduction by approximately 2 orders of magnitude [Blu98]. Mode dispersion or profile disper-sion can only be avoided by using singlemode fibers. As explained later on, due to the combination with the chromatic dispersion, certain polymer fibers, have some advantages as opposed to silica glass fibers.


2.1.5.9 Chromatic Dispersion

Chromatic dispersion describes the influence of the spectral width of a transmitter on a temporal broadening of the input pulse. This includes the material-dispersion and waveguide-dispersion types of dispersion. Both effects also occur in single-mode fibers. Waveguide dispersion is caused by the fact that light waves penetrate into the fiber cladding to various depths, depending on the wavelength of the light wave. Thus, the different speeds of the core and cladding parts result in pulse broadening. Since only a small portion of the light wave in higher modes of large diameter fibers spreads into the cladding, this effect is only considered for single-mode fibers.

However, even if only one mode is allowed to propagate, pulse broadening occurs due to material dispersion. Every light source has a spectral width > 0. The following applies for the pulse broadening tmax due to material dispersion:

ML²d

n²d

cLtmat

where is the spectral width of the transmitter n( ): wavelength-dependent refractive index, M( ): material dispersion parameters usually given in ps/km nm

Figure 2.38 shows the influence of material dispersion on pulse broadening, using polymer fibers as an example. Corresponding to the material dispersion, the longer wavelengths (red) propagate with a greater velocity than the shorter ones (blue).

t

wavelength

length

timeinput pulse

output pulse

Fig. 2.38: Temporal broadening as a result of material dispersion

The real influence of the chromatic dispersion from different polymer optical fibers to the system bandwidth will be shown in the next chapter which will contain detailed descriptions of the materials and fiber types.

2.2 Index Profiles and Types of Fibers 65

2.2 Index Profiles and Types of Fibers

After the theoretical descriptions on the properties of optical fibers in the section on the fundamentals of light propagation and the observations indicated above on mode propagation and the essential characteristics of fibers this following section will deal with concrete, available fibers. First, the different index profiles, as briefly mentioned in 1.1.6, will be introduced using examples.

The next section shows the historical development especially in regard to the different POF variants. Thereafter the important characteristics attenuation and bandwidth will be shown in a series of experimental results.

Three parameters are basically responsible for the actual properties of optical fibers. The core and cladding materials used determine the attenuation and chro-matic dispersion. The refractive index profile determines the mode dispersion and the core diameter is also responsible for the number of modes. Especially the core material and the index profile are at least recognizable from the name of the fiber, a designation method widely used in this book.

In the following section the historical development of the different polymer fibers is summarized. The POFs are dealt with in regard to their index profiles. Thereafter, different hybrid and glass fibers for short-range data transmission will also be introduced. The following chapter deals especially with the bandwidth of thick optical fibers since this characteristic is particularly important and also it makes the greatest demands on measurement techniques.

2.2.1 Step Index Profile Fibers (SI)

As was the case with silica glass fibers, the first polymer optical fibers were pure step index profile fibers (SI-POF). This means that a simple optical cladding sur-rounds a homogenous core. For this reason a protective material is always in-cluded in the cable. Figure 2.39 schematically represents the refractive index curve.

As already shown above, the refractive index step determines the numerical aperture (NA) and thus the acceptance angle. Some typical values are shown in Table 2.1. The refractive index of the core was always taken as 1.5, whereas the cladding has a correspondingly smaller refractive index. The last line is valid for wave guiding against air (n = 1). Here an acceptance angle of 90° is valid since the NA exceeds the value of 1.

ncore

jacket

optical cladding optical cladding

jacketcore

ncladd

Fig. 2.39: Structure of a step index profile fiber

66 2.2 Index Profiles and Types of Fibers

Table 2.1: Relationship between relative refractive index difference and numerical aper-ture (core refractive index = 1.50)

Relative Refractive-Index-Difference

Refractive Index of the Cladding

NumericalAperture

Acceptance Angle of the Fiber

0.22 % 1.497 0.10 6°

0.4 % 1.494 0.13 8°

0.8 % 1.488 0.19 11°

1.0 % 1.485 0.21 12°

1.5 % 1.478 0.26 15°

2.0 % 1.470 0.30 17°

2.7 % 1.460 0.35 20°

4.0 % 1.440 0.42 25°

5.8 % 1.413 0.50 30°

8.0 % 1.380 0.59 36°

12.0 % 1.320 0.71 45°

20.0 % 1.200 0.90 64°

33.3 % 1.000 1.12 90°

A larger acceptance angle of the fiber simplifies the launching of light, e.g. from a semi-conductor source. In addition, a high NA reduces the losses asso-ciated with fiber bending, as schematically illustrated in Fig. 2.40.

launched light rays

rays, exceeding the critical angle of total reflection behind the

bend

rays, guided behind the bend

bend radius

Fig. 2.40: Loss at fiber bends


Due to the effects of bending, the propagation direction of each individual ray is changed relative to the axis of the fiber. In the case of multi-mode fibers, a part of the rays is always extracted because the rays exceed the angle of total reflection at the interface between core and cladding. For fibers with a large NA, the effect of a change in angle for a certain amount of bending is not so significant so that the bending losses diminish. Likewise, when coupling fibers to each other (at connectors) the loss due to angle errors is less significant when there is a large numerical aperture.

A disadvantage of fibers with a large NA is the greater difference in time delay between the different light paths, and this in turn leads to a greater level of mode dispersion. This limits the bandwidth. In addition, the loss at coupling points in-creases if there is a gap between the abutting faces. Some advantages of larger or smaller numerical apertures are listed in Table 2.2.

Table 2.2: Influence of higher NA to various fiber parameters

Property of the Fiber Behavior with increasing NA

bending sensitivity becomes smaller

fiber coupled power becomes higher

connecting loss for fiber angular mismatch becomes smaller

connecting loss for axial fiber gap becomes higher

connecting loss for fiber axis lateral gap becomes higher

bandwidth becomes smaller

Silica glass multi-mode fibers usually have an NA of approximately 0.20. Silica glass fibers with polymer cladding have an NA in the range of 0.30 to 0.40 (some-times 0.65). The large refractive index difference between the materials that are used for the core and the cladding of polymer fibers allows significantly higher NA values. The majority of the initially produced SI-POF had an NA of 0.50 (e.g. [Asa96], [Esk97], [LC95]). SI-POF with an NA around this value are nowadays generally called standard NA-POF or standard POF for short. The bandwidth of such fibers is approximately 40 MHz for a 100 m long link (quoted as the band-width-length product 40 MHz · 100 m). For many years this was a completely satisfactory solution for most applications.

2.2.2 The Step Index Fiber with Reduced NA (low-NA)

However, when it became necessary to replace copper cables with polymer optical fiber to accomplish the transmission of ATM data rates of 155 Mbit/s (ATM: asynchronous transfer mode) over a distance of 50 m, a higher bandwidth was required for the POF. In the mid-nineties all three important manufacturers deve-loped the so-called low-NA POF.

POF with a reduced numerical aperture (low-NA POF) feature a bandwidth increased to approximately 100 MHz · 100 m because the NA has been reduced toapproximately 0.30. The first low-NA POF was presented in 1995 by Mitsubishi


Rayon ([Koi98]). Figure 2.41 shows that the fiber construction corresponds to the standard POF, the distinction being that the refractive index difference is smaller (approximately 2 %). Usually the same core material is used, but the cladding material has a modified composition.

ncore

jacket


jacketcore

ncladding

Fig. 2.41: Structure of a low-NA step index profile fiber

Unfortunately, practical testing showed that although this fiber met the require-ments of the ATM forum ([ATM96b]) with respect to bandwidth, it did not meet the requirements with respect to bending sensitivity. These requirements specify that for a 50 m long POF link the losses resulting from a maximum of ten 90° bends having a minimum bending radius of 25 mm should not exceed 0.5 dB. In order to meet both these requirements at the same time it became necessary to find a new structure.

2.2.3 The Double-Step Index Optical Fiber (DSI)

The double-step index POF features two claddings around the core, each with a decreasing refractive index (Fig. 2.42). In the case of straight installed links, light guiding is achieved essentially through the total reflection at the interface surface between the core and the inner cladding. This index difference results in an NA of around 0.30, similar to the value of the original low-NA POF.

ncore

jacket

inner / outer optical cladding

outer / inner optical cladding

jacketcore

ncladding1

ncladding2

Fig. 2.42: Structure of a double step index profile fiber


When fibers are bent, part of the light will no longer be guided by this inner interface. However, it is possible to reflect back part of the decoupled light in the direction of the core at the second interface between the inner and the outer cladding. At further bends, this light can again be redirected so that it enters the acceptance range of the inner cladding. The inner cladding has a significantly higher attenuation than the core. Light propagating over long distances within the inner cladding will be attenuated so strongly that it will no longer contribute to pulse propagation. Over shorter links the light can propagate through the inner cladding without resulting in too large a dispersion. A schematic illustration is shown in Fig. 2.43.

launched light rays

rays behind the bend

bend radius

1

2

3

4

1

1

2

2

3

4

rays, guided only by the inner cladding

rays, guided by the outer cladding behind the bend

rays, guided by the outer cladding over a limited distance

not guided rays behind the bend

Fig. 2.43: Operation of a bent double step index profile fiber

The first generation of DSI-POF primarily served the purpose of increasing the bandwidth of 1 mm fibers from 40 MHz · 100 m to 100 MHz · 100 m with an un-changed minimum bending radius of 25 mm. The respective applications are to be found in LANs and home networks.

The fiber producers offer these fibers under the same type names as the original “real” low-NA fibers. It has since become standard procedure to call the fibers low-NA and to indicate DSI as the index profile.

Currently, another goal is being pursued: the bandwidth of standard POF is sufficient for applications in vehicle networks, but the bending radius should be reduced. Presently being discussed are POFs, the index steps of which correspond to a NA of 0.50 or 0.65 respectively to the inner and outer cladding. The bending radius can thus almost be halved.


2.2.4 The Multi-Core Step Index Optical Fiber (MC)

As described above, the requirements of high bandwidth and low sensitivity to bending are difficult to accomplish together within one and the same fiber having a diameter of 1 mm. Fibers with a smaller core diameter can solve this problem since the ratio to the fiber radius is larger for the same absolute bending radius. However, this contradicts the requirements for easy handling and light launching. A PCS with a core diameter of 200 m and an AN = 0.37 permits, for example, a bending radius of 5 mm with very low bending losses.

As a compromise, Asahi developed a multi-core fiber (MC-POF, see [Mun94], [Mun96] and [Koi96c]). In this fiber many cores (19 to over 200) are put together in production in such a way that together they fill a round cross-section of 1 mm diameter.

First, the individual fibers are all perfectly round and each has its own optical cladding. Only a certain share of the total cross-section of the bundle enters the cores guiding the light, since the cladding areas and the spaces between the fibers have to be accounted for. Figure 2.44 shows the parameters which mark the per-centage of the filled-in area. The number N here indicates how many fibers lie next to each other over a diameter while n indicates the entire number of fibers.

R

dm

Rdm

N = 1

N = 5

n = 19

r

Fig. 2.44: Schematically arrangement of cores in a MC-POF

In the figure, R denotes the radius of the complete fiber (typically 0.5 mm) and d the thickness of the optical cladding (e.g. 5 μm). Let us assume first of all that the individual cores are arranged in a hexagonal shape with N = 2z + 1 cores positioned next to each other.

The next Fig. 2.45 shows how the arrangement for fibers is changed for z = 1 to 5. While these sketches can give a clear definition of the number of fibers that can be arranged within a circular shape, for smaller and smaller individual cores the possibilities are more complex. The arrangement at the bottom right shows one possible deviation. For the first five arrangements the number of indi-vidual fibers is calculated as follows:

n = 3z2 + 3z + 1.


It follows that the individual radius r is: r = R/N = R/(2z + 1).

N=3

N=9

N=5

N=11

N=7

N´=11

Fig. 2.45: Possible circular arrangements of cores in a MC-POF

In Table 2.3, the degree of coverage of the circle area is calculated for the cases shown. First, the number of individual cores is calculated from z. The radius r results from the overall radius of the fiber (here always 500 μm). Parameter ta indi-cates what percentage of the total circular area is covered by the individual circles (for the hexagonal arrangement of an infinite number of circles a maximum of 90.69 % of the area can be covered). When calculating parameter tb, the fact that part of the cross-section is lost to the optical claddings (all uniformly 5 μm thick) is taken into account.

Table 2.3: Core cross area degree of coverage for MC fibers (ideal)

z: N: n: r: ta: tb:

0 1 1 500 μm 100.00 % 98.01 %

1 3 7 167 μm 77.78 % 73.18 %

2 5 19 100 μm 76.00 % 68.59 %

3 7 37 71.4 μm 75.51 % 65.31 %

4 9 61 55.6 μm 75.31 % 62.36 %

5 11 91 45.5 μm 75.21 % 59.57 %

11´ 85 49.3 μm 82.47 % 66.57 %

6 13 127 38.5 μm 75.15 % 56.88 %

7 15 169 33.3 μm 75.11 % 54.27 %

8 17 217 29.4 μm 75.09 % 51.73 %

14 29 631 17.2 μm 75.03 % 37.82 %

- - - 90.69 % -


Figure 2.46 shows the proportion of core area tb as depending on the number of cores for four different thickness’ of the optical cladding.

0%

20%

40%

60%

80%

100%

1 7 19 37 61 91 127 169 217

dm = 5 μm

dm = 10 μm

dm = 20 μm

dm = 30 μm

use of the total cross area tb

number of single cores

Fig. 2.46: Proportion of core area for different cladding thickness

As can be expected, the proportion of the overall covered area decreases with an increasing number of cores because the proportion of cladding area will be-come larger and larger. A certain minimum thickness of cladding is necessary for it to be able to fulfill its function and still be technologically feasible. The four individual data points show the case of the optimized fiber arrangement with 85 individual cores in accordance with Fig. 2.45.

Given a minimum thickness of the optical cladding between 5 μm and 10 μm, these considerations indicate that a maximum number of some 100 single cores should be used, in which case the proportion of useable area will hardly exceed 70 %. It is easy to conclude that a smaller proportion of useable core area would lead to an increase in the losses encountered when connecting transmitters to, and fibers between each other.

Fig. 2.47: 37 core POF with deformed single cores (schematically)


Practical experience shows that a better utilization of the area can be achieved. During the manufacturing process the fibers are placed together at higher tempe-ratures which means that they change their shape and thus reduce the gaps bet-ween the fibers. Apparently, the resulting deviations from the ideal round shape do not play a significant role in light propagation (the causes for this are not yet completely understood; some points worth discussing can be found in the chapter on light propagation in POF). Figure 2.47 shows a schematic illustration of the cross-section of a fiber with 37 cores, such as e.g. in [Tesh98]. Data of available MC-POF and -GOF are grouped together later.

Figure 2.48 shows the refractive index profile of a MC-POF, shown as a cross-section through the diameter of the fiber. The index steps correspond to those of a standard POF.

ncore

jacket

optical cladding

jacketcores

ncladding

optical cladding

Fig. 2.48: Structure of a step index multi core fiber

Since the bandwidth only depends on the NA for SI fibers, it should be possible to measure values comparable to the standard POF. However, the fact is that the measured values are actually significantly higher, which has been explained in the chapter 2.1.5.2 discussing mode-selective attenuation mechanisms.

Glass fibers are also produced for use in many areas as fiber bundles. In lighting technology fiber glass bundles with a large NA are widely spread. (The lighting of the headlight outer ring at BMW via such a fiber bundle is well-known.) In the meantime, such fibers are also available for data communication ([Lub04b]).

2.2.5 The Double Step Index Multi-Core Fiber (DSI-MC)

In the MC-POF, too, an increase in bandwidth was achieved by reducing the index difference. Due to the smaller core diameters it was still possible to avoid an increase in bending sensitivity.

Even better values were achieved with individual cores having a two-step optical cladding such as illustrated in Fig. 2.49. The principle is the same as in the double-step index POF with an individual core. In this case a bundle with single cladding is completely surrounded by a second cladding material (“sea/islands” structure).


ncore

jacket

outer / inner optical cladding

jacketcores

ncladding1ncladding2

Fig. 2.49: Structure of a double step index profile multi core POF

2.2.6 The Graded Index Optical Fiber (GI)

When using graded index profiles (GI) an even greater bandwidth becomes pos-sible. In these profiles, the refractive index continually decreases (as a gradient), starting from the fiber axis and moving outwards to the cladding. Of particular interest are profiles that follow a power law (remember chapter 1.4.1).

g

radiuscore

axisfibertodistance-1n=nindexrefractive

axisfiber

The parameter g - often also - is characterized as the profile exponent. When g = 2 one speaks of a parabolic profile. The borderline case of the step index profile fibers corresponds to g = . The parameter states the relative refractive index difference between the maximum core and the cladding refractive index. Figure 2.50 shows a parabolic index profile.

ncore

jacket

optical cladding

jacketcore

ncladding

optical cladding

Fig. 2.50: Structure of a graded index profile fiber

Due to the continually changing refractive index, the light rays in a GI fiber do not propagate in a straight line but are constantly refracted towards the fiber axis. Light rays that are launched at the center of the fiber and do not exceed a certain angle are completely prevented from leaving the core area without any reflections occurring at the interface surface. This behavior is illustrated schematically in Fig. 2.51. The geometric path of the rays running on a parallel to the axis is still significantly smaller than the path of rays that are launched at a greater angle.

However, as can be seen, the index is smaller in the regions distant from the core. This means a greater propagation speed. In an ideal combination of para-meters the different path lengths and different propagation speeds may cancel each


other out completely so that mode dispersion disappears. In reality, this is only possible in approximation. It is possible, however, to increase bandwidths by two to three orders of magnitude compared with the SI fiber.

n

step index profile fiber graded index profile fiber

n

Fig. 2.51: Comparison of step and graded index profile (see also chapter 2.1.1)

When considering not only the pure mode dispersion but also chromatic disper-sion, i.e. the dependence of the refractive index on the wavelength and spectral width of the source, an optimum index coefficient 'g' deviating from 2 is achieved. This has been the subject of comprehensive investigations by the research group around Prof. Koike ([Koi96a], [Koi96b], [Ish00], [Koi97a], [Koi96c], [Koi98] and [Ish98]). In [Ish00] and [Koi00] the significance of this effect is particularly pro-nounced (see also Chapter 2). Due to the smaller chromatic dispersion of fluori-nated polymer compared with silica, the bandwidth of GI-POF theoretically achie-vable is significantly higher than that of multi-mode GI silica glass fibers. In parti-cular, this bandwidth can be realized over a significantly greater range of wave-lengths. This makes the PF-GI-POF interesting for wavelength multiplex systems. However, in this case the index profile must be maintained very accurately, a requirement for which no technical solution has as yet been provided.

Another factor involved in the bandwidth of GI-POF is the high level of mode-dependent attenuation ([Yab00a]) compared to silica glass fibers. In this case modes with a large propagation angle are suppressed resulting in a greater band-width. An example is the simulation that was carried out in [Yab00a]: the band-width of a 200 m long PMMA-GI-POF increases from 1 GHz to over 4 GHz, taking into account the attenuation of higher modes. This is also confirmed in practical trials. Mode coupling is less significant for GI fibers than it is for SI fibers since the reflections at the core-cladding interface do not occur.

2.2.7 The Multi-Step Index Optical Fiber (MSI)

Following the many technological problems experienced in the production of gra-ded index fibers having an optimum index profile that remains stable for the dura-tion of its service life, an attempt was made to approach the desired characteristics with the multi-step index profile fiber (MSI-POF). In this case the core consists of many layers (e.g. four to seven) that approach the required parabolic curve in aseries of steps. Here a “merging” of these steps during the manufacturing process may even be desirable. A diagram of the structure is shown in Fig. 2.52.


ncore

jacket


jacketcore

ncladding

Fig. 2.52: Structure of a multi step index profile fiber

In this case light rays do not propagate along continually curved paths as in the GI-POF, but on multiple diffracted paths as demonstrated in Fig. 2.53. However, given a sufficient number of steps, the difference to the ideal GI profile is relati-vely small so that large bandwidths can nevertheless be achieved. MSI-POF were presented in 1999 by a Russian institute (Tver near Moscow [Lev99]) and by Mitsubishi (ESKA-MIU, see [Shi99]). In the meantime, other companies are pro-ducing such fibers which are often called GI fibers. These GI and MSI fibers are classified in the same class of standards, e.g. A4e.

n

Fig. 2.53: Light propagation in the MSI-POF

2.2.8 The Semi-Graded Index Profile Fibers (Semi-GI)

A relatively new version of index profiles are fibers which have a gradient with a slightly varying index above the core cross section, but do have an optical clad-ding with a great index step as shown in Fig. 2.54 ([Sum00], [Sum03], [Ziem05f] and [Ziem06i]).

ncore

jacket


jacketcore

ncladding

Fig. 2.54: Structure of a semi-graded index profile fiber


At first sight this variety of fiber has enormous advantages. Light which propa-gates within the gradient is only subject to very little mode dispersion. If a ray of light has a greater propagation angle, e.g. after being bent, then it continues to be led to the core-cladding interface layer through total reflection. However, these rays do have a very much higher mode dispersion. Figure 2.55 shows how light spreads theoretically and what consequences this has for the pulse response.

input output

„GI“-modes

„SI“-modest

Fig. 2.55: Light propagation in semi-graded-index profile fibers theoretically

In principle, two different groups of modes can be seen in the picture. The paths designated as GI modes do not touch the cladding and only show a very slight difference in propagation times. The shares designated as SI modes are completely reflected at the core-cladding interface layer. These light paths are also bent in the core, but the light path, now very much longer, can no longer be com-pensated for in the outer areas by the lower refractive index. With very high data rates the second mode group is drawn out so widely that it is presented solely as a kind of DC offset in the eye diagram. At the POF-AC a data rate of 1 Gbit/s was transmitted over 500 m of a GI PCS fiber with a PRBS signal ([Vin05a]). Data rates up to 3 Gbit/s could be attained with a small surface APD receiver ([Kos95]). In order to do justice to the complex behavior of the semi-GI POF, corresponding modulation formats should be selected.

2.2.9 An Overview of Index Profiles

Figures 2.56 through 2.58 again show all index profiles described in an overview. Due to the wide range of possibilities offered in polymer chemistry further deve-lopments are certainly to be expected. For example, multi-core graded fibers, fibers with special cladding for a reduction of the losses at the core/cladding inter-face or to increase the bandwidth or even multi-core fibers with different indivi-dual cores are all conceivable. In the following figures POF variants are shown with typical parameters.


Low-NA-POF AN = 0.30 100 MHz 100 m

DSI-POF AN = 0.30 100 MHz 100 m

SI-POF AN = 0.50 40 MHz 100 m

Fig. 2.56: POF with single core and step index profile

Single-core fibers with diameters between 125 μm and 3 mm are available from different manufacturers at a reasonable price and in robust quality. Most of the polymer optical fibers used in practical applications are of these types.

MC-SI-POF e.g. 200 cores AN = 0.30 100 MHz 100 m

MC-DSI-POF e.g. 37 cores AN = 0.19 400 MHz 100 m

Fig. 2.57: POF with multiple cores and step index profile

MC fibers are available from various manufacturers. They are deployed in applications ranging from high data rates transmission systems through to optical image guides. Because of the short lengths produced, the prices are still signifi-cantly above expectations. However, further developments in this field can be ex-pected in the future.

GI-POF AN = 0.20 2 GHz 100 m

MSI-POF AN = 0.30 500 MHz 100 m

Fig. 2.58: Polymer fibers with graded index and multi step index profile

Graded index as well as multi-step index profile POF are commercially avai-lable today. Laboratory experiments and a series of practical installations in Japan and Europe, (e.g. [Mös04]) show the great potential in regard to the bit rates possible. Asahi Glass introduced them into the market around 2001. Lucent Tech-nologies, later called OFS and trading under the name of Chromis Fiberoptics as of 2004 ([Whi04], [Park05a]), also announced the possibility of producing large amounts of GI POF in case of demand.

2.3 The Development of POF 79

In Europe, fibers by Nexans are manufactured in Lyon ([Gou04]). All three fi-bers will consist of the fluorinated polymer material CYTOP®. The core diameter of the LucinaTM Fiber by Asahi Glass is 120 μm with an AN = 0.28. A protective cladding made from PMMA and measuring 500 μm is placed around an area of fluorinated polymer outside the core profile. The duplex cable has external dimen-sions of approximately 3 by 5 mm. The lowest attenuation achieved to date is ap-prox. 15 dB/km for a wavelength of 1,300 nm. The specified value is < 50 dB/km for 700 nm - 1,300 nm.

There has also been significant progress in the manufacture of GI or MSI-POF respectively on a PMMA basis (see Section 2.3.4).

2.3 The Development of Polymer Optical Fibers

The following sections will describe the polymer fibers presented so far, whereby particular attention will be paid to the chronological sequence of the develop-ments. Section 2.4 supplements these observations with some types of multimode glass fibers which were not discussed in the first edition.

2.3.1 Looking back

The first POF were manufactured by DuPont as early as the late sixties. Due to the incomplete purification of the monomer materials used, attenuation was still in the vicinity of 1,000 dB/km. During the seventies it became possible to reduce losses nearly to the theoretical limit of approximately 125 dB/km at a wavelength of 650 nm. At that point in time glass fibers with losses significantly below 1 dB/km at 1,300 nm/1,550 nm were already available in large quantities and at low prices. Digital transmission systems with a high bit rate were then almost exclusively used in telecommunications for long-range transmissions. The field of local com-puter networks was dominated by copper cables (either twisted-pair or coaxial) that were completely satisfactory for the typical data rates of up to 10 Mbit/s com-monly used then. There was hardly any demand for an optical medium for high data rates and small distances so that the development of the polymer optical fiber was slowed down for many years. A significant indicator for this is the fact that at the beginning of the nineties the company Höchst stopped manufacturing polymer fibers altogether.

During the nineties, after data communication for long-haul transmission had become completely digitalized, the development of digital systems for private users was commenced on a massive scale. In many spheres of life we are being in-creasingly confronted with digital end user equipment. The CD player has largely replaced analog sound carriers (vinyl records and cassettes). The MP3 format is leading to a revolution in music recording and distribution. The DVD (Digital Video Disc) and large hard disk drives could lead to the replacement of the analog video recorder within a few years. Even today more digital television programs

80 2.3 The Development of POF

are available than analog programs. Decoder boxes have become standardized(MPEG2 format) and will be integrated into television sets in the future. More and more households are using powerful PC and digital telephone connections (ISDN). With offers such as T-DSL (ADSL technology provided by Deutsche Telekom AG) as well as fast internet access via satellite or broadband digital ser-vices on the broadband cable network, private users are being offered access to additional digital applications even before the start of the new millennium. Likewise, in the automotive field the step towards digitalization has long been made. CD changers, navigation systems, distance-keeping radar and complex con-trol functions are increasingly part of the standard equipment being provided in all classes of vehicles. The development of electronic outside mirrors, fast network connections even from within an automobile as well as automatic traffic guidance systems will ensure a further increase in the range of digital applications for the motor vehicle. All these examples demonstrate that completely new markets for digital transmission systems are being developed for short-range applications. Polymer optical fibers can meet many of these requirements to an optimum degree and are therefore increasingly of interest.

A significant indicator for this development is the history of the International Conference for Polymer Optical Fibers and Applications which has been taking place annually since 1992 and represents the most significant scientific event in this specialized field. Many of the developments described below were presented for the first time at these conferences.

2.3.2 Step Index Polymer Fibers

The SI-POF is the oldest variant of all polymer fibers. Its development goes back to the beginning of the 1960’s, i.e. in a period when silica glass fibers were being developed. Today the SI-POF is by far the most common POF variant. In Table 2.4 data from different publications on this fiber type are summarized - without claiming to be complete.

Table 2.4: Published data of SI-POF

Ref. Year Producer Product Øcore

μm

Attenuation

dB/kmatnm

NA Remarks

[Min94] 1963 Du Pont CROFON - 1.000 650 st. first POF

[Koi97a] 1964 Du Pont - 500 650 st.

[Koi96c] 1968 Du Pont - 500 650 st. first SI-POF

[Sai92] 1976 Mitsubishi Eska - 300 650 st.

[Min94] 1978 Mitsubishi Super Eska - 300 650 st.

[Koi95] 1982 NTT - 55 568 st.

[Sai92] 1983 Mitsubishi Eska Extra - 124 650 st. 4 MHz km

[Sai92] 1983 Mitsubishi Eska Extra - 65 570 st.

[Koi95] 1983 Mitsubishi 1000 110 570 st.

[Min94] 1984 Mitsubishi Eska Extra - 150 650 st.


Table 2.4: Published data of SI-POF (continued)

Ref. Year Producer Product Øcore

μm

Attenuation

dB/kmatnm

NA Remarks

[Koi95] 1985 Asahi - 80 570 st.[Sai92] 1991 Mitsubishi Eska Extra - 125 650 st. up tp 85°C

[Sai92] 1991 Mitsubishi Eska Extra - 65 570 st.[Koi95] 1991 Hoechst 1000 130 650 st.[Tesh92] 1992 Asahi Luminous-F - 175 660 0.50 310 MHz 10m

AN, LED=0.50, 105°C

[Tesh92] 1992 Asahi X-1 - - - 0.37 540 MHz 10mAN, LED = 0.50

[Tesh92] 1992 Asahi X-2 - - - 0.28 >1.000 MHz 10m AN, LED = 0.50

[Eng96] 1992 Höchst EP51 970 190 650 st. 90 MHz 100 m with 650 nm LED

[Kit92] 1992 Mitsubishi Eska Premier 1000 135 650 0.51 up to 85°C

[Lev93] 1993 CIS Sveton MN-Series, Grade U

200-600

150 650 0.45 up to 70°C

[Lev93] 1993 CIS Sveton MF-Series, Grade U

200-1000

120 650 0.48 up to 70°C

[Non94] 1994 Sumitomo n. a. 480 150 650 0.51 200 MHz 50mn=0.055

[Koe98] 1998 Mitsubishi n. a. 1000 110 650 0.47 80 MHz 100 m [Mye02] 2002 Dig. Optr. n. a. 1000 - - 0.50 2003 announced

[Luv03] 2003 Luvantix SI type 1000 160 650 0.40 200 MHz bandwidth

[Nuv04] 2004 Nuvitech Nuvilight 1000 250 650 0.38 for illumination

[Luc05] 2005 Luceat SI-Type 1000 150 650 0.46 30 MHz 100 m [Wal05] 2005 Nanoptics A-POF 1000 100 650 - conception

[Hai05] 2005 Huiyuan SI-POF 1000 300 650 - coextrusion

[Zie06h] 2006 Luceat SI-POF 1000 135 65

650520

0.50 from preform

400 450 500 550 600 650 700 750 800

wavelength [nm]

attenuation [dB/km]

50

100

1,000

2,000

5,000

500

200

Fig. 2.59: Attenuation of different standard-NA SI-POF (measurement by POF-AC)


It was not until about 1980 that technology made possible the production of POF which came relatively close to the theoretical attenuation minima. Initial problems with the service life and with certain mechanical loads were quickly solved with on-going developments. In Fig. 2.59 the spectral attenuation curves of three SI-POFs are shown (data sheet information). All three fibers from Japanese manufacturers are close together. The visible differences may possibly be due to different methods of measurement.

Most manufacturers offer SI-POFs in different diameters. In [Zub01b] and [Nuv04] the properties of these fibers are compared (Table 2.5).

Table 2.5: Attenuation of POF with different diameter

Attenuation [dB/km]

diameter [μm] 250 500 750 1.000 Mitsubishi < 700 < 190 < 180 < 160 Toray < 300 < 180 < 150 < 150 Asahi Chem. n. a. < 180 < 180 < 125 BOF < 150 < 150 < 150 < 150 Optectron < 150 < 150 < 150 < 150 Nuvitech < 350 < 250 < 250 < 250

For Toray fibers, the losses of fibers with different diameters are listed in the data sheet and are shown in Fig. 2.60.

400 500 600 700 800 900

attenuation [dB/km]

wavelength [nm]

core 750 μm

core = 500 μm

core = 250 μm100

1000

10000

30

300

3000

Fig. 2.60: Attenuation of different PMMA-SI-POF by Toray

With a few exceptions the losses for all fiber diameters are similar. Some reasons for the increase in attenuation with thinner fibers could be that either the high attenuation of the optical cladding plays a greater role or that more stress is exerted on the thin fiber during manufacture. A fiber with a ¼ mm core diameter


has only one sixth the thermal capacitance. When the cladding and opaque jacketare applied this fiber is necessarily warmer. The process temperatures during ma-nufacture can indeed lie clearly above the glass transition temperature.

The youngest manufacturer of PMMA SI-POF is the Italian company Luceat. Here fibers for diverse applications, mainly in mechanical engineering, are pro-duced. The highest quality is still in the developmental stage. A comparison of the measured values of Luceat fibers (POF-AC 2006, [Ziem06h]) with the values from [Wei98], more or less the POF reference curve up until now, is shown in Fig. 2.61.

100

80

500

300

200

50

60

[Wei98]

attenuation [dB/km]

400 450 500 550 600 650 700wavelength [nm]

Luceat

Fig. 2.61: Attenuation of SI-POF by Luceat (2006)

In the area of 520 nm this fiber is even somewhat better that the data of the best fibers so far. Thanks to the availability of reasonably priced and fast green LEDs this advantage can be assessed very highly. As part of the European POF project POF-ALL (see www.ist-pof-all.org) the transmission of a 10 Mbit/s data stream was able to be demonstrated over 425 m (see System Chapter).

2.3.3 Double Step Index Profile Polymer Fibers

We have already discussed the principle idea of a double step index profile POF. All three important Japanese manufacturers presented such fiber types around 1995. After the expectations that ATM would become the dominating network technology in the home were not fulfilled, these fibers have more or less become niche products today, albeit at relatively high prices. Today in many areas there is a demand for data rates which require the use of these fibers instead of the normal SI-POFs. Technically, DSI-POFs are on a comparable level and would hardly be more expensive than SI-POFs when produced in high volumes.


Table 2.6 compares the properties of DSI-POFs of the three manufacturers ([Mit01], [Nich03], [LC00b]).

Table 2.6: Overview of DSI-POF

Mitsubishi Toray Asahi

MH4001 PMU-CD1001 AC1000(I)

diameter [μm] 980 1000 ± 45 1000 ± 60 attenuation (650 nm) [dB/km] 160 170 160 numerical aperture - 0.30 0.32 0.25 bandwidth MHz km 10 >10 15 temperature range [°C] -55 .. +75 -20 .. +70 -40 .. +70 bend radius [mm] 25 - 25

We would like to point out once again that the DSI-POFs are usually offered now as before as low NA POF. In the first few years manufacturers did not pro-vide any information at all about the double cladding structure. In [Eng98b] the double cladding structure was proven quite early on the basis of measurements of the far field and with optical microscopy. In Fig. 2.62 you can see the far field dis-tributions for different fiber lengths measured with the inverse far field method at the FH Gießen/Friedberg.

-30 -20 -10 0 10 20 30 [°]

1 m 10 m 50 m 90 m = 594 nm

1.0

0.8

0.2

0.0

0.6

0.4

Popt

Fig. 2.62: Inverse far field measurement of a DSI-POF

You can clearly see that after short distances much light from the interface layer between inner and outer cladding is still guided. After 50 m these shares have disappeared and the angle distribution corresponds to a true low NA POF.

Figure 2.63 shows two microscope photos of DSI-POF (Univ. of Ulm). Both optical claddings can be easily recognized.

At the 2003 POF Conference Mitsubishi was the first manufacturer to present the actual structure. The effect of suppressing higher modes by high attenuation of the inner cladding was also confirmed theoretically and experimentally. For example, Asahi gives a value of 6000 dB/km at 650 nm for the losses in the inner cladding.


Fig. 2.63: Double cladding structure of a POF

2.3.4 Multi-Core Polymer Fibers

Since 1994, polymer fibers as multi-core fibers have been introduced, e.g. in [Tesh98], [Mun94], [Asa97] and [Tesh98]. Table 2.7 shows a few parameters from these publications.

Table 2.7: Multi-core POF (Asahi Chemical)

Type Ref. No. of Cores

Structure NA Attenuation at 650 nm

Bandwidth

NMC-1000 POF´94 19 SI 0.25 125 dB/km 170 MHz 100 m PMC-1000 Data´96 217 SI 0.15 270 dB/km n. a. MCS-1000 Data´97 217 SI - 320 dB/km n. a. - POF´98 37 DSI 0.19 155 dB/km 700 MHz 50 m - POF´98 37 DSI 0.25 160 dB/km n. a. - POF´98 37 DSI 0.33 160 dB/km n. a. NMC-1000 Data´98 37 DSI 0.25 160 dB/km 500 MHz 50 m PMC-1000 Data´98 37 DSI 0.19 160 dB/km n. a. - POF´98 217 SI 0.50 160 dB/km n. a. - POF´98 217 SI 0.33 160 dB/km n. a.

The MC-POF features a noticeably reduced sensitivity to bending and only in-significantly increased attenuation as well as a significantly increased bandwidth compared to single core fibers, this being due to the possibility of smaller nume-rical apertures. Whether these fibers can be produced at the same price is still an open question. Should this be possible, data rates of 500 Mbit/s up to 1 Gbit/s over 50 m can easily be achieved in commercial applications. At the POF-AC a data rate of over 1 Gbit/s over 100 m MC-POF has already been achieved.

At present, only Asahi chemical offers MC-POF for data communication while other manufacturers offer this kind of fiber for lighting purposes or also as image guiding fiber. The following photos show the cross-sections of the three, presently available MC-POFs with 37, 217 and 631 cores (the 19 core variant is no longer available).


Fig. 2.64: Photo by microscope of MC-POF, 37, 217 respectively 631 cores

An overview of the technical data of the four different MC-POFs is summa-rized in the following Table 2.8 (Nichimen data sheets). The PMC 1000 permits the highest data rates on the basis of experiments conducted so far since it posses-ses a DSI structure.

Table 2.8: Data of MC-POF

Parameter Unit MCQ-1000 MCS-1000 NMC-1000 PMC-1000

number of cores - 613 217 19 37

single core μm 37 60 200 130 core material - PMMA PMMA PMMA PMMA cladding material - fluoro polymer FMA-copolymer 2nd cladding material - - VDF-copolymer NA - 0.5 0.05 0.50 0.25 0.19

fiber core mm 1.0 0.06 1.0 0.06 1.0 0.06 1.0 0.06

cable mm 2.2 0.07 2.2 0.10 2.2 0.10 2.2 0.10

jacket PE PE (black) PE (black) PE (black)

attenuation1) dB/km <200 320 1634) 1634)

bit rate (50 m) Mbit/s n. a. n. a. 3505) 5005)

temperature °C -40 .. +60 -40 .. +60 -40 .. +70 -40 .. +70

bend loss2) dB <0.1 <0.23) <0.13) <0.13)

1) Cut back 12 - 2 m, at 650 nm 4) 650 nm monochromatic light 2) R = 3 mm, 180°, no stress 5) 650 nm LD, BER = 10-12

3) R = 3 mm, 360°, Launch-NA: 0.2

We do have to point out one special feature of these four MC POFs: the fibers are tightly bound in the cable as opposed to the individual fibers in a fiber glass bundle or other MC POFs used in lighting technology. The share of the core surface is not only enlarged, but it is considerably easier to work the fibers. These strands can be mounted like quite normal 1 mm SI-POFs.

The two enormous advantages of MC-POF, namely the high band width and the low bending losses, have in the meantime been somewhat qualified since con-siderably cheaper GI-POFs on a PMMA basis have become available. The latter will be treated in the next paragraph.


2.3.5 Multi-Step Index Profile and Graded Index Profile Fibers

The greatest bandwidths of all fibers - with the exception of the singlemode fibers - are shown by graded index profile fibers. They have been used extensively for some time in the field of silica glass fibers and are a standard. In the USA, predo-minately fibers with a core diameter of 62.5 μm are used, whereas in Europe and most other countries fibers with a core diameter of 50 μm are used. This diameter is nevertheless 5 to 6 times greater than with singlemode fibers whereby the plug costs are greatly reduced and the coupling of lasers is also easier. The bandwidth-length product (BLP) of these multimode glass fibers lies in the range of 200 to 500 MHz · km. For the transmission of 10 Gbit/s a new fiber specification with a BLP of 2,000 MHz · km at a wavelength of 850 nm is even being developed (for example, see [Oeh02] and [Geo01]).

The advantages of the large core diameter and high bandwidth would be an optimal combination with POFs. Furthermore, numerous problems with the core-cladding interface area would cease to exist with GI fibers since the light guiding would take place exclusively in the core. Glass GI fibers are produced by applying many layers of a SiO2-GeO2 mixture with different compositions to a quartz glass pipe. Finally, the fiber is drawn (several 100 km) out of such a preform. Unfor-tunately, this is not possible with POFs. The different methods and combinations of materials with which attempts have been made to produce GI-POF will be des-cribed further on. Since GI fibers are difficult to produce - as we shall describe later on - a series of multi step index profile POFs have been introduced. These MSI-POFs also offer high bandwidth depending on the number of steps. For now, the optical characteristics are summarized here.

Table 2.9 shows an overview of the values for PMMA-based GI, MC and MSI fibers. To the best knowledge of the author, all PMMA-GI-POF published to date are produced by doping, whereas only MSI-POF are produced in a co-polymeri-zation process.

Table 2.9: Published data of PMMA-GI-, MSI- and MC-POF (IGPT: interfacial gel poly-merization technique; PFM: preform method)

Ref. Year Producer Material Øcore

μm

Attenuation

dB/kmatnm

NA Remarks

[Koe98] 1998 � �� 1� � � 11� � � � � �� 8� � � � 100 m

[Koi95] 1982 Keio Univ. MMAcoVPAc - 1070 670 - first GI-POF

[Koi96c] 1990 Keio Univ. PMMA - - - - 670 nm: 300 MHz km

[Koi95] 1990 Keio Univ. MMA co VB - 130 650 -

[Koi90] 1990 Keio Univ. MMA-VB - 134 652 - IGPT, 260 MHz 1 km

[Koi90] 1990 Keio Univ. MMA-VPAc - 143 652 - IGPT,125 MHz 1 km

[Koi92] 1992 Keio Univ. PMMA 200-1500 113 650 - IGPT, 1,000 MHz km

[Koi92] 1992 Keio Univ. PMMA 200-1500 90 570 -

[Non94] 1994 Sumitomo PMMA 400 160 650 0.26 n=0.014, 8GHz 50m

[Shi95] 1995 BOF PMMA 600 300 650 0.19 3 GHz 100 m


Table 2.9: Published data of PMMA-GI-, MSI- and MC-POF, continued

Ref. Year Producer Material Øcore

μm

Attenuation

dB/kmatnm

NA Remarks

[Ish95] 1995 Keio Univ. PMMA-DPS 500-1000 150 650 - 585 MHz km

[Koi97b] 1997 Keio Univ. PMMA - - - - 2 GHz 100 m

[Tak98] 1998 Kurabe PMMA 500 132 650 - 2 GHz 100m, PFM

[Tak98] 1998 Kurabe PMMA 500 145 650 - 2 GHz 90m, PFM

[Tak98] 1998 Kurabe PMMA 500 159 650 - 680 MHz 50m, PFM

[Tak98] 1998 Kurabe PMMA 500 329 650 - PFM

[Mye02] 2002 Dig. Optr. Polymer 180 350 685 0.20 no samples available

[Shin02] 2002 KIST Korea PMMA 1000 120 650 0.26 g=2.4; 3.45 GHz 100m

[Liu02a] 2002 Huiyuan PMMA - - - - since 2001

[Luv03] 2003 Luvantix PMMA ? 160 650 0.33 3.5 GHz bandwidth

[Fuj04] 2004 Lumistar PMMA 500 3 Gbps 50m

[Rich04] 2004 Optimedia PMMA 900 200 650 0.40 commercially available

[Yoo04] 2004 Optimedia PMMA 675 200 650 0.40 commercially available

[Nuv05] 2005 Nuvitech PMMA 500 180 650 0.25 3 Gbps 50m

[Nuv05] 2005 Nuvitech PMMA 900 180 650 0.30 3 Gbps 50m

[Fuj06] 2004 Lumistar-X new low loss 120 100 850 ? 10 GHz 50m

MSI-POF

[Shi99] 1997 Mitsubishi (Eska-Miu)

PMMA 700 210 650 0.30 500 MHz 50m, 4-7 layers (?)

[Lev99] 1999 RPC Tver PMMA/ 4FFA 800 400 650 7 layers, 310 MHz 100 m

From the beginning of the 90s, it became possible to produce PMMA-GI-POF having an attenuation at 650 nm that is similar in quality to that of SI-POF. In doing so, it was possible to attain bandwidths up to 50 times larger which are ade-quate for transmitting several Gbit/s across distances up to 200 m. Likewise, multi-core and multi-step index POF achieve similar values for attenuation and allow data rates up to 1 Gbit/s across distances of 50 m, e.g. for applications in compliance with IEEE1394 (up to S800). The core diameter of all these fibers typically lies between 0.5 mm and 1 mm which means that existing reasonably priced connectors can be used.

Multi-step index profile fibers - the last lines in the table - have been described in [She99] and [Lev99]. In the group headed by Prof. Levin, different materials were used for the production of layers with different refractive indices (P(MMA/4FFA), P(MMA/4FMA) and PMMA-naphthalene). The best results were obtained with the mixture PMMA/4FFA which has an attenuation of approximately 400 dB/km (at 650 nm) and a bandwidth of 310 MHz·100 m. The total number of 7 steps in the fiber with a core diameter of approximately 800 μm were produced in a preform and subsequently drawn.

The ESKA-MIU has a core/cladding diameter of 700 μm/750 μm and also has several layers (probably between 4 and 7) which are produced by co-polymeri-zation. Originally, 4 to 7 layers were presumably being aimed at but in the end this


fiber with 3 layers was achieved as a product. It is said to be produced in a con-tinuous drawing process. The bandwidth in [Shi99] is stated to be larger than 500 MHz 50 m. In several publications this fiber is called a GI-POF ([Sak98], [Num99]). The difference between this design and “genuine” GI fibers is prima-rily the larger core diameter. In [Num99] the attenuation of the fiber is stated as being 210 dB/km with an AN = 0.30, i.e. values that are comparable with the DSI-POF. Materials and measurements of the index profile will be discussed in the Section Production and Materials.

Institutes and companies from South Korea have been very successful in producing PMMA GI-POF. In the past few years publications have come from:

Department of Materials Science and Engineering, Kwangju Institute of Science and Technology (KIST), Kwangju Center for Advanced Functional Polymer, Department of Chemical Enginee-ring, KAIST, Taejon, Korea E-Polymer Laboratory, SAIT, Taejon, Korea Optics Laboratory, Seoul, Korea Optimedia, Korea Nuvitech, Korea Luvantix, Korea

The production method for GI POF is described in [Shin03]. A MMA-BzMA mixture is poured into a rotating cylinder. The purpose of the rotation is simply to form even concentric layers. The polymerization takes place thermally and the concentration of BzMA is continuously increased to 15%. This emerging preform is then drawn into a fiber. Figure 2.65 shows the pulse broadening for a 66 m long fiber which corresponds to a BLP of 3.45 GHz · 100 m. The smallest measured attenuation of the fiber is given at 120.6 dB/km.

16000 200 600 1000 1200 1400400 800-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2Intensity [a.u.]

t [ps]

1 m

99.5 ps

66 m

129.5 ps

Fig. 2.65: Pulse broadening in PMMA-GI-POF ([Shin03])


Since 2004, a new GI-POF on a PMMA basis has been available on the market. The OM-Giga (see [Rich04] and [Yoo04]) has a core diameter of 900 μm or 675 μm respectively and a nearly parabolic profile. It is produced through poly-merization of several layers, although the steps are almost completely smoothed through thermal treatment. According to the data sheets available in the Internet the fibers have the following parameters (Table 2.10).

Table 2.10: Parameter of GI-POF OM-Giga

Property Unit B-075 B-100 Remarks

core diameter μm 750 (675) 1,000 (900) (GI-region) diameter variations % ±5 ±5tensile strength N > 35 > 65 at break bend radius mm 25 25 temperature range °C -30 .. +60 -30 .. +60 attenuation dB/km < 200 < 200 at 650 nm bandwidth GHz > 1.5 > 1.5 for 100 m

The fact that this fiber possesses thermal stability comparable to a standard POF, different from GI-POF with doping, must be rated as a particularly great step. Even after 5,000 hours of operation at 80°C no change in the bandwidth could be determined. The cross-section of a 1 mm OM-Giga is shown in Fig. 2.66 (microscope photograph shown in wrong colors). The approx. 10 index steps can still be seen quite well.

Fig. 2.66: Cross section of an OM-Giga (POF-AC) and a MSI (Tver, [Ald05])

In Fig. 2.67 the change in the refractive index profile of a doped PMMA GI-POF is shown after accelerated aging (122 hours at +109°C, from [Bly98a] and [Bly98b]). You can see quite well that the index profile is still parabolic at the beginning of the aging process. The share of the dopants is the greatest in the center of the fiber which is why the glass transition temperature has sunk the most.


The dopant diffuses outwardly. Consequently, the concentration increases out-wardly, Tg also drops there and the diffusion process continues until the profile has almost become rectangular. The attenuation of the fiber will hardly increase, but the bandwidth drops dramatically. Decisive for the application temperature of the fiber is the dopant concentration in the axis.

Fig. 2.67: Change of the refractive index profile of a GI-POF by ageing

Measurements on OM-Giga at the POF-AC will subsequently be introduced. The results of a long-term temperature test are shown in Fig. 2.68. The bandwidth was measured for over 5,000 hours on a 50 m long sample in order to be able to determine changes in the index profile.

0

500

1000

1500

2000

2500measured bandwidth [MHz]

0 400 800 1200 1600 2000 2400

time [h]

70°C 80°C

Fig. 2.68: Long-term behavior of OM-Giga


The frequency range of the network analyzer extended to 1.3 GHz. The values represented were determined through extrapolation and thus burdened with a rela-tively large error. A clear deviation from the parabolic index profile would in any event have caused a very strong decrease in the bandwidth.

The stable bandwidth proves that co-polymerization is obviously a suitable means to produce thermally stable and thus long-life PMMA GI-POF.

A comparison of the measured attenuation of ESKA-MIU and OM-Giga is shown in Fig. 2.69. The attenuation of the OM Giga is somewhat higher at 650 nm than that of the Mitsubishi fiber and also of the SI-POF. However, it clearly shows the greatest bandwidth.

100

1000

400 450 500 550 600 650 700

attenuation [dB/km]800

600

400

300

200

wavelength [nm]

OM-Giga

ESKA-MIU161 dB/km

217 dB/km

Fig. 2.69: Spectral attenuation of ESKA-MIU and OM-Giga

The Korean manufacturer Luvantix offers preforms for PMMA GI-POF ([Luv03] and [Kim03]). The index profiles from both refernces - measured values and approximation of each - are shown in Fig. 2.70.

The authors do not know what relations exist between Luvantix as preform manufacturer and Nuvitech and Optimedia as fiber producers as well the different research institutes. Overall, however, POF production in South Korea seems to enjoy greater attention and more progress in the field is foreseeable.

Other announcements concerning the production of GI-POF came from the USA (Digital Optronics and Nanoptics, [Wal02], [Mye02] and from China [Liu02a]). Since no data or even fibers are known from these producers they will not be considered in greater detail.

In Sections 2.5 and 2.6 data on bending behavior and bandwidth are summa-rized. The production methods are presented in Section 2.8.

2.4 Glass Fibers for Short-Range Data Transmission 93

1.492

1.494

1.496

1.498

1.500

1.502

1.504

1.506

1.508

1.510

1.512

0.0 0.2 0.4 0.6 0.8 1.0

refractive index

0.2 0.4 0.6 0.8 1.0

refractive index

theory

measured

1.460

1.465

1.470

1.475

1.480

1.485

1.490

1.495

1.500

1.505

1.510

theory

measured

AANN == 00..2211 AANN == 00..3322

rel. radius (r/rc) rel. radius (r/rc)

Fig. 2.70: PMMA GI-POF index profile (left: [Kim03], right: [Luv03])

2.4 Glass Fibers for Short-Range Data Transmission

2.4.1 200 μm Glass Fibers with Polymer Cladding

Thanks to their simple production and great robustness silica glass fibers with polymer cladding have been used for a long time. Figure 2.71 shows the principle structure. A core (typically with a diameter of 200 μm) of homogeneous SiO2 is surrounded by a high-strength, transparent polymer with smaller refractive indices (about 15 μm thick).

2.3 mm outer jacket

500 μm inner jacket

230 μmpolymer

200 μmSiO2-core

Fig. 2.71: Structure of a 200 μm PCS

Production is so easy because the core is drawn from a quartz glass cylinder. The polymer cladding is applied by extrusion after it has cooled off. First of all, all glass fibers are extremely sensitive to water and must be protected by a plasticcoating as thick as possible. Furthermore, pure glass fibers do not have a great

94 2.4 Glass Fibers for Short-Range Data Transmission

mechanical load capability. The polymer cladding gives the fibers the capacity to bear extreme loads. The jacketed fiber can thus hardly be shattered. Pure glass-glass fibers (glass core with an optical glass cladding) are always surrounded by similar protective layers, e.g. acrylates which, however, do not have any optical function.

Because of its refractive index and attenuation the polymer cladding determines to a great extent the optical parameters of the PCS. In short wavelength ranges the attenuation nearly corresponds to pure SiO2 fibers. Above approx. 1,000 nm the losses in the polymers are so high that the effective PCS attenuation also rises ra-pidly. Silica glass can endure temperatures up to 1,000°C, but not the polymer cladding. Consequently, the primary coating material determines the thermal and chemical characteristics. Most PCSs available in the market have been specified for an application temperature of +70°C. Some more recent types have been di-mensioned for use in automobile networks for temperatures up to +125°C. Infor-mation on such PCSs can be found for example in [Hub03] and [Schö03]. Fig 2.72 has been taken from the latter work. You can clearly recognize how strongly the attenuation spectra of different PCSs can depend on the cladding materials selected.

0.1

1

10

100

1000

10,000

200 400 600 800 1,000 1,200 1,400 1,600 1,800

attenuation [dB/km]

wavelength [nm]

theoretical limit

diff. PCS

Fig. 2.72: Attenuation of different 200 μm PCS according to [Schö03]

Just as with glass-glass fibers the absence of water plays an important role for PCS for keeping losses low especially in the long-wave ranges. So-called all-silica fibers in which the optical cladding consists of silica glass are used at high tem-peratures. These fibers are also employed for the transmission of very high light power (working with lasers) since it is very important that no light is absorbed at the core-cladding interface layer.


Table 2.11 lists some of the representative types taken from a number of diffe-rent PCS variants which differ in cladding material, core diameter and NA (data from [Hub03] and [OFS02]).

Table 2.11: Properties of different PCS

Parameter Unit All SilicaHigh OH

All SilicaLow OH

HCSHigh NA

HCSLow OH

PCS[Hub03]

producer OFS OFS OFS OFS Polymicro core/cladding

μm 200/240 365/400 550/600

940/1000

200/240 365/400 550/600

940/1000

200/230 400/430

125/140 200/230 300/330 400/430

200/230

NA - 0.22 0.22 0.43 0.37 0.37 (820 nm) dB/km 10

101010

88810

68

12688

6

bandwidth MHz km n. a. n. a. n. a. 20 201513

20

bend radius (long term)

mm 14 4794118

144794

118

1647

15162447

16

temperature °C -65..+135 -65..+135 -65..+125 -65..+125 -40..+125

PCS are generally used in lengths of up to a maximum of 200 m. The attenu-ation then only amounts to a few dB which can for the most part normally be dis-regarded. If a LED is used as a transmitter, considerably less light will be coupled into the fiber than into a 1 mm POF. On the other hand, the light can be coupled more effectively into the photodiode. Different manufacturers even offer trans-mission systems which can work with the same plug construction with POF as well as with 200 μm PCS, e.g. [HP01].

system with 1 mm POF

system with 200 μm PCS

-4 -6 -8 -10 -12 -14 -16 -18 -20 -22 -24 -26 -28 -30

Popt [dBm]

LED-power range (launched into the fiber)

receiver sensitivity range

allowed path loss (with margin)

Fig. 2.73: Link power budget for POF and PCS


Fig. 2.73 shows power budgets for both possibilities, each with the same trans-mitters and receivers (system for 125 Mbit/s).

The result for PCS is a permissible fiber attenuation of at least 11 dB - taking the system margin into consideration - thanks to the greater input power. In this way at least 20 m of POF can be bridged. The guaranteed loss for PCSs is only 7 dB. However, at least 100 m of fiber can be bridged, limited here due to the bandwidth.

10 10020 604030 8010

100

200

20

40

70

recommended application

area with POF

max. bit rate [Mbit/s]

max. at +25°C

10 100020 20010050 500

max. bit rate [Mbit/s]

max. at +25°C

fiber length [m]fiber length [m]

recommended application

area with PCS

Fig. 2.74: System parameters of the HP-system with POF and PCS (according to [HP01])

The bandwidth for PCS indicated in the data sheets has to be viewed with a certain degree of skepticism. Measurements conducted at the POF-AC show that all PCSs investigated with An = 0.37 at full launch have a BLP in the range of 5-7 MHz · km. This lies clearly below the specified data of 10-20 MHz · km. This is not a contradiction, however, since none of the manufacturers as a precaution provided any information about the measurement conditions. One reason may be that the PCS was developed for relatively low data rates (10 Mbit/s and less). The fiber bandwidth therefore did not play any role whatsoever while the POF was also designed from the very beginning for higher data rates. Diverse information and publications on bandwidth exist for the different polymer fibers as summa-rized in Chapter 2.5. The most recent draft for the standardization of PCS is viewed by the IEC as having a bandwidth of 5 MHz · km for fibers with a NA of 0.40 ± 0.04.

A specific problem with PCS in the past was that the temperature coefficients of glass and plastic did indeed deviate considerably from one another. In the case of some fibers this resulted in a refractive index difference - and NA, too - which dropped to zero at low temperatures. This effect is shown in Fig. 2.75 taken from [Dug88].

Far field distributions are represented in the picture after 2 m of fiber at diffe-rent temperatures. They were measured with laser stimulation at altered angles. In this case the optical cladding was a silicone plastic. Modern PCSs no longer show this effect.


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

-5 0 5 10 15 20 25 30 35

rel. power

+40°C -2°C-31°C-51°C-65°C-72°C-92°C-98°C

[°]

Fig. 2.75: Temperature dependence of PCS-NA, presented as far field

2.4.2 Semi-Graded Index Glass Fibers

Up until some time ago this class of fibers was only available as a product from the manufacturer Sumitomo ([Sum03]). Except for the gradients introduced this fiber corresponds to conventional PCS. The index variation is attained by adding germanium which is also usual for silica glass. Even with normal 50 μm GI fibers the germanium share represents a considerable cost factor. The semi-GI PCS, however, has a 16-fold cross-section. This type of fiber is still extremely expen-sive. It is still open how far the price can drop when manufacturing greater lengths. In the meantime, OFS has appeared as a second manufacturer ([Ziem06i]).

450 500 550 600 650 700 750 800 850 900 950 1000

10

100

8

80

6

60

4

40

20

30

spectral attenuation [dB/km]

Sumitomo

OFS

wavelength [nm]

Fig. 2.76: Spectral attenuation of the semi-GI-PCS


Figures 2.76 and 2.77 show the attenuation curve and the pulse response of the semi-GI-PCS based on measurements made at the POF-AC. The following table gives the parameters from the data sheet - the bending radius and the operating temperature are not specified. The bandwidth and maximum data rate measure-ments are dealt with in the corresponding sections.

0.0

0.2

0.4

0.6

0.8

1.0opt. power [a.U.]

0 10 20 30

t [ns]

FWHM: 6.8 ns 32 MHz km

full mode launch500 m PCS

Fig. 2.77: Pulse response of Semi-GI-PCS

Table 2.12: Parameters of Semi-GI-POF

Parameter Unit HG-series Sumitomo

Semi-GI V2 OFS

core μm 200 200 cladding μm 230 230 core structure n. a. n. a. VAD/MCVD GI NA - 0.40 0.36 GI-NA n. a. 0.275

(820 nm) dB/km 6 8 bandwidth MHz km 100 48 (overfilled)

2.4.3 Glass Fiber Bundles

2.4.3.1 Quartz Glass Fiber Bundles

Glass fiber bundles are employed in the most diverse areas. Above all, it makes sense to use them when a large light-guiding cross-section is to be combined with high cable flexibility. In optical measurement techniques bundles of quartz glass fibers are employed which permit a continuous high transmission rate in the range between 380 nm and 2000 nm. If you lay out the fibers differently at both ends of


the cable, then they can also serve as cross-section converters, e.g. monochro-mators. The end surfaces are usually prepared: the bundle is glued in the plug and then polished. Figure 2.78 shows an example.

Fig. 2.78: Example for a quartz glass fiber bundle

The transmission of such a bundle is shown in Fig. 2.79 (acc. to [Ori01]). The greatest part of the 100% missing share is determined by the only about 60% part of the core surfaces and the Fresnel losses. The numerical aperture of the bundle shown is 0.22, the length is about 1 m, and the single fiber diameter is 200 μm.

0%

10%

20%

30%

40%

50%

60%

70%

200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000

transmission

wavelength [nm]

Fig. 2.79: Transmission of a quartz glass fiber bundle [Ori01]


2.4.3.2 Glass Fiber Bundles

Pure quartz glass is many times more expensive than normal polymers but also as conventional mineral optical glasses. Attenuations of some 100 dB/km is absolu-tely acceptable for many applications, e.g. in lighting technology. Schott has been producing bundles of thin glass-glass fibers for quite some time. The index diffe-rence can be varied in diverse areas by choosing a particular glass composition. The spectral attenuation of a typical glass fiber bundle compared with a PMMA POF is shown in Fig. 2.80. The glass has higher losses in the blue areas whereby the cable length is limited when guiding white light. The POF has greater losses in the near infrared range.

100

1,000

200

500

50500 550 600 650 700 750 800 850500 550

[nm]

MC-GOF

POF

[dB/km]

Fig. 2.80: Spectral attenuation of glass fiber bundles and POF

An entirely new application for such glass fiber bundles has come about with the ever increasing use of optical networks in vehicles. The previous systems are specified with 1 mm POF. Two parameters especially limit the use: the tempera-ture range is limited to a maximum of +85º and the relatively large bending radius. Both limitations can be reduced considerably with glass fiber bundles, whereby the usual optical characteristics are for the most part retained so that the identical active components can be used. Table 2.13 from [Lub04b] compares the parame-ters of a glass fiber bundle (MC-GOF) with those of a POF for vehicle networks.

The construction of the plug is especially problematical. The usual method of cementing and polishing takes too much time for mass production and results in the core surface having too low a share with correspondingly high losses with the plug connections.

Megomat TS AG, working together with Schott, has developed a new kind of assembly procedure ([War03]). The actual fiber bundle has a diameter of 1.2 mm. The plug has a metal ferrule with a corresponding opening. During production the fiber bundle is heated to such an temperature that the glass can be compressed. The fibers are pressed closely together when crimped so that the diameter of the bundle is lowered to 1 mm.


Table 2.13: Comparison MC-GOF and POF

Parameter Unit MC-GOF MOST-POF

core diameter [μm] 53 1,000 45

cladding thickness [μm] 3 10

number of cores - approx. 400 1

ncore/ncladding - 1.585 / 1.49 1.49 / 1.40

numerical aperture - 0.50 0.50

attenuation at 650 nm [dB/km] 250 160

bandwidth (full launch) [MHz·20 m] 150 >50 (200 typ.)

bend radius [mm] 5 25

temperature [°C] -40 .. 125 -40 ..85

The core share of the plug end face then amounts to about 85%. After the crimping the bundle is broken off and polished. Figure 2.81 shows a photo of the plug end face.

Fig. 2.81: Photograph of a MC-GOF

The irregular arrangement of fibers in the bundle leads to different patterns when pressed together. Neighboring fibers mostly form regular hexagonal struc-tures which can, however, also have big gaps. Every once in a while linear struc-tures with pentagonal fibers are formed. The individual fibers are deformed at the edge of the bundle, something which happens quite irregularly.


Fig. 2.82: Details of MC-GOF connector end faces

Fig. 2.83: Details of MC-GOF connector end faces

Since the bundle consists of about 400 individual fibers this irregular defor-mation of individual fibers does not play any role overall. Since the deformations only arise over a few millimeters there is no significant additional attenuation.

In conclusion, Fig. 2.84 shows an x-ray photo of the bundle within the cable. The individual fibers have to move freely within the cable. When there is a tight bend the change in length is distributed on the inside and outside for a long stretch so that the fibers are only subject to a slight load. That is why the bundle can take bending radii of only a few mm.

Fig. 2.84: X-ray picture of a bundle

2.5 Bandwidth of Optical Fibers 103

2.5 Bandwidth of Optical Fibers

In order to be able to determine the bandwidth of an optical fiber, several different influencing factors need to be considered. Mode dispersion and chromatic disper-sion are the most important factors involved in multimode fibers. Particularly in the case of POF, mode dispersion depends on various parameters such as wave-length, the light launching conditions, refractive index profile, fiber-laying condi-tions as well as the homogeneity of the fiber's characteristics. In the following sections we intend to show how the conclusions drawn from basic physical pro-cesses can be used to explain values that are measured in reality.

2.5.1 Definition of Bandwidth

It is possible to define the term bandwidth in quite different ways. Essentially it describes the frequency range of a system within which the transmission of signals can be achieved with reasonable attenuation. In POF systems, the limiting factor is usually the bandwidth of the fiber itself, which is created by modal dispersion. As we will demonstrate later on, one can describe the SI-POF as being very close to a Gaussian low pass filter. In this book we will use the following definition of band-width:

f3dB : Frequency at which the amplitude of a sinus modulated monochromatic signal has been reduced to ½ of the optical level (see Fig. 2.28).

Figure 2.85 schematically illustrates this definition.

frequency f

rel. opt. amplitude at fiber output

0.0

0.5

1.0

bandwidth f3dB

frequency response

Fig. 2.85: Definition of POF bandwidth

Nonetheless, knowledge of the bandwidth alone is not adequate for estimating what the actual capacity of the complete link will be. In order to determine this parameter, it is further necessary to know the actual transmission procedure as well as the complete transmission function. For example, it is possible to transmit signals of significantly broader bandwidth if an electrical compensation of the frequency response takes place, as illustrated schematically in Fig. 2.86.

104 2.5 Bandwidth of Optical Fibers

P(f)POF

0.0

0.5

1.0

0 1 2

P(f)compensating filter

0 1 2

P(f)resulting

0 1 2

f [a.U.] ff

Fig. 2.86: Compensation of the POF low pass characteristic

A high pass filter is used for compensation. In the case of low frequencies, the signal is attenuated - in the case of higher frequencies the signal is passed through without attenuation. The resulting function has a significantly higher bandwidth; however, due to the overall existing level of attenuation, a higher level of signal is necessary.

In addition, the type of signal involved (digital or analogue) is also of signifi-cance and finally the required system reserves must be considered. The following general relationship can be used as a rule of thumb for digital systems:

maximum bit rate [Mbit/s] = 2 bandwidth [MHz].

We intend here to look at bandwidth as a function of fiber characteristics. For this reason, the effect of chromatic dispersion will be initially neglected because it is directly proportionally dependent on the spectral width of the source.

In this section we will show experimental investigations on the bandwidth of SI-POF fibers. After explaining the measurement procedures, we will show to what extent bandwidth is particularly dependent on the launching conditions.

2.5.2 Experimental Determination of Bandwidth

As a frequency response multimode fibers show an almost Gaussian-like behavior:

20

2 f/f0 eP)f(P

As can be easily demonstrated, the amplitude of a Gaussian low pass filter (P(f) = P0 · exp (f²/f0²)) for f = 1.17741 · f0 has dropped to half the value that applies for f = 0. When using a spectrum analyzer to measure the frequency response of a fiber link, it is necessary to determine the electrical 6 dB width because the photodiode will convert the optical power proportionally into a current. Therefore the following applies:


2optel PP

Figure 2.87 shows an example for such a bandwidth measurement for a 30 m standard NA-POF.

-18

-15

-12

-9

-6

-3

1 10 100 1,000

electr. power [dBm] measured

Gaußian approx.

15 m SI 650 nm

239 MHz NA: 0.34

f [MHz]

Fig. 2.87: Bandwidth measurement at a SI-POF

Due to the limited dynamics of the measurement system, the frequency res-ponse can be measured only up to a certain distance. In this case a measurement of up to 200 MHz was easily possible. The 3 dB bandwidth is found simply by deter-mining the point at which the electrically measured transmission function has dropped by 6 dB, here approximately 150 MHz.

Apart from the values actually measured, an approximation with a Gaussian low pass function has been entered into the figure. By determining the frequency f0 it is then possible to determine the bandwidth even when the measurement is not possible because of the limited dynamics or bandwidth of the measuring system.

Figure 2.88 shows the measured transmission functions for a SI-POF and a DSI-POF of 50 m length each. The optical 3 dB bandwidth for an SI-POF is approximately 67 MHz, corresponding to a bandwidth-length product of 33 MHz · 100 m, with the NA of the fiber being 0.52. It follows that the measured value is substantially greater than had been theoretically expected (approximately 14 MHz · 100 m, see Fig. 2.31). For DSI-POF (AN = 0.30) the measured value is 130 MHz, corresponding to 65 MHz · 100 m, with the theoretical value being 42 MHz · 100 m.

The measurement was carried out with a 520 nm LED. The LED had a wide emission angle so that approximate equilibrium mode distribution can be assumed.


-12

-9

-6

-3

0

1 10 100 1,000

rel. power [dB]

frequency [MHz]

50 m St.-NA-POF

50 m DSI-POF

Fig. 2.88: Bandwidth measurement for SI-POF and DSI-POF

When measuring bandwidth, a two to four ranging factor of deviation from the theoretical value of an ideal SI fiber can be generally expected, even when wor-king in an EMD condition. The reason for this is the combination of mode depen-dent attenuation and mode coupling described in Chapter 1. As a result of the con-tinuous energy exchange that takes place between the faster and slower modes, the delay does not rise in proportion to the length. The increased attenuation of those beams having a particularly large propagation angle - many reflections at the cladding - has the additional effect of reducing the pulse width.

Figure 2.89 shows the bandwidth measurement of a standard NA-POF for 3 different wavelengths for samples between 20 m and 100 m in length.

fiber length [m]

bandwidth [MHz]

10 1005020

650 nm 590 nm 525 nm

ACU 1000, Low-NA

10

100

1000

30

300

fiber length [m]

bandwidth [MHz]

650 nm590 nm525 nm

10 1005020

GH 4000, St.-NA

Fig. 2.89: Bandwidth of a SI-POF and a DSI-POF at different wavelengths


Once again, the measurements in Fig. 2.89 were carried out with an LED having an emission characteristic near to EMD (see [Gor98] and [Rit98]). The figure reveals 2 significant items of information:

The bandwidth of the POF does not decrease in proportion to the length-1; its decrease is less than proportional.The bandwidth of the POF is nearly identical for the 3 attenuation windows.

2.5.3 Experimental Bandwidth Measurements

The section on experimental bandwidth measurements first summarizes the results from earlier technical literature. As we shall see, determining the bandwidth is among the most difficult metrological challenges with thick-core fibers. A series of systematic measurements on the most diverse fibers is introduced as a supple-ment to the first edition (Alexander Bachmann is responsible for the bandwidth measurements at the POF-AC, cf. for example [Bun02a] and [Ziem04a]). Since there are still no standards for the definition and measurement of bandwidth, the following description will therefore not represent the definitive assessment.

2.5.3.1 Bandwidth of SI-POF

After presenting some examples of our own measurements in the previous section we will now compare these with measurements carried out by other authors. Some of the first efforts made to systematically investigate the bandwidth of SI-POF were undertaken by [Tak91], [Tak93] and [Rit93]. As shown in Fig. 2.90, the bandwidth was measured for SI-POF having lengths between 20 m and 100 m. The fiber used was an EH4001 by Mitsubishi with an NA of 0.47. The bandwidth was measured through the pulse broadening of a 660 nm laser pulse (150 ps). Different launching devices were used to vary the NA between 0.10 and 0.65. The detector used consisted of a wide area photo multiplier.

fiber length [m]

bandwidth [MHz]

AN Launch = 0.10AN Launch = 0.65

launch conditions

10 10020 50

theory

10

100

1,000

20

50

200

500

2,000

Fig. 2.90: Bandwidth measurement according to [Tak91]


The results allow three significant conclusions:

The bandwidth of a SI-POF is always significantly higher than the theoretical values for a SI-POF under theoretical UMD conditions, even with full laun-ching in the acceptance area.The measured bandwidth is strongly dependent on the launching conditions.Although the bandwidth difference for measurements using different laun-ching conditions decreases with the increasing length of fiber, it is still clearly in evidence even after 100 m.

Figure 2.91 shows measurements of bandwidths for different detector NA, also taken from [Tak91]. In principle, detection with a small NA means the same limi-tation in mode number as launching light at a small angle so that it is not surpri-sing to find that the value curves are similar.

fiber length [m]

AN Det = 0.22AN Det > 0.65

receiver detection angle range

10 10020 50

theory

10

100

1,000

30

300

bandwidth [MHz]

Fig. 2.91: Bandwidth measurement according to [Tak91] with different receiver NA

Another measurement of bandwidth on standard NA-POF (1 mm) is presented in [Tak93] (Fig. 2.92). Again, the measurement was carried out using the pulse method at 650 nm. Apart from measuring the bandwidth, the half far field width following the corresponding sample length was also determined.

The bandwidth was calculated from the far field width as follows:

const.)(Ct

CzB:thereforeand

cn2

At

mod

2FFN,

mod

whereby tmod is the modal pulse propagation and B · z is the product of bandwidth and length. Parameter C is a free selectable constant which depends on the coup-ling conditions. The speed of light is c. In the formula AN , FF is not the fiber para-meter indicated, but the value measured depending on length.

For a sample length of 10 m, the difference between the measured bandwidth for launching the light with AN = 0.10 and AN = 0.65 is more than one order of magnitude. For lengths up to 100 m this factor decreases to 2.


fiber length [m]

bandwidth [MHz]

AN Launch = 0.10AN Launch = 0.65

launchingconditions

10 10020 50

theory based on far field width

100

1,000

10,000

30

300

3,000

Fig. 2.92: Bandwidth according to [Tak93]

When launching light with a small NA, the bandwidth drops disproportionately, from approximately 80 MHz km to approximately 16 MHz km. This suggests an increasing filling out of modal field. By comparison, when launching with a large NA, the bandwidth is reduced somewhat more slowly than the length, from approx. 4 MHz km to approx. 5 MHz km. This is due to the effect of mode coupling and mode related attenuation.

The bandwidth values determined by means of the far field width correlate very well with the results of the bandwidth measurements made by pulse propagation. This suggests that mode dependent attenuation and mode conversion are the deter-mining processes because they affect the bandwidth by changing the mode distri-bution. In contrast, if mode coupling were more pronounced, the bandwidth would also change without affecting the far field. However, any estimated quantification based on these measured results alone would be questionable. In [Rit93] measured results for the bandwidth of standard NA-POF at launching conditions of AN = 0.10 and AN = 0.65 (Fig. 2.93) are also shown.

launching conditions

AN Launch = 0.10

AN Launch = 0.65

fiber length [m]20

100

1,000

2,000

50

200

500

5,000

10 100 1,00030 300

bandwidth [MHz]

Fig. 2.93: Measured bandwidth of a SI-POF according to [Rit93]


Here too, the measured bandwidth for short lengths (20 m) differs by more than an order of magnitude. For large lengths the difference is reduced corresponding-ly. The authors calculate the bandwidth based on their own theory that follows the concept of the diffusion model. Instead of investigating separate modes, this model investigates modal groups that differ in their 2 angles of propagation (radial and azimuthally).

The coupling between the modes is described by a diffusion constant that only takes into account the energy transfer in neighboring mode groups. The model also takes into account mode dependent attenuation.

In this work the remaining deviation between theory and measured values is explained by means of the mechanism of mode coupling. In variance to the model, this is a factor that is not independent of the angle. Simulations provide good results if elongated scattering centers of 37 μm length and 2.5 μm diameter are assumed in the fiber with random distribution and orientation along the axis of the fiber (caused by the drawing process), as shown schematically in Fig. 2.94.

scattering centers

Fig. 2.94: Model for scattering centers in POF

An indication of a non-uniform inner structure of the PMMA fiber is the photo (from Fei00 ) of the surface of a cut POF taken by a scanning electron micros-cope and shown in Fig. 2.95. The fibril-like structures in the sub- m range can clearly be seen.

Fig. 2.95: Microscopic structure of a PMMA POF cut ([Fei00])


Figure 2.96 shows further experimental results for the bandwidth of polymer optical fibers [Kar92]. In each case collimated light or light with an angle adapted to the fiber's NA (UMD) was launched into the POF. As was the case in the results previously shown, very large differences result for short lengths of fibers. The parameter shown in the figure here is the product of bandwidth and length.

10 1005 5020 500200

10

2

5

20

50

sample length [m]

bandwidth length product [MHz km]

1.0 mm0.5 mm

0.5 mm

1.0 mm

collimated

UMD

launching conditions

Fig. 2.96: Measured POF bandwidth of a SI-POF according to [Kar92]

Apart from the effect of the launching NA, [Kar92] also investigates whether the bandwidth depends on the size of the launched beam. In fact, for UMD laun-ching, a larger bandwidth was found as well as a smaller light spot, compared with complete illumination of the fiber cross-section; however, the differences are not as pronounced as when the launch angle is changed. For collimated light the relationship is reversed.

Because all processes described up to this point are only dependent on the angle, it seems surprising to find that the size of the launching spot has an effect on the measured bandwidth of SI fibers. However, when considering the fact that mode conversion can cause deviations in location and deviations in angle after just a short length of the specimen (see schematic in Fig. 2.97), the result becomes understandable [Kar92].

deviation in location

bending

deviation in angle

Fig. 2.97: Conversion of spatial and angular distances


In [Poi00] the results of [Kar92] are compared with current measurements on 2 standard NA-POF by Toray and Mitsubishi (Fig. 2.98). These measurements qualitatively confirm the previous results. For very short lengths of samples the differences between small and large launching angles are even greater.

30,000

10,000

3,000

1,000

300

100

301 10 1003 30 300

bandwidth [MHz]

length [m]

theory

NALaunch: Toray 0.09 Mitsubishi 0.09 [Kar92] collimated Toray 0.64 Mitsubishi 0.64 [Kar92] UMD

Fig. 2.98: Measured bandwidth of different SI-POF according to [Poi00]

2.5.3.2 Bandwidth Measurements on SI-POF

This section as well as the following four sections deals with bandwidth measure-ments conducted at the POF-AC Nürnberg. All measurements were carried out under uniform measurement conditions.

Semiconductor diodes with a wavelength of 650 nm or 850 nm respectivelyserved as transmitters. Both lasers can be modulated analogously up to 2 GHz. A singlemode glass fiber is mounted firmly to the laser diodes. Using a combination of different microscope lenses and optical apertures the coupling angle in the area AN Launch = 0.01 to 0.64 can be varied. The coupling spot is directly visible through a beam splitter. With the aid of adjustment screws the size as well as the position of the light spot can be changed. Figure 2.99 shows the complete setup of the measurement device.

A commercial product on the basis of a 400 μm Si-PIN photodiode with an integrated preamplifier and about 1.5 GHz bandwidth was used as a receiver. In order to attain mode independence the receiver was connected to a 1 mm mixed glass fiber bundle with a large NA.


Fig. 2.99: Experimental setup of bandwidth measuring by POF-AC

The lengths and NA-dependent bandwidths were systematically measured for a series of different step index profile fibers. An overview can be found in [Bun02a]. The following Figs. 2.100 to 2.102 show the results for three types of fiber:

1 mm standard PMMA POF with AN = 0.46 1 mm standard POF made of cross-linked PMMA with AN = 0.54 1 mm polycarbonate POF with AN = 0.75

10 1005 5020

100

1,000

50

200

500

2,000

5,000

20

B3dB [MHz]

length [m]

0.64

0.48

0.33

0.19

0.09

0.05

NAlaunch:

Fig. 2.100: Bandwidth measurement of a 1 mm SI-PMMA-POF


For a 1 mm PMMA POF (Toray PFU CD1000, see also [Ziem04a]) 3 dB band-widths for lengths between 5 m and 100 m were measured. The coupling angle was changed for NA values between 0.05 and 0.65 with the unit described above.

For short fiber lengths the bandwidths measured differ by almost a magnitude which demonstrates once again the importance of correct measurement conditions for correctly indicating the bandwidth values. After a 100 m test length there still is a factor of two between the values measured. The curves for under filled launch (small NA) fall more steeply than with length caused by a predominance of mode mixing. For overfilled launch (large NA) the curves run flatter. Here the mode-dependent attenuation dominates. The next figure shows the results with a 1 mm POF made of modified PMMA (Toray PHKS CD1001). The fiber is specified with a NA of 0.54.

5 10 10020 50 length [m]

B3 dB [MHz]

30

100

1,000

3,000

300

AN = 0.05

AN = 0.09

AN = 0.19

AN = 0.33

AN = 0.48

AN = 0.64

NAlaunch:

Fig. 2.101: Bandwidth measurement of a 1 mm SI-mod. PMMA-POF

Since the losses of this fiber lie at about 300 dB/km at 650 nm, test lengths of only up to 50 m could be measured. Incidentally, the measurement results are similar to a large degree to the results of the PMMA POF.

1 10 205230

100

1.000

3.000

300

AN = 0,05

AN = 0,09

AN = 0,19

AN = 0,33

AN = 0,48

AN = 0,64

NAlaunch:B3 dB [MHz]

length [m]

Fig. 2.102: Bandwidth measurement of a 1 mm SI-PC-POF


The third fiber tested is the polycarbonate POF FH4001 from Mitsubishi. The NA of the fiber lies at 0.75, the attenuation amounts to 650 nm at about 800 dB/km, whereby the maximum measurement length remains limited to 20 m.

Surprisingly, the bandwidth differences between the three types of fiber are only very slight although there were clear differences in the NA. One explanation for this could be the greater effects for mode mixing and above all for the mode-dependent attenuation which occurred in the fibers made of modified PMMA and polycarbonate. Figures 2.103 and 2.104 illustrate the far fields of the three fibers in comparison (cf. [Bun02a]).

length [m]

B3 dB [MHz]

PC

PHKS

PMMA

100

1.000

30

300

3.000

10 1002 205 50

NALaunch = 0.33

Fig. 2.103: Comparison of the bandwidths of different SI-POF

0

200

400

600

800

power [a.U.]1000

-40 -30 -20 -10 0 10 20 30 40

PMMA

PC

mod. PMMA

[°]

Fig. 2.104: Comparison of the farfields of different SI-POF

The fibers of PMMA and PC - each after 10 m - have half-value widths of about 27°. The fibers of modified PMMA have only 17°. Here the share of mode-dependent attenuation predominates over the nominally larger NA.


Fig. 2.105: Comparison of the far fields of different SI-POF (3-d representations)

As part of the European Project POF-ALL (www.ist-pof-all.org) other compre-hensive measurements of both length and launch-dependent bandwidths of diffe-rent fibers were carried out. The following Figures 2.106 and 2.107 show the measurement results for a 1 mm standard POF (Luceat, high quality fiber) and for a 500 μm standard POF.

B3 dB [MHz]

fiber length [m]

100

1,000

10,000

200

2,000

500

5,000

501 5 20 1002 10 50

0.05

0.10

0.19

0.34

0.47

0.65

NALaunch

Fig. 2.106: Bandwidth measurements of a 1 mm SI-POF (Luceat, HQ)

Both fibers essentially show comparable results. Since the fibers also have very similar attenuation values they can be used in almost all the same applications. The advantages of the thinner fibers are primarily the smaller space needed, an im-portant point with multiple cables, and the smaller bending radius. The argument that the fibers with a smaller core diameter would enable higher bit rates or better receiver sensitivity because of the smaller photodiodes has for the most part since been dropped because of technical developments.


1 5 20 1002 10 50

B3 dB [MHz]

fiber length [m]

100

1,000

10,000

200

2,000

500

5,000

50

0.05

0.10

0.19

0.34

0.47

0.65

NALaunch

Fig. 2.107: Bandwidth measurements of a 0.5 mm SI-POF

2.5.3.3 Bandwidth Measurements on MC- and MSI-POF

Multicore and multistep index profile POFs allow significantly greater bandwidths than conventional step index profile fibers. In the case of MC-POFs the diffe-rences in propagation time between the different individual cores, in addition to mode dispersion, have to be added. The pure length differences, however, should hardly play a role. The pure path differences alone between the modes amount to about 6% at a maximum propagation angle of 20° in the fiber. Since the fibers lie well-ordered in the MC-POF the geometric differences in length lie at the most in the thousandth range.

Of greater significance is the fact that the fibers in the MC-POF are deformed in different ways. For example, differences among the fibers in the middle and at the edge of the bundle can already be seen in the attenuation. These differences are also formed for the mode selective processes resulting in different average propagation speeds in the individual cores.

launching with magnified light spot: - medium fibers obtain small angles only - outer fibers obtain large angles

launching with mode field converter: - all fibers obtain around the same optical power and rays of all angles

Fig. 2.108: Optimal launching into multicore fibers


In order to register these effects when making a measurement, a so-called mode field converter (MFC) is used. The coupling unit shines into a short piece of SI fiber with a large NA. The far field distribution remains intact as the light is distri-buted over the fiber cross-section. This ensures that the individual fibers receive approximately identical light intensity and comparable angle distributions. The difference between this arrangement as opposed to a simple widening of the light spot is depicted schematically in Fig. 2.108.

As indicated above, there are in the meantime different MC-POFs. Here we wish to present the results of the bandwidth measurements of two 1 mm MC-POFs with 37 cores (Fig. 2.109) and 217 cores (Fig. 2.110, see also [Ziem02a]).

100

1,000

200

500

2,000

20 10030 40 60 80

B3 dB [MHz] PMC 100037 cores

length [m]

AN = 0.09

AN = 0.19

AN = 0.33

AN = 0.48

AN = 0.64

launch NA:

Fig. 2.109: Bandwidth measurements of a 37-cores MC-POF (measured on single samples)

B3 dB [MHz]

length [m]

100

1,000

200

500

2,000

20 10030 40 60 80

MCS 1000217 cores AN = 0.09

AN = 0.19

AN = 0.33

AN = 0.48

AN = 0.64

launch NA:

Fig. 2.110: Bandwidth measurements of a 217-core MC-POF (measured on single samples)


Both fiber types show considerably greater bandwidth values compared to standard SI-POF. The 37 core fiber above all shows hardly any drop in the band-width over great lengths, especially since the dependence on the coupling condi-tions is very small. The reason is the very great mode dependence of the attenu-ation. This fiber possesses a double step index structure. Using a laser coupling it has been possible to transmit 1 Gbit/s over this fiber for 100 m.

Systematic investigations of the bandwidth have been carried out at the POF-AC on a 37 core POF sample with a relatively small diameter (400 μm). The results for the two different NA are shown in Fig. 2.111. The bandwidth of this fiber is almost independent of the launch conditions. The reason for this is the strong mode-dependent attenuation which, as described above, occurs intensively with very thin fibers and leads to a equilibrium mode distribution after very short lengths.

1,000

10,000

200

2,000

500

5,000

0.1 1 10 1000.3 3 30

B3 dB [MHz]

fiber length [m]

0.05

0.65

NALaunch

Fig. 2.111: Bandwidth measurements of a MC-POF (measurements on a single fiber sample, cut-back method)

Multi step index fibers have already been introduced by different manufac-turers. However, they are not yet ready to go into mass production. The youngest product so far is the ESKA MIU from Mitsubishi-Rayon, a fiber with three diffe-rent layers. Using a sample length of 100 m of this fiber, a bandwidth of almost 300 MHz was ascertained. Figure 2.112 shows the frequency response.


-24

-21

-18

-15

-12

-9

-6

-3

0

3

NA 0.10

NA 0.34

NA 0.64

rel. level [dB]

10 100 100020 20050 500

frequency [MHz]

Fig. 2.112: Frequency responses of a 100 m MSI-POF

2.5.3.4 Bandwidth Measurements on GI-POF

The bandwidth measurements of graded index profile fibers are associated with a series of particular difficulties. First of all, the attenuation is relatively high with polymer fibers so that the sample lengths can not be very great because of the limited dynamics of the measuring system. For PMMA POF the maximum measuring lengths lie between 50 m to 100 m. For PF GI-POF lengths of some 100 m can be used. On the other hand, POFs have a relatively large core diameter. The detector must be relatively large in order to record most all modes and thus obtain a meaningful bandwidth measurement. Then again the size of the diode limits the bandwidth of the detector. The only commercially available measuring system that can be used for this special task is the optical oscilloscope from Hamamatsu (described in the Chapter on measurement techniques). However, only measurements in the time domain are possible.

The bandwidths of PMMA GI-POF and PF-GI-POF have been measured at the POF-AC. The transmission functions for the PMMA GI-PF OM-Giga from Opti-media (see also [Yoo04], [Rich04]) and a PF-GI-POF (Nexans, see [Gou04]) are illustrated in Fig. 2.113 and 2.114.

The optical bandwidth of the fiber at 1.504 MHz was ascertained by adapting the measuring curve to a Gaussian function. This value, however, can clearly fluctuate with slightly altered launch conditions. The measuring conditions lie close to the “worst case scenario”. When there is under-launching, e.g. with a VCSEL transmitter, even greater values can be attained.


-6

-5

-4

-3

-2

-1

0

10 20 50 100 200 500 1,000

rel. electr. level [dB]

frequency [MHz]

50 m OM-Giga

LD = 650 nm ANLaunch = 0.34

f3 dB opt. = 1,504 MHz

Fig. 2.113: Frequency response of 50 m OM-Giga (AN = 0.34; 650 nm)

In order also to be able to measure bandwidths of several GHz with thick core fibers, an optical oscilloscope is a practicable device, whereby the widening of a short laser pulse (about 120 ps) is measured. In [Lwin06] the results for the OM-Giga are shown compared with the microstructured POF (with effective graded index profile).

100

150

200

250

300

350

15 25 35 45 55 65

length [m]

pulse broadening [ps]

Optimedia 1,000 μm

MPOF: 500 μm

Abb 2.114: Pulse broadening measurement of a MPOF and GI-POF

A pulse width of about 340 ps corresponds approximately to an optical band-width of 1.4 GHz. The value matches pretty much the measurements in the fre-quency range when taking the various problems of measurement techniques with such large frequencies into account.


The next illustration shows the frequency response for a PF-GI-POF at the wavelengths 650 nm and 850 nm together with the fitted Gaussian functions.

-6

-5

-4

-3

-2

-1

0

+1

0 200 400 600 800 1000 1200


frequency [MHz]

300 m PF-GI-POF

LD = 650 nm/850 nm ANLaunch = 0.10

measurement 850 nm

Gaussian fit 850 nm

measurement 650 nm

Gaussian fit 650 nm

Fig. 2.115: Frequency response of a PF-GI-POF

The 3 dB bandwidths are around 1,600 MHz for both fibers. The bandwidth-length product is at about 500 MHz · km, somewhat in the range of conventional multimode graded index glass fibers (cf. further results in [Bach01]).

2.5.3.5 Bandwidth Measurements on MC-GOF and PCS

The bandwidth measurement of multimode glass fibers proceeds according to the same principles. First, the glass fiber bundles from the Schott manufacturing com-pany are measured. The bundle consists of about 400 individual fibers each having a diameter of 53 μm. The fibers have been hot pressed into the plug so that the overall diameter is about 1 mm.

In Fig. 2.116 the frequency response for various coupling conditions are shown. The NA of the fiber lies at 0.50. The fiber bandwidth does not recognizably change when coupling in at great angles.

With full mode launch the measured bandwidth amounts to about 150 MHz · 20 m which is almost exactly the same value as for SI-POF with a comparable NA. Consequently, this fiber type can be used alternatively to the 1 mm St.-NA POF when either high temperatures or very tight bending radii are necessary ([Lub04b]).


-16

-14

-12

-10

-8

-6

-4

-2

0

2

AN = 0.10

AN = 0.34AN = 0.46

AN = 0.60AN = 0.64

launch NA:

10 100 100020 20050 500


f [MHz]

20 m fiber at 650 nm

Fig. 2.116: Bandwidth measurement of 20 m MC-GOF

In another series of measurements the length dependence of the bandwidth of MC-GOF was investigated. Fig. 2.117 shows the results for 3 different launch conditions. The bandwidth decreases almost linearly with the length so that one can assume that the influence of mode mixing is relatively small.

length [m]

B3 dB [MHz]

100

1,000

300

3,000

2 10 505 20

AN = 0.64

AN = 0.34

AN = 0.10

launch NA:

1 mm MC-GOF 375 cores

NAfiber: 0.50 = 650nm

Fig. 2.117: Length- and NA-dependent bandwidth of MC-GOF

Finally, the bandwidth for different lengths was determined using a 650 nm laser. In order to be able to make measurements relatively independently of mode, a 1 m long SI-POF was used as an adapter fiber at both the transmitter and the receiver. Figure 2.118 shows the results.


excitation by laserNAlaunch 0.30

= 650 nm

length [m]100

1000

500

200

10 20 30 40 50 60

B3 dB, opt. [MHz]

Fig. 2.118: Bandwidth of a MC-GOF excited by a laser source

This type of fiber is suitable for the transmission of data rates in the Gbit/s range over lengths of 10 m to 20 m.

Another glass fiber version which has gained increasing attention is the PCS, i.e. silica glass fibers with a polymer cladding. The typical NA lies around 0.37. However, there are versions available with a NA up to 0.48. Accordingly, the bandwidth of PCS should lie in the range of DSI-POF. At the POF-AC predo-minantly fibers with a core diameter of 200 μm - the most commonly used value - were measured. In Fig. 2.119 the length and launch-dependent results for a typical PCS are represented. The fiber, 200/230 μm with a 500 μm primary coating, was laid out for this measurement as a loose bundle with a diameter of about 30 cm (see also [Ziem04a]).

200 μm PCS loose bundle

length [m]

100

1000

20

200

2000

50

500

10 10020 20050 500

B3 dB [MHz]

0.02

0.09

0.17

0.26

0.34

0.46

launch NA:

Fig. 2.119: Bandwidth of a 200 μm PCS

This fiber was specified with a bandwidth of 100 MHz · 100 m. This value can be achieved for an under filled launch. For a full launch, however, you can only attain about 60 MHz · 100 m. The differences between the different launch condi-


tions hardly decreases with fiber lengths up to 250 m. Mode mixing hardly occurswith this measurement. The measurement was repeated for the same type of fiber, whereby the fiber was wound around a spool. The results are shown in Fig. 2.120.

100

1000

20

200

2000

50

500

10 10020 20050 500

200 μm PCS fiber on a spool

length [m]

bandwidth [MHz]

0.02

0.09

0.17

0.26

0.34

0.46

launch NA:

Fig. 2.120: Bandwidth of a 200 μm PCS

The results pretty much agree for short fiber lengths. For longer lengths, how-ever, the differences roughly disappear between the different launch conditions for the rolled up PCS. This can only be explained by a recognizable increase in the mode mixing. The bandwidths dependent on the coupling NA are compared for 250 m long samples in Fig. 2.121.

0

10

20

30

40

50

60

70

80

90

0.00 0.10 0.20 0.30 0.40 0.50

loose bundle

Bopt, 3 dB [MHz]

fiber on a spool

250 m PCS

launch NA

Fig. 2.121: Bandwidth comparison of 250 m PCS


Examinations of the different types of 200 μm SI-PCS confirm the measure-ments mentioned above. The specified bandwidth-length products of 10 to 20 MHz · km could be attained for all fibers examined only with under filled launch. Unfortunately, none of the currently active manufacturers provided any data on measurement conditions for the bandwidths indicated. Even the corres-ponding standards are completely missing. Should PCS seriously advance into areas of application in which the available bandwidth is to be completely used, then a lot of work still has to be done in this area. A comparison between the launch-dependent bandwidths is represented in Fig. 2.122.

B L [MHz km]

0.0 0.1 0.2 0.3 0.4 0.5

10

20

40

30

4

8

6

launch NA

15

Fig. 2.122: Bandwidth dependence on launch conditions for 5 different PCS types

A number of publications, especially in regard to applications in the Gigabit-Ethernet and 10Gigabit-Ethernet ranges, exist for a bandwidth of 50/125 μm GI glass fibers. Articles providing an overview include [Oeh02] and [Bun03a].

With GI glass fibers, too, keeping the exact parabolic index profile as well as the mode-selective coupling play the most important role in achieving large band-widths.

Originally, two different types of fibers were specified:

customary in the USA: 62.5/125 μm fiber (AN = 0.275 ± 0.015) customary worldwide: 50/125 μm fiber (AN = 0.200 ± 0.015)

The typical bandwidth-length product is 160 MHz · km to 200 MHz · km (62.5 μm) when using an 850 nm LED as emitter. 500 MHz · km is attained with 1,300 nm laser emitters. The limiting factor is the refractive index dip in the middle of the fiber which is caused by the production technology.


For fast Ethernet (125 Mbit/s) the bandwidths entirely suffice to bridge distan-ces of up to 1 km. The first range limitations (maximum of 275 m at 850 nm emit-ters and 62.5 μm fiber) arise with Gigabit-Ethernet so that a new class of fibers (OM2) has been defined which generally guarantees a transmission range of 550 m.

In the worst case a data rate of 10 Gbit/s could be transmitted on OM1 fibers over about only 30 m. OM2 fibers are also limited to about 80 m. In order to be able to transmit high data rates, three different procedures have been suggested:

Splitting the data rate into 4 × 2.5 Gbit/s which are then transmitted by WDM on a fiber. Emitter with so-called Restricted Mode Launch (RML) or Effective Laser Launch (EL) respectively, whereby the power is coupled if possible within the annulus with a diameter of between 4.5 μm and 19 μm. Moreover, the NA of the emitter may not be too large. Use of the new OM3 fiber class which has been optimized for the employ-ment of 850 nm VCSEL.

An overview of the specified characteristics of the different GI-GOFs is presented in Table 2.14. Specific products can on occasion clearly surpass these parameters.

Table 2.14: Properties of MM-GI glass fibers

Class Unit OM1 OM2 OM3 OM3

550m

typical applications Fast

EthernetGigabit

Ethernet10Gbit

Ethernet10Gbit

Ethernet

core- [μm] 50/62.5 50/62.5 50 50

at 850 nm [dB/km] 3.5 3.5 3.5 3.0

at 1.300 nm [dB/km] 1.5 1.5 1.5 1.0

BW 850 nm (OFL) [MHz km] 200 500 1,500 3,500

BW 1.300 nm (OFL) [MHz km] 500 500 500 500

BW 850 nm (LD) [MHz km] n.d. n.d. 2,000 4,700

(OFL: Overfilled Launch)

Other less customary fiber types are, for example, GI-GOF with a core dia-meter of 100 μm and a cladding diameter of 140 μm. Fig. 2.123 shows the fre-quency response of a 500 m long sample with three different launch conditions. At 200 MHz · km the results lie in the range of the fiber specifications.

The last fiber presented here is the semi-GI-PCS described above. The measurement conditions become extremely more noticeable here so that the measurement results shown may not be conclusively representative.


-8

-7

-6

-5

-4

-3

-2

-1

0

1

1 10 100 1000

AN = 0.10

AN = 0.34AN = 0.64

launch NA:


f [MHz]

500 m fiber at 650 nm

Fig. 2.123: Frequency response of a 100 μm GI-GOF

Fig. 2.124 first shows the frequency response with a 500 m long sample for 6 different launch conditions measured at a wavelength of 650 nm.

-12

-10

-8

-6

-4

-2

0rel. opt. power [dB]

frequency [MHz]

1 10 1003 30 300

AN Launch = 0.03 .. 0.64

500 m Semi-GI-PCS

Fig. 2.124: Bandwidth of a Semi-GI-PCS

The bandwidth-length product of the fiber was determined as having values between 24 and 55 MHz · km which clearly lies above the specification of 100 MHz · km. Bandwidths with their length and launch dependence are also determined for this type of fiber. Fig. 2.125 summarizes the results.


length [m]

100

1000

B3 dB [MHz]

30

300

3000

10 100 100020 20050 500

AN = 0.02

AN = 0.09

AN = 0.17

AN = 0.26

AN = 0.34

AN = 0.46

launch NA:

Fig. 2.125: Bandwidth measurement of Semi-GI-PCS

What is striking is the low dependence of the bandwidth on the launch con-ditions with longer sample lengths. Evidently, there is a significant exchange of energy between the SI and GI modes in the fiber. The specified bandwidth value could only be determined in short fiber lengths with under filled launch.

Bandwidth measurements on semi-GI PCS have also been published by [Aiba04] and [Aiba05], whereby a method was used in which a light pulse circu-lates in a 100 m long ring and passes an acousto-optic modulator after every pass. The numerical aperture of the coupling optics amounts to only 0.25 and SI modes are for the most part suppressed. The results for the frequency response, deter-mined by Fourier transformation, are shown in Fig. 2.126.

rel. opt. power [dB]

-10

-8

-6

-4

-2

0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

f [GHz]

1st circulation

10th circulation

Fig. 2.126: Frequency responses of a Semi-GI-PCS according to [Aiba04]

The bandwidths thus determined are shown in Fig. 2.127. The values lie higher by a factor of ten than the values measured with full launch on long fibers. This


reveals impressively how important correct specifications of the measurement conditions are when indicating bandwidth values.

The succinct statement in the Sumitomo data sheet that “the bandwidth can change under other measurement conditions” is of little help.

B3 dB, opt. [GHz]

semi-GI -PCS

length [m]

100 10006004002000.3

0.6

1.0

2.0

0.4

0.8

1.5

Fig. 2.127: Bandwidth of a Semi-GI-PCS according to [Aiba04]

2.5.3.6 Comparison of Bandwidth Measurements and Calculations

The diverse measurements of fiber bandwidths show that the same principles are essentially valid for thick glass and polymer fibers. Important effects are:

The bandwidth drops with the square of the numerical aperture by increasing the differences in propagation time among the individual modes. The diameter of the fiber does not play any role in regard to the bandwidth. Strong mode-dependent attenuation increases the bandwidth of fibers, but it also leads to a rise in transmission losses. Multicore fibers and fiber bundles permit smaller NAs with the same bending radius and thus greater bandwidths. The bandwidth of fibers greatly depends on the launch and detection condi-tions. The difference can be » 10 for short fiber lengths. When stating the bandwidth in data sheets, measurements should always be made with UMD (full launch) or EMD (equilibrium mode distribution). Graded index profiles increase the bandwidth up to two magnitudes. How-ever, the index profile must be as ideal as possible - it should be parabolic when the chromatic dispersion is disregarded. In the case of a non-ideal GI profile a large bandwidth can still be attained through a selective launch. In addition, the chromatic dispersion especially with glass GI fibers has to be taken into account (this will be discussed in the next paragraph). It is technically easier to produce a multi-stepped index profile, with which the bandwidth can clearly be increased, than a GI profile.


Semi-GI fibers have large bandwidths, above all over short lengths and when coupling into small angles. The bandwidth of individual fibers - not yet placed in cables - under labora-tory conditions can depend to a great extent on the external conditions, depen-ding on the degree of induced mode coupling.

A comparison between POF and PCS is particularly interesting since both can be used alternatively in many applications. The length-dependent bandwidths of both types of fiber with full launch are illustrated in Fig. 2.128.

length [m]

PCS, NALaunch = 0.48POF, NALaunch = 0.64

100

300

1.000

303 10 30 100

B3 dB, opt. [MHz]

Fig. 2.128: Bandwidth comparison of POF (fiber-NA: 0.50) and PCS (NA: 0.37)

Theoretically, the PCS should show about 50% greater bandwidth because of its smaller NA - which has just about been confirmed by measurements. Both measurement curves run approximately parallel which suggests similar magni-tudes in mode-dependent processes. The angle-dependent attenuation of a typical PCS fiber is illustrated in Fig. 2.129.

-25 -20 -15 -10 -5 0 5 10 15 20 25

fiber length 50 m 100 m200 m

excess loss [dB/km]

[°]0

50

100

150

200

25

75

125

175

225

Fig. 2.129: Mode dependent loss of a PCS (at 650 nm)


-30 -20 -10 0 10 20 300

100

200

300

400attenuation [dB/km]

[°]11 dB/km

100 m

50 m

Fig. 2.130: Mode dependent loss of a Semi-GI-PCS (at 650 nm)

PCS does indeed show very large mode-dependent attenuation, the intensity of which is comparable to POF. This explains the similar behavior even if the core material itself has a very much lower attenuation.

A schematic comparison of typical bandwidth values for the different multi-mode fibers described above are illustrated in Fig. 2.131. The values, as already mentioned several times, can clearly deviate for specific products or under diffe-rent measurement conditions.

bandwidth [MHz·km]1 10 100 1,000 10,000

MC-GOF

200 μm PCS

DSI-POF

SI-MC-POF

DSI-MC-POF

MSI-POF

OM-Giga

PF-GI-POF

PC-POF

St.-NA-POF

GI-GOF OM1

GI-GOF OM2

GI-GOF OM3

OM3 mit LD

Semi-GI-PCS

Ø: 1000 μm

Ø: 1000 μm

Ø: 1000 μm

Ø: 200 μm

Ø: 1000 μm

Ø: 1000 μm

Ø: 200 μm

Ø: 1000 μm

Ø: 750 μm

Ø: 900 μm

Ø: 62.5 μm

Ø: 50 μm

Ø: 120 μm

Ø: 50 μm

Ø: 50 μm

Fig. 2.131: Bandwidth comparison of different optical fibers (typical values)


The bandwidths of the fibers presented vary over more than 3 magnitudes. If singlemode fiber is used, however, then nowadays there is practically no longer any bandwidth limit. Mode dispersion no longer arises. Chromatic and polariza-tion mode dispersion can be compensated for as one likes. The significance of chromatic dispersion will be discussed in the next section.

2.5.4 Chromatic Dispersion in Polymer Optical Fibers

In all optical media we can observe the effect that the speed of propagation of light of different wavelengths differs. When we differentiate the propagation con-stants according to wavelength, we obtain the so-called chromatic dispersion, usually expressed in ps/nm·km. This constant indicates by how much a signal's delay will vary with the wavelength. In the typical application range of optical fibers this value is negative which means that with increasing wavelength the delay becomes smaller (corresponding to greater speed). Figure 2.132 shows the chromatic dispersion for silica glass, PMMA and a typical fluorinated polymer (according to [Koi97a]).

-1,200

-1,000

-800

-600

-400

-200

0

200

400 1,000500 600 700 800 900 1,200 1,400 1,600

dispersion [ps/(nm km)]

wavelength [nm]

PMMA

silica glass

PF-Polymer

Fig. 2.132: Dispersion of different materials

Typical semiconductor sources feature certain spectral widths that range from some 10 nm for LED up to a few MHz for lasers (corresponding to some 10-5 nm). In addition, there is the fact that when a light source is modulated there is always a spectral broadening that cannot be less than a certain theoretical limit. This effect only plays a role, however, with spectral singlemode lasers and with very high data rates.

Figure 2.133 shows a schematic illustration of the effect of chromatic disper-sion on a light pulse that has a given spectral width. A pulse with a certain spec-


trum of the width is launched into the fiber. After passing through the fiber (length L) and experiencing a certain amount of dispersion D, the pulse has the width = D · L · , whereby the shorter wave components arrive first. (cf. Fig. 2.38 as well).

t t

t = D L

spectral shape of the source

input pulse output pulse broadening by time

POF

length L

Fig. 2.133: Influence of chromatic dispersion

For silica singlemode fibers, the value for chromatic dispersion at 17 ps/nm·km lies within the range of the smallest fiber attenuation at 1,550 nm wavelength. Today, DFB-laser diodes are predominantly used for long-distance systems, the spectral width of which is a maximum of a few MHz. What matters here essen-tially is the broadening effect that is brought about by the data itself. In this case, 1 nm corresponds to approximately 125 GHz of spectral width. This means that for a data rate of 10 Gbit/s a spectrum in the range of one-tenth nm is generated. Where the permissible bit broadening is 0.05 ns, the fiber link may have a length of approximately 30 km. For 2.5 Gbit/s this value rapidly increases to approxima-tely 500 km due to the narrower spectrum and the greater pulse broadening per-mitted. Conventional 2.5 Gbit/s systems can operate without specific actions against dispersion. However, all systems that have many inline fiber amplifiers or higher bit rates require devices to counteract chromatic dispersion. The most common method today is the use of dispersion compensating fibers with strong negative dispersion. Since these fibers utilize waveguide dispersion they can only be produced as singlemode fibers.

The situation is significantly different for POF. The chromatic dispersion of PMMA-POF with over 300 ps/nm·km at 650 nm wavelength is over 20 times lar-ger than of silica fibers at 1,550 nm wavelength. For POF it is also usual to use LED with a typical spectral width of 20 nm to 40 nm and not lasers that have just a few tenths of a nanometer of spectral width. On the other hand, there are the typically short distances of POF systems and the moderate bit rates. Table 2.15 lists some examples for the effect of chromatic dispersion in POF systems.


Table 2.15: Influence of chromatic dispersion in POF systems

Example Bit Rate/ POF-Length

Wavelength/ Spectr. Width

Pulse Broadening/ rel. to the Bit Length

SI-POF 50 Mbit/s / 50 m

650 nm LED 20 nm FWHM

0.375 ns 2 % of the bit time

ATMF DSI-POF

155 Mbit/s / 50 m



ATMF DSI-POF

155 Mbit/s / 100 m



IEEE1394 MC-POF

500 Mbit/s / 70 m



STM16 PF-GI-POF

2,500 Mbit/s /200 m

650 nm LD 2 nm FWHM


The first three examples are based on LED for transmitting data rates up to 155 Mbit/s over a maximum length of 100 m. Even in the unfavorable case of using green LED, pulse broadening is less than one ½ the bit length so that there is only a small effect on the system. In the fourth example, the intention is to trans-mit an IEEE1394 S400 data stream (with 500 Mbit/s physical data rate) over a distance of 70 m using a green LED. Here pulse broadening is nearly in the same range as the bit length. When this deteriorating effect due to mode dispersion is added, one can see that this system can only work with considerable additional efforts. It may, for example, be possible to partially provide electrical compen-sation, whereby higher optical receiving power is required. When using data rates from ½ Gbit/s to 1 Gbit/s, the use of spectrally narrower sources becomes neces-sary. These primarily include RC-LED and VCSEL (see Chapter 4), and for even higher requirements DFB laser diodes. In most cases this selection is required anyway due to the limited modulation bandwidth of LED.

Fluorinated graded index profile polymer fibers feature significantly reduced chromatic dispersion compared with PMMA-POF. These fibers are designed for use in Gbit/s systems operating at spectral ranges between 800 nm and 1,300 nm. It is for these demands only that laser diodes can be considered, not least due to the smaller core diameters, the spectral width of which is a few nanometers at most. The last row shows that in such a case chromatic dispersion can be neglec-ted even for a transmission length of a few 100 m.

2.5.5 Methods for Increasing Bandwidth

Generally, the theoretical bandwidth of polymer fibers is calculated on the basis of two essential assumptions. One assumption is that the launch of light at the fiber entrance takes place in uniform mode distribution and that the detector will re-ceive all modes. The second assumption is that the attenuation of all modes is nearly constant. However, in practice polymer fibers, and in particular step index fibers, show completely different behavior. In the first place, it is relatively diffi-cult to illuminate all modes uniform at the entrance of the fiber. In many cases


laser diodes are used where the emitting angle is significantly smaller than the angle of acceptance of a SI-POF. The use of solid-state lasers or gas lasers, the exact wavelength of which is often required for measuring purposes, is even more problematical. These lasers emit collimated light so that only a small proportion of the POF modes can be excited. When using glow lamps or discharge lamps, opti-cal devices are used to collimate the light to the fiber. For this reason it is difficult to find lenses that actually work with consistent efficiency in the given acceptance range. All this has the effect in a concrete experiment of increasing the deviations of the actual bandwidth in comparison with the theoretical limit value. This is a very undesirable effect when attempting to define characteristics by making measurements of this kind, as shown in Chapter 7. However, for high bit-rate data transmission this situation can in practice also be beneficially exploited as shown by the following examples.

Figure 2.134 demonstrates the most important methods for increasing the bandwidth of a POF.

launch with small angle fiber without

connectors, bends and splices IN OUT

high pass filter for dispersion

precompensation

high pass filter for dispersion

postcompensation

detection with small angles

Fig. 2.134: Methods for increasing bandwidth (cf. p. 441 as well)

Launching light at a small angle as well as detecting just a selected angle range has the effect of restricting the modes involved in signal transmission and thus reducing pulse broadening. It is possible to electrically compensate for the resul-ting low pass behavior, both before as well as after the POF link. To date the most significant increases in bandwidth for a POF system have been described in [Bat92] (see also [Bat96a] and [Yas93]). The following components were utilized:

Launch with a small AN = 0.11, thereby exciting only a few modes with only small differences in delay.Pre-distortion of the LD excitation signal (peaking); high pass (33 pF 51 ).Detection with low NA (modes with large delay differences are blanked out).Dispersion compensation behind the receiver; high pass (8 pF 200 ).

It was possible to transmit at more than 500 Mbit/s across a distance of 100 m of standard NA-POF (see also chapter 6). However, all these measures are usually at the expense of a reduced power budget, as summarized in Table 2.16.


Table 2.16: Consequences of different bandwidth increasing methods

Method Penalty for the Power Budget due to:

peaking lowering of source modulation depth

low NA launch decreasing POF coupled optical power for sources with broad emission angle

low NA detection loss of light with high propagation angle at the fiber output

post compensation amplified noise at higher frequencies

It follows that the use of such methods is of particular interest in systems that have adequate power reserves. POF attenuation across very short distances is hardly of importance; on the other hand, the use of high data rates is of interest in various applications. Chapter 6 will describe experiments for transmitting Gbit/s over distances of 10 m to 100 m conducted by T-Nova GmbH, the University of Ulm, Daimler Chrysler, the Fraunhofer Institute for Integrated Circuits Nuremberg and the POF-AC Nürnberg.

Figure 2.25 shows theoretical considerations with respect to the POF bandwidth at different launching angles (Gaussian shaped far field with 3 dB width calculated relative to fiber NA) according to [Bun99a]. With short lengths and small launch NAs the light remains concentrated in areas with small propagation angles. The small differences in propagation time result in large bandwidths. After approx. 100 m of fiber equilibrium mode distribution is just about reached through mode mixing, and the influence of the launch conditions gradually disappears. This be-havior corresponds to a great degree to the measurement results described above.

length [m]

B3 dB [MHz]

10 100 5005020 20010

100

1,000

20

50

200

500

0.50.71.01.2

1.51.72.0

rel. launch NA (NAfiber = 1)

Fig. 2.135: Theoretical bandwidth with different launching conditions ([Bun99a])

The principle of peaking is demonstrated in Fig. 2.136 and 2.137 ([Zam00b], [Ziem00a] and [Ziem00c]). A high pass filter which dampens lower frequencies and lets high frequencies pass through without losses is switched between the modulation input and the laser. We begin with an illustration of the electrical spec-trum of the emitting signal at the laser with and without peaking (1.2 Gbit/s, NRZ, pseudo-random bit sequence).


-80

-70

-60

-50

-40

-30

-20

-10

0 rel. power [dB]

frequency [GHz]

without peaking

Giehmann

with peaking

1.50.5 1.0 2.0 2.5 3.0 3.5 4.00.0

Fig. 2.136: Modulation spectrum with and without peaking

In the experiment, the data rate was 1,200 Mbit/s with NRZ coding. The twin-stage pre-distortion filter dampens the signal by 12 dB in the low frequency range so that the higher frequencies can create a stronger modulation. For the pulses this means steeper edges and overshoot at the beginning and end, hence the term peaking, as shown in Fig. 2.137.

0 10 20 30 40 500

2

4

6

8

10

12

14

16

18

Giehmann 2000

relative amplitude

time [ns]

with peakingwithout peaking

60

Fig. 2.137: Laser modulation signal with/without peaking

The disadvantage of peaking can be clearly recognized in the diagram. The peaks at the beginning and the end of the pulse must lie within the admissible ope-rating range of the laser, i.e. between the threshold current and the maximum cur-rent. This reduces the actual power per pulse compared with rectangular pulses.

Figure 2.138 summarizes the bit rates and transmitted distances of different high-rate transmission systems using SI-POF ([Scha00], [Ziem00a], [Kich99] and[Yas93], [Vin04a], [Vin04b] and [Ziem03g]). Chapter 6 contains detailed expla-


nations of the different systems. The diagram also shows the theoretical limits for the bandwidth of standard NA-POF and DSI-POF (assuming NRZ coding and bit rate = 2 3-dB bandwidth).

100

1,000

3,000

200

500

10 100 20020 50

bit rate [Mbit/s]

POF-length [m]

AATTMM 115555

110000BBaasseeFFXX

BBaatteess ´́9933

IIEEEEEE 11339944 BBaatteess ´́9922

KKaaiisseerr ´́9922

UUNNII UUllmm

DDaaiimmlleerrCChhrryysslleerr

TT--NNoovvaa

PPOOFF--AACC

PPOOFF--AACC

PPOOFF--AACC

system with St.-NA system with DSI-POF system with MC-POFUMD-limit St.-NAUMD-limit DSI-POF

PPOOFF--AACC

Fig. 2.138: Bit rates of different POF systems (status 2003)

It is easily discernible that a number of systems with standard NA-POF are sig-nificantly above the theoretical limits. Particularly for greater lengths the potential for exceeding the limit is considerable. As shown in the next section, the practical application presents some problems such as the bending behavior. It is generally true that extreme dispersion compensation must be adaptive in its execution. That means that above all the limit frequencies of the high passes must be adapted very precisely to the frequency response of the link. If the frequency response changes, the result will be too much or too little compensation so that the pulses become distorted. Such a change may, for example, occur as a result of different lengths of cable; however even a bend in the fiber may have the same effect. In commercial systems it is desirable to avoid having to use automatic adaptations, such as is necessary, for example, in 1000BaseT-systems, or having to provide specific receivers for different cable lengths. One practical solution is to adjust the com-pensation in such a way that there is just-tolerable over-compensation for short lengths of fiber; based on this level of compensation, the next step is to select the maximum fiber length for which this compensation is still just sufficient. A sche-matic illustration is shown in Fig. 2.139.


increasing transmission distance

f

frequency response of the POF link (dispersion limited)

f f f

f

compensation filter (fixed)

f

resulting frequency response

f f f

overcompen-sation

optimized compensation

undercompen-sation

Fig. 2.139: Compensation of dispersion for various transmission distances

A proposal for increasing the bandwidth by direct interference with the optical path is described in [Kal99]. By using a mode filter immediately after the trans-mitter, the light angle range in the fiber is reduced as shown schematically in Fig. 2.140. With this method it was possible to achieve an improvement in band-width by 53% and 89% respectively for two standard NA-POF provided by Mitsubishi and Toray. The losses of the mode mixer are approximately 2.5 dB which is perfectly acceptable in many applications.

sourcemode filter

POFreceiver

AN 0.43 AN 0.29

Fig. 2.140: Increased bandwidth with a mode filter ([Kal99])

Basically, this method is equivalent to the method of light launching using a smaller NA, though probably much easier to implement because no optical com-ponents are required and only a simple mechanical clamp needs to be placed on the fiber. If required, this can be repeated in the middle of the link or before the receiver.


2.5.6 Bit Rates and Penalty

In general, optical transmission systems are set up in such a way that the system bandwidth amounts to at least 50% of the bit rate with NRZ transmission. Thus, 500 MHz are needed to transmit 1000 Mbit/s. This means that the eye is complete-ly open with ideally adapted filtering. In other words, the transition from the zero symbol to the one symbol and vice versa takes place within the bit duration. If the system bandwidth is smaller than half of the bit rate, then the symbol transition needs more time resulting in a reduction of the vertical eye opening. This effect must either be compensated for through adapted filtering or the reduced eye opening is compensated for by a correspondingly higher receiving level. The deterioration of the signal-to-noise ratio at the receiver through the bandwidth limitation is called penalty (measured in dB). The relationship between signal-to-noise ratio, receiving level and penalty is shown in Fig. 2.141.

system without noise and with sufficiend bandwidth - the eye open completely

t

U

t

U

U1U2

system without noise and with limited bandwidth - the eye in closed partially penalty: 20·log(U2/U1)

system with noise and with sufficient bandwidth - the eye is open completely SNR = 20·log(US/UN)

system with noise and with limited bandwidth - the eye is more closed SNR is decreased by penalty

t

U

t

U

US

UN

Fig. 2.141: Definition of penalty


When describing the sensitivity of a receiver, measurements are always made at the maximum bit rate. A possible penalty is always included. If there is sufficient-ly large bandwidth, then only the noise should limit the sensitivity. The large diode capacitance produces as a rule a relatively dramatic low-pass effect when using large photodiodes, which are necessary for POF or PCS. The noise rises in proportion to the signal with decreasing receiver impedance, i.e. one will accept some penalty and work with as large an input resistance as possible.

Under laboratory conditions data communication can also be carried out with high penalty. Modern bit error analyzers can transmit error free as long as the eye opening amounts to some 10 mV. A typical eye diagram with high penalty is shown in Fig. 2.142. Subsequently, the connection between system bandwidth and penalty for a broad-band receiver at the POF-AC is illustrated.

Fig. 2.142: Data transmission with large penalty

In the example shown 820 Mbit/s were transmitted over 100 m of DSI-POF. Although the eye was almost completely closed, an error free transmission was possible. In a real system, however, certain detection would be relatively difficult since the sampling moment and the decision threshold have to be re-adjusted very exactly. Furthermore, there are no margin whatsoever for fluctuations in the laser power or bending losses.

bandwidth/bit rate [MHz/Mbit/s]

0

5

10

15

20

25

simulated

with fiber

penalty [dB]

0.10 1.000.20 0.30 0.40 0.60 0.80

Fig. 2.143: Effect of system bandwidth on the penalty

2.6 Bending Properties of Optical fibers 143

The simulated values were determined by calculating the penalties with the aid of PSpice analyses. A Gaussian-shaped filter was used as a low-pass system. The measuring points were determined on a 20 m long standard POF with different bit rates. The penalty was estimated from the eye diagram. The measured values tallied greatly with the simulation down to 25% of the system bandwidth, e.g. a transmission of 1 Gbit/s with a system bandwidth of 250 MHz. With higher bit rates the penalty increases more quickly than in the simulation. One main reason is that the frequency response only corresponds to a certain degree to idealized Gaussian behavior. It hardly makes sense to use practical systems with more than a 10 dB penalty.

The results show that an exact relationship does not have to necessarily exist between the maximum bit rate and the fiber bandwidth. Furthermore, even with bandwidth-limited systems relatively high data rates can be achieved under labora-tory conditions if enough emitting power is available.

2.6 Bending Properties of Optical fibers

The sensitivity of optical fibers to bending is of special significance. In practical applications installed links are never completely straight. Often they are fitted around corners where 90º bends are a common occurrence. Even along a link in a straight cable duct there are many small bends, for example, in places where cables are hold with cable ties.

When being assembled, the fiber must also be able to withstand mechanically tight bends. In many applications there is continuous bending during operation, for example in drag chains or with a data cable in a car door. That is why one diffe-rentiates between different bending loads:

Static bending: involves the assessment of how much light is loss in bends. These losses are to be taken into account in the power budget of the system. The bending loss is measured in dependence of the bending radius. Minimum bending radius during assembly: only characterizes what bends the fiber can tolerate for a short time without being mechanically destroyed. Repeated bending: in certain applications fibers must be able to tolerate 105 to some 106 bends without being mechanically destroyed. Reel change bending: arises in particular in drag chains (see also Chapter 9).

The following results have either been taken from data sheets or come from measurements made at the Deutsche Telekom and as of 2000 at the POF-AC Nürnberg. Now as before no standards exist for measurements of bending attenu-ation. We mostly used a long fiber sample stimulated with as large a NA as pos-sible and all modes were detected with the aid of an integrating sphere. On the other hand many manufacturers measure with a small NA whereby much better values automatically come about because the outer modes are more strongly radiated in the bends.

144 2.6 Bending Properties of Optical fibers

Nevertheless, the series of measurements cannot always be compared exactly. In addition to the wavelength the bending attenuation can also depend to a great extent on the primary coating material and on the lateral forces within the bend. The coupling and detection conditions are also always included with short samp-ling lengths.

2.6.1 Bending Losses in SI-POF

The essential parameters which determine the bending sensitivity of a fiber are the diameter and the numerical aperture. The larger the NA, the narrower the permis-sible bending radii may be in relation to the fiber diameter. Figures 2.144 and 2.145 show the losses for bends of different commercially available fibers accor-ding to information in the data sheets ([Tor96a] and [Asa97]).

The Fig. 2.144 shows the bending losses of two different SI-POFs with some-what different NAs. You can clearly see that larger NAs reduce the bending losses.

bend radius [mm]

0,0

1.0

2.0

3.0

4.0

5.0

0 5 10 15 20 25 30 35 40

bend losses [dB]

fiber

PFU-CD-1001 AN = 0.46

PGU-CD-1001 AN = 0.50

Fig. 2.144: Loss for 360° bend according to [Tor96a]

Figure 2.145 shows losses resulting from bends in a standard NA-POF, a low-NA-POF and a multi-core fiber (see Chapter 2.3).

The low-NA-POF shows significantly larger losses compared to a standard NA-POF. Due to the smaller individual core diameters, the bending sensitivity of the multi-core fiber is comparable with that of the standard NA-POF despite the smaller NA.

If many bends directly follow each other, attenuation does not increase propor-tionally with the number of bends because there is less and less energy present in the higher mode groups. Figure 2.146 shows a measurement of the bending losses for different POF according [Hen99].


bend radius [mm]0

1

2

3

4

5

6

0 10 20 30 40 50 60 70 80

bend losses [dB] fiber

TC 1000 (AN = 0.485)

NC 1000 (AN = 0.25)

NCM 1000 (AN = 0.25)

Fig. 2.145: Loss for a 360° bend according to [Asa97]

0 1 2 3 4 5 6 7 8 9 10

loss [dB]NC 1000 (Low-NA) no longer available at

the market

AC 1000 (DSI) PFU 1000 (St.-NA)

MH 4000 (DSI-POF)0.10

1.00

10.0

0.02

0.05

0.20

0.50

2.00

5.00

number of turns

Fig. 2.146: Bending loss depending on number of turns ([Hen99])

The measurements were taken at 650 nm with LED-launch and a mode mixer. The bending radius was 32 mm and the bends were located at the beginning of a 50 m sample length.

PFU 1000 is a standard NA-POF, while MH 4000 and AC 1000 are double-step index POF. Their losses are approximately identical and up to 10 windings are significantly below 1.0 dB. By comparison, the low-NA-POF NC 1000 is in the range of 10 dB, which is too much for deployment in practical applications. The ATM forum stipulates an admissible bending radius of 25 mm and at this radius the attenuation was already above the range of measurement. Meanwhile, DSI-POF offer significantly improved bending characteristics at comparable NA.

Figures 2.147 and 2.148 demonstrate the losses over the inverse bending radius and the number of windings for a (genuine) low-NA-POF (NC 1000) and a stan-dard NA-POF [Hen99].


39 mm

26 mm32 mm

15 mm

21 mm

12 mm

0

2

4

6

8

10

12

0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

inverse bend radius [mm-1]

loss [dB]

2 turns

4 turns

6 turns

8 turns

10 turns

Fig. 2.147: Bending loss of a PFU-CD-1000 ([Hen99])

Under UMD conditions, the bending losses should increase proportionally to the inverse bending radius. In practice, however, this only takes place below a bending radius of around 20 mm. It appears that the real equilibrium mode distri-bution reduces the losses above a certain radius. The reason for this is the smaller weighting of modes that have a large propagation angle, which, as already men-tioned, are particularly sensitive to losses at bends.

Basically the low-NA-POF in Fig. 2.148 shows the same behavior, albeit for significantly greater radii and due to the smaller NA.

inverse bend radius [mm-1]

0

2

4

6

8

10

12

14

16

0.020 0.022 0.024 0.026 0.028 0.03 0.032 0.034 0.036

loss [dB]

50 mm

32 mm

39 mm

28 mm

2 turns

4 turns

6 turns

8 turns

10 turns

Fig. 2.148: Bending loss of a NC-1000 ([Hen99])


2.6.2 Bending Losses in GI Fibers

For graded index POF slightly different conditions apply for bending sensitivity compared with step index profile fibers. Here it is not the total reflection at the core-cladding interface but the continuous bending in the index profile that is res-ponsible for light guiding. In addition, there is a fundamentally different distri-bution in the near and far field. Figure 2.149 shows a measurement for GI-POF according to [Ish95].

0.2

1

10

20

0.5

5

2

0 10 15 20 25 30 35 40 45 50 55

bend radius [mm]

GI-POF, doped by:MMA/DPS, AN = 0.29

MMA/BBP, AN = 0.21

bend losses [dB]

Fig. 2.149: Loss of two GI-POF ([Ish95]) for a 90° bend

Due to the different dopants used, the two samples with a core diameter of 0.5 mm each have a different NA, which has a very significant effect on the ben-ding losses. Despite the smaller core diameter the losses for a 25 mm bend are still significantly higher than the values for a SI-POF or a DSI-POF. Here, too, a reduction in the core diameters leads to lower bending losses.

[Aru05] describes how the bending losses in PMMA GI POF can be signifi-cantly reduced. In addition to an optimized index profile an additional PVDF layer (polyvinylidenfluoride) was applied to the core with parabolic profile resulting in a semi-GI-POF which combines high bandwidth with low bending losses. The losses of a 90° bend are shown later in Fig. 2.205 compared with a conventional PMMA GI-POF. (The sample length was 100 m.) Even with a bending radius of 5 mm there was no measurable increase in attenuation. The different methods for reducing bending losses in PMMA GI-POF and PF-GI-POF are described in greater detail in Section 2.8 on fiber production. Examples of measurements are also shown.

2.6.3 Change of Bandwidth by Bends

However, bends do not only contribute to additional losses, but also have an effect on bandwidth because certain mode groups are selectively attenuated. This effect is exploited in mode filters and mixers.


Figure 2.150 (according to [Rit93]) shows what the effect of a 720º bend at the beginning of a 50 m long POF link has on the measured bandwidth. In this case the light is launched with AN = 0.10.

60

80

100

120

140

160

0.00 0.04 0.06 0.08 0.10 0.12

bandwidth [MHz]

ØPOF:

250 μm

500 μm

750 μm

1000 μm

rel. inverse bend radius [ØPOF-1]

0.02

Fig. 2.150: Change of bandwidth by bending the fiber according to [Rit93]

Due to the low launch NA, the bandwidth is relatively large (80 MHz · 100 m). In the case of tight bending radii at the beginning of the fiber there is mode mixing so that the bandwidth is significantly reduced sometimes. This effect is naturally more pronounced for smaller diameters. In the illustration selected here above the inverse relative bending radius, relative to the core diameter, the effect of the core diameter should disappear. It seams to be, that the effect described above of the larger bandwidth for thinner fibers is already dominant here due to the more mode dependent processes.

Comprehensive investigations of the effect of bends on the bandwidth of POF links were presented in [Mar00]. The test fiber consisted of a 100 m long standard NA-POF; 360º bends were inserted at the beginning of the link, after 25 m, after 50 m, after 75 m or at the end of the link. The source consisted of a 655 nm laser diode, the NA of which could be adapted through different optics from 0.10 to 0.65. The bandwidth and the attenuation of the overall link were measured without bends and with bending radii of 6.4 mm, 11.1 mm and 13.8 mm. The results are shown in Fig. 2.151.

When light is launched into the fiber using a large NA, the original bandwidth of approximately 33 MHz can be increased significantly. However, large improve-ments with small bending radii occur at the expense of large additional losses. The biggest gain in bandwidth is obtained with a bend in the middle because this means that many modes of the first 50 m are filtered out and EMD is not com-pletely regained in the remaining 50 m. The changes in attenuation are largely independent of the length since the mode field is well filled out everywhere.


0 25 50 75 100

bandwidth over 100 m [MHz]

30

40

50

60

70

80

90

100

-5-4-3

-2-1

loss [dB] bend position [m]

0 25 50 75 100

bend position [m]

launch NA: AN = 0.65 launch NA: AN = 0.10

radius 6.4 mm radius 11.1 mm radius 13.8 mm

bandwidth without bendings

Fig. 2.151: Influence of a bend to bandwidth and attenuation ([Mar00])

When light is launched into the fiber using smaller NA, the relative gain in bandwidth compared to the original - approximately - 60 MHz is not as big. Therefore the optimum position for the bend is clearly nearer to the end since the mode field must first be filled. Again, tight radii have more effect. The additional attenuation increases significantly when the bends are moved to the end, since at the beginning of the fiber there are hardly any higher mode families in existence. These results also confirm clearly for the existing assumptions with respect to mode propagation in a coupling length of some 10 m.

2.6.4 Bends on PCS, Multicore Fibers and thin POF

A very simple method to decrease bending radii is to reduce the core diameter while otherwise retaining identical parameters. If you wish to maintain the advantage of the simple handling of ready-made thick fibers, then there is the possibility of fiber bundles or multicore fibers respectively.

Fig. 2.152 and 2.153 show the measured bending losses, each with a bend of 360° in the middle of the sample, with UMD launch and measured with an inte-grating sphere. A 10 m long fiber was used for the MC-GOF. The range of the bending radii lay between 2 mm and 100 mm. The bending attenuation measured lies below 0.1 dB.


0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.06

0.05

0.04

0.03

0.02

0.01

0.00

bending loss [dB]

one bend by 360°

inverse bending radius [mm-1]

Fig. 2.152: Bending losses of MC-GOF, Schott

The bending losses of the MC-POF were measured on a 100 m long sample in order to guarantee as much mode equilibrium as possible. The bend (360°) was made in the middle of the fiber length. Due to the different relations between mode coupling and absorption the EMD conditions for 520 nm and 650 nm only differ slightly. That is the reason for the somewhat different bending losses.

0.01

0.10

1.00

0 5 10 15 20 25 30

bending loss [dB]

bending radius [mm]

at 520 nm

at 650 nm

100 m fiber length

Fig. 2.153: Bending losses of MC-POF, 37 cores, 1 mm total diameter

In many areas the 200 μm PCS is used because it permits smaller bending radii. Fig. 2.154 illustrates quite graphically that the same physical characteristics are also valid for these fibers. Here the bending losses are given versus the relative bending radius in relation to the fiber diameter. The numbers in brackets indicate the bending radius in millimeters for the PCS. Both fibers thus have in relative terms an identical bending sensitivity.


0 5 10 15 20 25 30 35 40 45 50

3.0

2.5

2.0

1.5

1.0

0.5

0.0

4.0

3.5

bending radius [ Kern]

200 μm PCS

1 mm POF

bending loss [dB]

(0) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Fig. 2.154: Bending losses of PCS and POF in comparison

Thin POF could be used as an alternative to PCS in many areas when tight bending radii are indeed required, but the attenuation and the temperature range of the POF are satisfactory. A comparison between a 250 μm SI-POF and a 200 μm PCS, measured at 650 nm with full launch for 5 m long samples, is illustrated in Fig. 2.155.

0 1 2 3 4 5 6 7 8 9 100.01

0.1

1.0

10.0

0.03

0.3

3.0

bending loss [dB]

250 μm POF

200 μm PCS

bending radius [mm]

Fig. 2.155: Bending losses of small diameter POF and PCS in comparison

The somewhat thicker POF also has somewhat higher bending losses. A tenth of a dB is attained for the POF at a bending radius of 8 mm and 6 mm for PCS.

The bending losses of three different SI-POFs with different NAs are compared in Fig. 2.156. The lowest losses are shown by the 300 μm thick POF with a high


NA. The 250 μm and 500 μm thick POFs have almost identical bending attenu-ations. It is thus indicative that the NA is by far the most important factor for the bending losses. Consequently, you should always choose fibers with the largest possible NA for particularly tight radii, unless you decide to go back to multicore fibers. In addition, the latter have the advantage of offering an even greater bandwidth.

0.001

0.01

0.10

1.00

10.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13

bending radius [mm]

bending losses [dB] 250 μm POF (AN = 0.63) 500 μm POF (AN = 0.50) 300 μm POF (AN = 0.63)

Fig. 2.156: Comparison of bending losses of various SI-POF (different NA)

More recent measurements of bending losses of four different SI fibers, each with cladding and made available from Toray Germany, are shown in Fig. 2.157. Fibers with a large NA (0.63) were used for this measurement. They allow considerably smaller bending radii without remarkably decreasing the attenuation and bandwidth.

[dB]

0.01

0.1

1.0

0.03

0.3

3.0

1.0 10.00.3 3.0 30.0

250 μm

500 μm

750 μm 1000 μm

7.5 r 7.5 r

7.3 r

8.0 r

r [mm]

Fig. 2.157: Comparison of bending losses of various SI-POF


Bending radii are drawn in the picture with which a bend (360°) results in exactly 1 dB additional attenuation. With the four fibers with their 250 μm to 1000 μm core diameters this is the case each with a seven-fold to eight-fold fiber radius, i.e. a bending radius between 0.9 mm and 4 mm. As a comparison, the bending losses of a 125 μm SI-POF ([Witt04]) are shown in Fig. 2.158.

bending losses [dB]

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00.1

1.0

0.2

2.0

0.5

bending radius [mm]

Fig. 2.158: Bend losses of a 125 μm SI-POF ([Witt04])

Optimedia has made available samples of a thinner PMMA GI-POF. The ben-ding losses of this fiber with overfilled launch (LED) and a launch with a laser (AN = 0.10) are shown in Fig. 2.159. Both measurements were carried out at 650 nm with a 5 m long fiber.

0 20 40 60 80 100bending radius [mm]

bending losses [dB]

OM-Giga 500 μm/750 μm

360°-bends = 650 nm 5 m fiber

0.1

1.0

0.2

2.0

0.5

5.0

laser launch

overfilled (LED)

Fig. 2.159: Bending losses of a 500 μm PMMA GI-POF

A bending radius of 15 mm is still not a problem for collimated light whereas a high bending attenuation arises below a bending radius of 30 mm with an LED launch. You could argue, of course, that laser sources should always be used for GI fibers in order to utilize the high bandwidth. Nevertheless, it is imperative that manufacturers reduce the bending losses.


Finally, some results from a project work [Bau06] are shown. First, Fig. 2.160 compares the bending losses of three Toray fibers with different core diameters: 500 μm, 750 μm and 1,000 μm. The NA of the three fibers is the same. As expec-ted, the bending radius for a given attenuation is reduced nearly proportional to the fiber diameter. Only with very thin fibers does the effect of stronger mode-depen-dent attenuation make itself noticeable.

0.001

0.01

0.1

1

10

0 10 20 30 40 50

bending losses [dB]

bending radius [mm]

fiber type: Toray PFU AN = 0.47

measured with 650 nm LED 1 bend 360°, 10 m fiber

750 μm

1,000 μm

500 μm

Fig. 2.160: Bending losses of various standard-NA-POF

The bending losses of 1 mm POF from three manufacturers are compared in Fig. 2.161. Since the NAs of the fibers are not exactly equal, the bending attenu-ations differ somewhat. In practice, however, these small deviations should hardly play a role.

0 10 20 30 40 50bending radius [mm]

0.01

0.1

1

10bending losses [dB] fiber type: St.-NA

measured with 650 nm LED 1 bend 360°, 10 m fiber

Fig. 2.161: Bending losses of various standard-NA-POF

2.7 Materials for POF 155

2.7 Materials used for POF

2.7.1 PMMA

The material most frequently used for polymer fibers is the thermoplastics PMMA (Polymethylmethacrylate), better known as Plexiglas . Figure 2.162 shows the structure of the monomer and its polymer chain.

OC

O

C

C

C

MMA

OCH3

C

O

C

CH3

CH2

PMMA

OCH3

C

O

C

CH3

CH2

OCH3

C

O

C

CH3

CH2

OCH3

C

O

C

CH3

CH2

H

H

HH

H

H

H

HC

Fig. 2.162: Molecular structure of PMMA

PMMA is produced from ethylene, hydrocyanic acid and methyl alcohol. It is resistant to water, lyes, diluted acids, petrol, mineral oil and turpentine oil. PMMA is an organic compound forming long chains with typical molecule weights around 105. Essential from the point of view of optical transparency of the material is the amorphous structure of the polymerized material. The density of PMMA is 1.18 g/cm3. Its tensile strength is approximately 7-8 kN/cm2 ([SNS52]). The refractive index of PMMA is 1.492 and the glass transition temperature Tg

lies between +95°C and +125°C. At room temperature and 50% relative humidity the material can absorb up to 1.5% water, which also affects the attenuation characteristics.

Table 2.17 presents further properties of PMMA:

Table 2.17: Properties of PMMA (typical values)

Parameter Unit Value

refractive index - 1.492 glass transition temperature Tg °C 115 density g/cm³ 1.18 absorption of water up to saturation % 0.5 thermal conductivity: W/m K 0.17thermal heat expansion coefficient: mm/m K 0.07Rockwell hardness (M), Shore hardness (D)

- 95 70

tensile strength N/mm² 76 resistivity Ohm cm 1015

breakdown strength kV/mm 20 - 25 spontaneous combustion temperature °C approx. 430

156 2.7 Materials for POF

As can be seen in the illustration, each MMA monomer has a total of eight C-H bonds. The vibrations of this compound, or more precisely its harmonic waves are a main cause for the losses encountered in PMMA polymer fibers. The attenuation resulting from absorption at the respective wavelength is shown in [Mur96] and [Koi96c] (see Fig. 2.163 and table 2.18). In particular the harmonic waves at 627 nm (6th harmonic wave) and 736 nm (5th harmonic wave) essentially deter-mine the level of attenuation within the application range of PMMA-POF because these are not narrow absorption lines but relatively wide bands. Further causes for attenuation will be discussed in the chapter titled Characteristics.

10-8

10-6

10-4

10-2

100

102

104

106

108

500 1000 1500 2000

attenuation [dB/km]

wavelength [nm]

C - H

C - D

C - F

C - Cl

molecule

Fig. 2.163: Absorption lines of C-X-bounds according to [Gra99] and [Mur96]

Quite early in the history of this technology, the idea came up to reduce the ab-sorption losses of polymer fibers by using different materials in which less or no C-H bonds were present. However, it is not easy to eliminate these; instead, the hydrogen atoms are replaced by other atoms of the 7th main group. A heavier core will result in a lower vibration frequency, thus moving the attenuation bands to a larger wavelength. The illustration shows the attenuation bands for deuterium (heavy hydrogen with the atomic weight 2), fluorine (atomic weight 19) and chlorine (atomic weight 35 or 37, see also [Bau94]). Generally, the materials for polymer fibers can be divided into three groups:

compounds containing hydrogen compounds with partial substitution of hydrogen compounds with complete substitution of hydrogen


Table 2.18: Absorption bands position of carbon bonds ([Gra99])

Oscillation C-H [nm]

C-D [nm]

C-F [nm]

C-Cl [nm]

C=0 [nm]

O-H [nm]

0 3,390 4,484 8,000 12,987 5,417 2,818

1 1,729 2,276 4,016 6,533 2,727 1,438

2 1,176 1,541 2,688 4,318 1,830 979

3 901 1,174 2,024 3,306 1,382 750

4 736 954 1,626 2,661 1,113 613

5 627 808 1,361 2,231 934 523

6 549 704 1,171 1,924 806

7 626 1,029 1,694 710

8 566 919 1,515 635

9 830 1,372

2.7.2 POF for Higher Temperatures

Fibers with high resistance to heat are especially needed for use in certain areas of automotive engineering (engine compartment) and automation technology. In the passenger compartment of a vehicle a maximum of +85°C will arise. PMMA-POF can easily be used with such temperatures. In the area near the center console or under the roof temperatures can also go up to over +100°C and near the engine to +125°C. Summaries of the data published so far and of comprehensive investi-gations at the POF-AC Nürnberg can also be found in [Poi03a] and [Poi03b]. On the whole the following methods for increasing the resistance to heat of polymer fibers have been presented:

Cross-linking of PMMA: cross-linking between polymer chains can be generated by chemical effects or by UV irradiation which results in a rise of Tg. At the same time, however, the scattering and the mechanical charac-teristics become worse. Polycarbonate: PC has a considerably greater Tg compared with PMMA and is likewise transparent. Fibers made of this material have been produced on a large scale. PC fibers, however, age relatively quickly in combination with humidity. Elastomers: fibers made of this material could be used up to +170°C and show very low attenuation. So far, they have only been produced as labora-tory samples. Alternative polymers: a series of other polymers such as cyclical polyolefins have Tg up to +200°C.

When determining the thermal stability, a maximum increase in the kilometric attenuation is established over a maximum period of aging. In case the aging pro-cedures are thermally activated, then the permissible operating period decreases almost logarithmically to the temperature. An example of the behavior of a stan-


dard PMMA-POF can be seen in Fig. 2.164 (measurements were made at the POF-AC). The increase in losses is represented here vs. the temperature. With an approx. 10 K increase in temperature the speed of ageing increases about one order of magnitude.

1

10

100

1000

70 75 80 85 90 95

520 nm

590 nm

650 nm

increase of attenuation coefficient

dB/(km 1000 h)

temperature [°C]

Fig. 2.164: PMMA-POF ageing

2.7.2.1 Cross-Linked PMMA

One of the most obvious methods for more heat-resistant POFs is the use of cross-linked PMMA, generally referred to as modified PMMA. The attenuation curves of such fibers are summarized in Fig. 2.165. The fibers of the PHK Series are sold by Toray ([Tor96a] and [LC00a]). Important parameters are:

Core/cladding: PMMA/fluoropolymer Diameter. 0.5 mm, 0.75 mm. 1.0 mm and 1.5 mm NA/aperture angle: 0.54/65° Lowest attenuation at 650 nm (for 1 mm): <300 dB/km Permissible bending radius: 9 mm Operating temperature: -40°C to +115°C Available as single fiber and bundle with 18 fibers à 0.5 mm

The attenuation measurement at the POF-AC is shown in the figure. A first version of the Toray fiber has already been presented [Tan94a]. Another fiber manufacturer active in this field is Asahi Chemical. A first sample was introduced in 2003 under the designation H-POF and the measurement results were presented in [Poi03a]. The manufacturer provides the following data:

Core/cladding: cross-linked PMMA/P-FEP Primary coating: ETFE (black Tefzel®)Numerical aperture (after 2 m): 0.65 Core diameter: 1.0 mm Primary coating: 1.51 mm/2.3 mm (MOST specification)


Attenuation: 540 dB/km (measured at 657 nm) Minimum bending radius: 5 mm Bandwidth: 30 MHz · 100 m Operating temperature: -40°C to +130°C

Samples of PMMA-POF with different degrees of cross-linking were produced in 2002 to 2004 by the RPC Institute in Tver near Moscow. The measurement results of a sample are shown in the picture. The maximum application tempe-rature lies at +130°C with an attenuation of about 800 dB/km at 650 nm.

attenuation [dB/km]

400 500 600 700 800100

1.000

200

2.000

500

4.000

wavelength [nm]

H-POF[Tan94a]PHKS

Tver-POF1

Fig. 2.165: Attenuation of cross-linked PMMA-POF

On the whole it is true for this type of fiber that a higher degree of cross-linking leads to higher application temperatures, whereby the scattering is also greater so that the losses increase. A short piece of a Tver fiber sample exposed to red light is shown in Fig. 2.166. The high degree of scattering leads to a clearly visible lateral emission.

Fig. 2.166: Cross-linked PMMA-POF (sample from Tver)


2.7.2.2 Polycarbonate POF

The first polymer fibers on the basis of polycarbonate were introduced in 1986 by Fujitsu ([Ish92b] and [Koi95]). The attenuation lay at 800 dB/km at 660 nm and 450 dB/km at 770 nm respectively. The maximum operating temperature was given at +130°C. Similar data were published in [Min94] - see also Fig. 2.167.

In 1992 [Tesh92], Asahi introduced another PC-POF called Luminous H. With an application temperature of +125°C the attenuation was 600 dB/km at 660 nm. The fiber NA was 0.78 and the bandwidth 17 MHz · 100 m. The relatively large NA of most PC-POFs can be explained by the high refractive index of PC which amounts to about 1.59. If PMMA is used with n = 1.49 as cladding material, the result is then AN = 0.55.

Mitsubishi sells a type of fiber, the ESKA FH4001-TM, with a temperature capability up to +125°C. The specific parameters for this type are:

Application temperature range: -55°C to +125°C Application temperature at high humidity: +85°C Maximum attenuation at 770 nm: 800 dB/km Minimum bending radius: 25 mm Core/cladding material: polycarbonate/fluoropolymer Refractive index core/cladding: 1.582/1.392 Numerical aperture: 0.75 ± 0.01 Core/ cladding diameter: 910 ± 50 μm / 1000 ± 60 μm Primary coating: 2.2 mm polyolefin elastomer

Laser Components GmbH offered another PC fiber. The last PC-POF shown in Fig. 2.167 was introduced by Furukawa ([Hatt98], [Nish98] and [Irie94]), whereby a material was used in which hydrogen atoms were partially replaced by fluorine.

400 500 600 700 800 900

1,000

10,000

2,000

500

5,000Minami 1994Laser Comp.Mitsubishi

Furukawa

wavelength [nm]

attenuation [dB/km]

Fig. 2.167: Various PC-POF


The fibers produced by Furukawa with a core diameter of 0.5 mm had a NA of 0.35 and 0.53 (elastomer as cladding material). The low-NA version attains a bandwidth of 200 MHz 100 m. A data rate of 156 Mbit/s could be transmitted over 80 m of fiber (200 Mbit/s over 70 m).

No change in length could be ascertained in ageing tests over 10 days at tempe-ratures of +100°C to +155°C (Fig. 2.168) which corresponds to an improvement by 20 K over conventional PC-POF.

10

8

6

4

2

0105 115 125 135 145 155

length variation [%]

temperature [°C]

PC-A

PC(AF)

Fig. 2.168: Temperature resistance of PC ([Hatt98])

The different PC-POF from Furukawa are summarized once again in Fig. 2.169. Unfortunately, we do not know of any other work carried out by this company.

PC(AF) [Hatt98]

D-POF[Irie94]

PC-POF [Irie94]

400 500 600 700 800 900300

1,000

2,000

500

5,000attenuation [dB/km]

wavelength [nm]

Fig. 2.169: Data by Furukawa 1994-1998 (Polycarbonate)

The greatest disadvantage of PC-POF is its poor stability in regard to humid heat. BAM tests on the aging of different POFs are summarized in Fig. 2.170. What is surprising here is that PC-POF broke down before standard PMMA-POF.


120%

100%

80%

60%

40%

20%

0%0 500 1,000 1,500 2,000 2,500 3,000 3,500

at 92°C / 95 % RH wavelength: 650 nm; sample length: 10 m

relative transmission

SI-PMMA

SI-mod. PMMA

SI-PC

ageing time [h]

Fig. 2.170: Ageing behavior of various POF ([Daum03c])

2.7.2.3 Elastomer POF

Possibly the most suitable material group for heat-resistant POF are elastomers. A number of institutes have already produced samples, but real product development is still missing.

450 500 550 600 650 700 750 800 850 900400

1,000

10,000

2,000

5,000

attenuation [dB/km]

HPOF-S

HPOF-Sb

[Ish92]

[Suk94]

[Zei03]

wavelength [nm]

Fig. 2.171: Attenuation of various EOF

The attenuation curves of different EOFs (elastomer optical fiber) are compared in Fig. 2.171. Particulars of the following fibers are compared:


Elastomer POF, produced by G. Zeidler (see [Zei03]) HPOF-S (Hitachi), data sheet information HPOF-Sb (Hitachi), POF-AC measurements (1.5 mm core diameter) 2 mm silicone elastomer POF, AN = 0.54 [Ish92b] POF made of ARTONTM, cyclical olefin, [Suk94]

You can clearly see that the attenuation spectra are quite similar to those of PC fibers. The lowest losses lie in the range around 500 dB/km which is entirely acceptable for use in vehicle networks or for parallel connections.

Typical representative fibers for both EOF and PC POF are compared in Fig. 2.172. The similar functional groups lead to only slight differences in the loss spectra.

attenuation [dB/km]

500 550 600 650 700 750 800 850500

1000

2000

3000

wavelength [nm]

PC-POF

silicone-POF

Fig. 2.172: PC-POF in comparison with silicone-POF

The most recent development by Asahi is particularly interesting. The HPOF-S possesses the following specified parameters (see [Poi03b]).

Core/cladding material: elastomer/P-FEP Primary coating: ETFE (black Tefzel®) Numerical aperture (after 2 m): 0.65 Core/cladding diameter: 1.00 mm / 1.50 mm Primary coating: 2.3 mm Attenuation: 800 dB/km (measured at 660 nm) Minimum bending radius: 7 mm Bandwidth: 250 MHz · 100 m Operating temperature: -40°C to +150°C

Since the cladding could not be extruded directly, it was produced as a tube. At 250 μm it is relatively thick. Practical technical production methods for such fibers are undoubtedly possible.

When aged under high temperatures, the attenuation of this fiber even dropped to values around 300 dB/km. Whether this was due to the possible drying of the fiber or by improving the adhesion of the cladding on the core could not be determined. The material system thus shows some enormous potential.


2.7.2.4 Cyclic Polyolefines

A theoretically useful group of materials for POF are also the polyolefins. Figure 2.173 shows a possible structure. These materials can also be produced transparent. Low losses are theoretically possible because of their amorphous structures.

C

RH

H

C

H

x

C

H

C

H

R’ R’ y

Fig. 2.173: Molecule structure COC

Some principal characteristics of such materials are:

Low water absorption Theoretically more transparent than PMMA Refractive index n = 1.56, makes another range of NA possible as well as the production of different index profiles Tg typically > 150°C

Manufacturers of such polymers are among others Ticona and JSR. It is not foreseeable when test fibers made from this very promising material system will be produced again.

2.7.2.5 Comparison of High-Temperature POF

So far the following temperature-resistant fibers have been summarily described:

Cross-linked PMMA (>130°C) Polycarbonate (115°C) Partially fluorinated polycarbonate (145°C) Silicone elastomers (>150°C) Thermoplastic resins (145°C) ARTONTM (Fujitsu) (170°C)

The data of these different fibers have been compiled in Tables 2.19 to 2.21. *) temporary data sheet, the fiber is not presently available **) modified PC - partially fluorinated according to the authors’ information ***) different data on materials, but with identical attenuation curves


Table 2.19: Polycarbonate-POF

Parameter MitsubishiFH 4001-

TM

ProducerB

*)Furukawa

[Irie94][Hatt98]

Furukawa**)

[Hatt98][Nish98]

Laser Comp.

core diameter 910 50 μm 940 20 μm 910 μm 500 μm 1 mm cladding thickness 40-50 μm 30 μm n. a. n. a. n. a. NA 0.75 0.54 n. a. 0.30 0.61 x dB/km @ y nm 800@770 2000@633

1500@780 400@660 700@760

460@650 300@780

800@770

bandwidth n. a. n. a. n. a. 200MHz 100m n. a. max. temperature +125°C +125°C +125°C +145°C n. a. core material n = 1.582 n = 1.586 PC(A) D-POF, PC-AF***)

PCcladding material n = 1.392 n = 1.491 n. a. n. a. n. a.

Table 2.20: Properties of modified PMMA-POF

Parameter Toray PHKS-

CD1001-22

HitachiH-POF

Tver-POF (Sample

2002)

Toray [Tan94a]

principle mod. PMMA cross linked PMMA Copolymer Copolymer core diameter n. a. 1 mm 1 mm 1 mm cladding thickness n. a. 250 μm 30 μm n. a. NA 0.54 0.65 >0.50 n. a. x dB/km @ y nm 300@650 540@660 800@660 250@650 bandwidth n. a. 30 MHz 100m n. a. n. a. max. temperature +115°C +130°C +130°C Tg = 135°C core material PMMA PMMA PMMA Copolymer cladding material n. a. P-FEP n. a. n. a. jacket PP ETFE n. a. n. a.

Table 2.21: Properties of different high temperature-POF

Parameter Hitachi HPOF-S

Hitachi[Sas88]

Bridgestone [Ish92b]

Zeidler[Zei03]

Fujitsu[Suk94]

principle silicone resin silicone elastomer elastomer core diameter 1.0 mm 1 mm n. a. 1 mm 1 mm cladding thickness 0.25 mm 0.5 mm n. a. n. a. n. a. NA 0.65 0.62 0.54 0.44/0.25 n. a. x dB/km @ y nm 800@660 660@650

900@780 700@660 450@770

800@770 800@680

bandwidth 25 MHz km n. a. n. a. n. a. n. a. max. temp. n. a. >150°C +150°C +150°C Tg = 171°C

core material n. a. ester based thermosetting resin

silicone elastomer ARTON

cladding material

P-FEP ethylen tetrafluoride - propylene hexafluo-

ride copolymer

n. a. elastomer fluorcopol.

n. a.

jacket Tefzel (ETFE) n. a. n. a. without n. a.


Fig. 2.174 shows an aging experiment at +130°C with different fibers described above. The most suitable ones at these temperatures were evidently the EOF and the PC-POF.

0 100 200 300 400 500 600 700100

1,000

10,000

200

2,000

500

5,000

attenuation [dB/km]

measuring time [hours]

T = +130°C

FH 4001

PHKS

HPOF-S

TVER 2002

Fig. 2.174: Ageing of various POF at high temperatures

The PC-POF from Mitsubishi (FH4001) only shows a moderate increase while the two POFs made of cross-linked PMMA aged more quickly. The EOF even gets better during the measurement period. Particularly noticeable is the clear drop in attenuation after 15 hours. This was the point at which the temperature was raised in the climate test chamber. It was noticeable that the bandwidth of the EOF had dramatically diminished after this treatment. The combination of both events provides the explanation that the adhesion of the cladding onto the core was clearly improved by the high temperature so that even higher modes can now be guided.

2.7.3 Polystyrene-Polymer Fibers

Another candidate for the production of polymer optical fibers is polystyrene (PS), the molecular structure of which is shown in Fig. 2.175 ([Ram99]).

C

H

HC

C H

CC H

H

H

H

C

C

C

H

C

H

HC

C H

CC H

H

H

H

C

C

C

H

n

Fig. 2.175: Molecule structure of PS


Theoretically, the attenuation of PS is partly below that of PMMA, as the following theoretical estimate of losses in [Kai89] shows - without taking into account propagation effects and the effects of claddings (see table 2.22).

Table 2.22: Theoretical attenuation of different polymers according to [Kai89]

Material Wavelength Rayleigh-Scattering

UV-Ab-sorption

C-H-Absorption

Sum Total

520 nm 28 dB/km 0 dB/km 1 dB/km 29 dB/km


PMMA





PS




PMMA-d8


To date, PS-POF have been manufactured e.g. by Toray (first PS-POF 1972), NTT (1982) and CIS in Tver (1993). The initial fibers had an attenuation of over 1,000 dB/km; later on it was possible to reduce this to 114 dB/km at 670 nm ([Koi95]). The NA of these fibers which can be used at temperatures up to 70°C is 0.56, i.e. a little higher than that for the standard PMMA-POF. Figure 2.216 shows the attenuation behavior of a PS-POF ([Ram99], red curve and [Zub001b]).

500 550 600 650 700 750 800 850

attenuation [dB/km]

100

1000

800

400

600

200

wavelength [nm]

[Zub01b]

Fig. 2.176: Attenuation spectrum of PS-POF acc. to [Ram99] and [Zub01b]


The refractive index of PS is n = 1.59 so that it is possible to use PMMA for the optical cladding (n = 1.49), as is possible for PC (n = 1.58). The glass transition temperature of PS is approx. 100°C and therefore approx. 5 K lower than that of PMMA. Hitherto there has been no reason to replace the PMMA-POF by PS so that this material is not of any practical significance.

2.7.4 Deuterated Polymers

As has been illustrated in Fig. 2.163, a significant reduction in the absorption losses of polymers can only be achieved by substituting the hydrogen with heavy atoms. This would seem to be achieved most simply by replacing it with deu-terium. This isotope has twice the atomic mass compared to hydrogen. In nature, approximately 0.0156% of all hydrogen atoms are deuterium (1 atom in every 6,400). Chemically, deuterium behaves the same way as hydrogen so that it simply makes sense to use so-called heavy water (D2O) as a base material for this synthesis. Table 2.23 shows data of different POF based on deuterated polymers.

Table 2.23: Data of deuterated materials

Ref. Year Producer AttenuationdB/km

at:nm

Remarks

[Koi95] 1977 Du Pont 180 790 first deuterated SI-POF

[Koi96c] 1982 NTT 20 680 SI-POF

[Lev93] 1993 CIS 120180

650850

core: 200-1000 μm, AN = 0.48, to 70°C

[Koi92][Khoe94]

1993 Keio Univ. 5694

688780

core: 500 μm, MMA-BBP-d8, 2.000 MHz km

[Kon02] 2002 Keio Univ. 58109127

650780580

g = 3.4; 511 MHz 300 m Tg = 105°C

[Kon03] 2003 Keio Univ. 58 650 g = 2.0; 1020 MHz 250 m

[Kon04] 2004 Keio Univ. 80 650 g = 2.3; 1200 MHz 300 m

According to [Koi95], the first deuterated SI-POF was produced by DuPont in 1977. In 1982, NTT ([Koi96c]) produced a SI-POF in deuterated material with a minimum attenuation of 20 dB/km at 680 nm. It was not until the year 2000 that this attenuation value was improved with the introduction of LucinaTM-POF. Figure 2.177 shows further attenuation curves for POF made with deuterated poly-mers; all examples are GI fibers.

Using POF made with deuterated polymers would offer a number of advan-tages. Chemically these materials behave identically to the substances made from "normal" hydrogen. The attenuation is approximately one order of magnitude less than the values achieved for PMMA fibers. The behavior over temperature and the options for index profile design should be the same as those of PMMA-POF. However, the decisive disadvantage is that there is always water vapor present in the atmosphere which will be absorbed by the fibers. This will lead to a situation,


where which protons (normal hydrogen nuclei) slowly replace the deuterium so that the absorption losses will increase again.

Although it is possible to solve the problem with a watertight coating of the fiber (including all connections), this would defeat the object of obtaining a parti-cularly low priced cable system.

500 600 700 800 900 1000 1100 1200 1300 1400

[Koi95]

[Ish92a]

[Koi96b]

[Koi96d]

[Mur96]

10

100

1,000

10,000

20

50

200

500

2,000

5,000

wavelength [nm]

attenuation [dB/km]

Fig. 2.177: Loss spectra of GI-POF (deuterated, 1996)

In the past few years work has once again been conducted in Japan on the production of deuterated POF. GI fibers exclusively have been investigated - see [Kon02], [Kon03] and [Kon04]. The attenuation of these fibers from [Kon04] is compared in Fig. 2.178 with the values from 1995 and those of a PMMA-POF.

0

500

1000

1500

450 550 650 750 850wavelength [nm]

d8-PMMA

PMMA

attenuation [dB/km]

2002

1995

Fig. 2.178: Loss spectra of GI-POF ([Kon02])


Different production versions are compared in [Kon02]. The effect of an additional PMMA cladding is investigated among other things. The best results compared with a pure PMMA-POF are shown in Fig. 2.179. With about 60 dB/km and 650 nm the attenuation ranges approximately between pure PMMA and PF-GI fibers. On the other hand, the attenuation of the PMMA POF at 520 nm also does not lie much higher.

500 550 600 650 700 750 80010

100

1.000

10.000

20

50

200

500

2.000

5.000

wavelength [nm]

attenuation [dB/km]

PMMA PMMA

d8-POF

Fig. 2.179: Attenuation of deuterated POF ([Kon04])

Since 2003, Fujifilm has been announcing the development of a new fiber “Lumistar” in the versions I, V and X. According to their own statements this is: “the first POF with a large diameter which is able to transmit over 1 Gbit/s”. This is somewhat exaggerated, of course, since PMMA GI-POF and MC-POF have been able to do this for many years.

power [dB]

0.0 0.5 1.0 1.5 2.0 2.5 3.0-6.0

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

frequency [GHz]

1.9 GHz 100 m 3 dB-bandwidth

Fig. 2.180: Frequency response of the Lumistar GI-POF ([Nak05b])


Details of a fiber with a core diameter of 500 μm and a cladding diameter of 750 μm are described in [Nak05b]. The bandwidth of the fiber is 1.9 GHz over 100 m. Figure 2.180 shows the frequency response.

Furthermore, the work shows that the index profile produced by gel poly-merization technology also remains stable after 2000 hours of aging at +90°C (Fig. 2.181) which is very astonishing.

Fig. 2.181: Refractive index profile of the Lumistar GI-POF after ageing (+90°C)

Parameters for the Lumistar fibers are mentioned in different sources. Accor-ding to this information a particularly low-attenuation polymer is used. Since the company works closely with Keio University, where until 2004 there were reports on the development of deuterated fibers with very similar parameters, we must assume that we are dealing here with d8 PMMA-POF.

Table 2.24: Data of the d8-POF Lumistar

Lumistar-I Lumistar-V Lumistar-X

core material n. a. n. a. n. a.

core diameter 500 μm 300 μm 120 μm

cladding diameter 750 μm 316 μm 500 μm

attenuation 160 dB/km (650 nm)

180 dB/km (650 nm)

<100 dB/km (850 nm)

bandwidth 1 GHz 50 m 3 GHz 50 m 10 GHz 50 m

According to [Kon05] the fiber is drawn from a 22 mm thick preform (60% core). By means of a two-stage polymerization process the bandwidth is improved (Fig. 2.182 shows the losses of two current versions). The optimal NA lies bet-ween 0.2 and 0.3. The goal is the transmission of at least 3 Gbit/s over 200 m.


500 550 600 650 700 750 80050

100

1000

200

500

attenuation [dB/km]

wavelength [nm]

PMMA-d8 core PMMA-cladding

PMMA-d8 complete

Fig. 2.182: d8-POF variants according to [Kon05] - conventional gel-polymerization with all-PMMA-d8 (79.8 dB/km at 650 nm) - two level gel-polymerization with PMMA-d8 core and PMMA-cladding (red)

The bandwidth of the fibers was determined through pulse broadening in the time domain, a SI-POF was used as a mode mixer. The fiber with the PMMA d8 core and the PMMA cladding attains 1.2 GHz · 300 m (overfilled launch). This version does indeed have a somewhat higher attenuation, but also has a higher bandwidth due to the index dip at the core-cladding interface.

In October 2004, Fujifilm introduced a DVI transmission system on the basis of the Lumistar fiber. Using a 850 nm VCSEL a data rate of 10.3 Gbit/s over 40 m could be transmitted (eye diagram in Fig. 2.183).

Fig. 2.183: 10.3 Gbit/s-data transmission over 40 m PMMA-d8-GI-POF

To what extent this fiber is actually available on the market cannot yet be assessed since there are no channels of distribution yet in Europe. Even the actual production costs are still unknown.

The use of fluorine instead of deuterium is indeed more complicated, but does promise even lower attenuation values and above all long-life fibers. The following section describes the development of these fibers.


2.7.5 Fluorinated Polymers

The atomic mass of fluorine is many times greater than that of hydrogen so that the absorption bands are moved significantly further into the infra-red zone. The theoretical minimum values are less than 0.2 dB/km ([Mur96]), i.e. comparable to silica fibers in the wavelength range of about 1,500 nm. Figure 2.184 compares the attenuation values theoretically possible for fluorinated polymers with those achieved for singlemode glass fibers.

0.1

1

10

100

400 600 800 1,000 1,200 1,400 1,600wavelength [nm]

attenuation [dB/km]

silica glassPF-polymer

Fig. 2.184: Theoretical comparison of PF polymer and silica

However, practical experience shows that these impressive theoretical values are in fact difficult to achieve. The most important question is whether it will be possible to find a fluorinated polymer that can be processed into a fiber in its amorphous state. For example, Teflon materials tend to crystallize. Due to scatte-ring losses, this will significantly reduce the transparency of the material. Even this very first problem proved to be quite difficult to solve. The second question relates to the production of the optical waveguide itself. For a step index fiber one needs a cable material with a slightly smaller refractive index ( n 0.02 - 0.05). However, fluorinated polymers already have the lowest refractive index of all existing transparent plastics (n = 1.340 at 650 nm, or n = 1.336 at 1,300 nm), which is why they are the preferred material for claddings. The reason why no PF-SI-POF have been produced to date is simply the fact that there are no suitable cladding materials available for this purpose.

In principle, graded index POF do not require an optical cladding. On the other hand, it is necessary to find a way to continually increase the refractive index towards the axis. Essentially this can be achieved through doping and co-poly-merization. In the case of silica glass, the index variation can be easily achieved by replacing the silicon atoms with germanium because these two substances be-have identically within the glass structure. However, the components used for po-lymer optical fibers do not allow such a simple replacement of individual atoms.


The process of doping involves inserting small molecules between the long chains of the actual core material which increases the refractive index. What is important is that the dopants do not diffuse out of the polymer material too easily and do not show too strong absorption in the desired wavelength range. The doping process always lowers the glass transition temperature. It is therefore desirable to insert a molecule that accomplishes the required change in the refrac-tive index even at small concentrations (a few percent).

In co-polymerization one uses chains composed of different monomers. The ratio of monomers determines the refractive index. In this case it is important that the sequence should be irregular - no long chains of one monomer are formed - since otherwise the losses due to scattering increase considerably. This means that the bonding force of monomers amongst each other must not be greater than the bonding force to the respective other monomer. Of course, both monomers must have sufficient transparency. Figures 2.185 and 2.186 show a schematic illustra-tion of the principles.

monomer

dopant

Fig. 2.185: Index variation by dopants

monomer A

monomer B

Fig. 2.186: Index variation by copolymerization


Some fluorine polymers are listed, for example by [Mur96].

HFIP 2-FA hexafluoroisopropyl 2-fluoroacrylate PTFE polytetrafluoroethylene FEP tetrafluoroethylene-hexafluoropropylene PFA tetrafluoroethylene-perfluoroalkylvinyl-ether

To date the best results in producing low attenuation POF have been achieved with the material CYTOP (cyclic transparent optical polymer), developed at Asahi Glass in Japan. This material no longer contains hydrogen. Its molecular structure is shown in Figs. 2.187 and 2.188.

CYTOP®

CF2=CF-O-CF2-CF2-CF=CF2

CF

O

CF2

CF

CF2

CF2

CF2

CF

O

CF2

CF CF2

CF2

CF2

CF

O

CF2

CFCF2

CF2

CF2

momomer polymer

Fig. 2.187: Fluoropolymer CYTOP® from Asahi Glass

Fig. 2.188: CYTOP® molecule structure

It was possible to reduce the attenuation of fibers step by step from initially over 50 dB/km to 30 dB/km and finally to less than 10 dB/km at a wavelength of 1,300 nm, as shown in the data for different PF-GI-POF in Table 2.25.

Different attenuation spectra of GI-POF are compared in Fig. 2.189. The years indicate the history of the development of this technology. Estimates in [Mur96] suggest that attenuation for CYTOP will be less than 1 dB/km, bearing in mind that the need for a GI profile will have a negative effect on this value.


Table 2.25: Data of different PF-GI-POF

Ref. Year Producer Øcore

μmdB/km at nm Remarks

[Koi96c] 1995 Keio Univ. n. a. 50 1300 [Mur96] 1996 Asahi

Glass Co. 300-500

14056

8501300

n=1.34 (589 nm, Tg=108°C, n = 0.115, = 2.4

[Koi96c] 1996 Keio Univ. n. a. n. a. n. a. 10 GHz 100 m | 660 nm [Yos97] 1997 Asahi

Glass Co. 125-300

56 1300 AN = 0.2, = 2.4, nKern = 1.34, 600 MHz km

[Koi98] 1998 n. a. 40 1300 [Oni98] 1998 Asahi

Glass Co. 210 41

45850

1300 10,000 h/70°C, AN = 0.18

[Khoe99] 1998 n. a. 120 56

8501300

[Khoe99] 1998 130 110 43.6 31

650840

1310 [Koi00] [Kog00]

2000 Asahi Glass Co.

120 15 1300 9 ps/nm km dispersion 509 MHz km@1300 nm 522 MHz km@850 nm

[Wat03] 2003 Asahi Glass Co

120 15 8

1300 1070

till now lowest POF-attenuation

[Gou04] 2003 Nexans 120 40 850 1500 MHz 100 m [Whi04b] 2004 Chromis 120 25 850 400 MHz km[DuT07] 2007 Chromis 120

5040 800-

1300 800 MHz kmcontinuously drawn

500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600

wavelength [nm]

attenuation [dB/km]

10

100

1,000

200

500

20

50

1996

1995

1998

2000

Fig. 2.189: Development of the attenuation for PF-GI-POF


Values below 20 dB/km allow transmission ranges of up to 1,000 m. This covers not only the field of application for copper data cables but also for glass multimode fibers. Likewise, deployment in access networks would become possible.

The best values so far are shown in Fig. 2.190 from [Whi02] and [Wat03], whereby OFS - in the meantime under the company name of Chromis Fiberoptics - has used a continuous production process for the first time (more information below).

600 700 800 900 1,000 1,100 1,200 1,300 1,400

AGC

OFS

10

100

60

20

40

6

wavelength [nm]

attenuation [dB/km]

Fig. 2.190: Current attenuation values for PF-GI-POF

2.7.6 Overview over Polymers for POF Jackets

Apart from the materials used in the fiber core, the material used for the jacket is also important. It has a significant contributing effect on thermal resistance. In addition, the jacket determines the mechanical properties of the cable such as re-sistance to compressive load and tensile strength as well as flexibility. Tables 2.26 through 2.30 list different possible materials with some of their characteristics.

The use of PVC, PE or PA as typical jacket materials for applications within buildings allows for maximum temperatures ranging from 70°C up to 90°C. The materials in the last two rows (trade names are Teflon FEP or Teflon PTFE) can be used at significantly higher temperatures.


Table 2.26: Materials for POF jackets (thermal properties)

ShortName

Material VDE-label

Allowed Conti-nuous Operation Temperature

Thermal Over-load Capacity240 h 20 h

PVC polyvinylchloride Y 70°C 80°C 100°C PVC 90° polyvinylchloride 90°C Y 90°C 100°C 120°C PVC flame ret. polyvinylchloride flame retardant Y 70°C 80°C 100°C PE LD; MD polyethylene (low, medium density) 2Y 70°C 100°C 100°C PE flame ret. polyethylene flame retardant /with halogen 2Y 70°C 100°C 100°C PE HD polyethylene (high density) 2Y 80°C 110°C 120°C PP polypropylene 9Y 90°C 110°C 130°C PA-6 polyamide - 6 4Y 80-90°C 120°C 150°C PUR polyurethane (thermoplastic) 11Y 90-100°C 120°C 140°C VPE cross linked polyethylene 2X 90°C 140°C 160°C EVA ethylene-vinylacetate-copolymere 4G 120°C 160°C 180°C FEP perfluorethylenpropylene 6Y 180°C 230°C 240°C PTFE polytetrafluorethylene 5Y 260°C 300°C 310°C

Table 2.27: Materials for POF jackets (thermal/mechanical properties)

ShortName

Proces-sing *)

Flame Resistant

OxygenIndex LOI

Thermal Value Ho

MJ·kg-1

Thermal Conductivity

W·K-1·m-1

LinearExpansion-Coefficient K-1

PVC E partly 23-28% O2 17 - 25 0.17 10 - 20·10-5

PVC 90° E partly 23-28% O2 17 - 25 0.17 10 - 20·10-5

PVC flame red. E yes 30-40% O2 15 - 20 0.17 10 - 20·10-5

PE LD; MD E and S no 22 % O2 42 - 44 0.30 20 - 50·10-5

PE flame red. E and S partly 24-27% O2 35 - 40 0.30 20 - 50·10-5

PE HD E and S no 22 % O2 42 - 44 0.40 40 - 45·10-5

PP E and S no 22 % O2 42 - 44 0.19 15·10-5

PA-6 E and S no 22 % O2 29 - 30 0.23 7 - 10·10-5

PUR E and S no 20-25% O2 23 - 27 0.25 15 - 20·10-5

VPE E V no 22 % O2 42 - 44 0.30 20 - 30·10-5

EVA E V no 22 % O2 19 - 23 n. a. n. a. FEP E yes >95 % O2 5 0.26 8 - 11·10-5

PTFE W (E) yes >95 % O2 5 0.26 6 - 15·10-5

*) E: extrusion, S: injection molding, V: vulcanization, W: wrap technology

Table 2.28: Materials for POF jackets (physical/chemical properties)

Short Name Melting Temperature

Low Tempe-rature Limit

Densityg·m-3

Corrosive Harmfull Agents in the Flue Gas

-RaysResistance

PVC from 130°C -10°C 1.20-1.50 yes 10 Mrad PVC 90° from 130°C -10°C 1.20-1.50 yes 10 Mrad PVC flame ret. from 130°C -10°C 1.30-1.60 yes 10 Mrad PE LD; MD 90-110°C -50°C 0.87 no 100 Mrad


PE flame ret. from 110°C -50°C 0.98 yes 50 Mrad PE HD 125-135°C -50°C 0.95-0.98 no 100 Mrad PP from 145°C -20°C 0.91 no 10 Mrad PA-6 from 175°C -50°C 1.10-1.15 ? 10 Mrad PUR from 150°C -50°C 1.15-1.20 no 500 Mrad VPE - -50°C 0.92 no 100 Mrad EVA - -50°C 1.30-1.50 no 100 Mrad FEP 255-275°C -65°C 2.00-2.30 yes 0.1 Mrad PTFE 325-330°C -65°C 2.00-2.30 yes 0.1 Mrad

Table 2.29: Materials for POF jackets (physical/chemical properties)

ShortName

Oil and fuel resistance

WeatherResistance

Shore-Hardness1) = A; 2) = D

TensileStrength

ExtensionBreak

PVC middling good 70-951) 10-20 N·mm-2 150-350 % PVC 90° middling good 70-951) 10-20 N·mm-2 150-350 % PVC flame ret. middling good 80-901) 10-20 N·mm-2 150-250 % PE LD; MD bad medium 43-502) 15-20 N·mm-2 300 % PE flame ret. bad medium 502) 15-20 N·mm-2 300 % PE HD middling medium 60-622) 15-25 N·mm-2 300 % PP middling medium 40-602) 30-50 N·mm-2 300 % PA-6 middling good 40-75 70-120 N·mm-2 50-200 % PUR good excellent 75-1001) 35-45 N·mm-2 300 % VPE middl. /good good 40-502) 12-20 N·mm-2 300 % EVA bad good 70-901) 5-15 N·mm-2 300 % FEP very good excellent 55-602) 15-25 N·mm-2 250 % PTFE very good excellent 55-652) 80 N·mm-2 50 %

Table 2.30: Materials for POF jackets (electrical properties)

ShortName

Loss Factor tan at 20°C and 800 Hz

Permittivity at 20°C and 800 Hz

Resistivityat 20°C

PVC 20 - 100·10-3 4 - 6 1013 ·cmPVC 90° 50 - 100·10-3 4 - 6 1013 ·cmPVC flame ret. 70 - 150·10-3 5 - 7 1013 ·cmPE LD; MD 0.2; 0.4·10-3 2.3 1016 ·cmPE flame ret. 1.1·10-3 3 1016 ·cmPE HD 0.3·10-3 2.3 1016 ·cmPP 0.5·10-3 2.3 - 2.5 1016 ·cmPA-6 30 - 50·10-3 3 - 7 1014 ·cmPUR 30·10-3 8 1012 ·cmVPE 0.5·10-3 2.4 - 3.8 1016 ·cmEVA 20 - 30·10-3 4 - 6 1012 ·cmFEP 0.0003·10-3 2.1 1016 ·cmPTFE 0.0003·10-3 2.1 1017 ·cm

180 2.8 Fiber and Cable Production

2.8 Fiber and Cable Production

The processes for producing POF have been continuously improved in the last few years. The fundamental methods have indeed always remained the same, but vari-ous details have been improved. A very comprehensive treatment of POF produc-tion and its history can be found in [Nal04]. Many fine points concerning the materials can also be found in [Har99].

As opposed to the production of glass fibers there is a number of unusual features with POF. First of all, the polymer chemistry involved, in part very com-plicated and with its occasional safety aspects, has to be mastered. On the other hand the process temperatures are very much lower - almost always below +200°C. The demands on POF production can be sub-divided into four areas:

The core material must be produced uniformly without any impurities, air bubbles, etc. and with a correct distribution of the molecular masses. The fiber must be drawn or extruded exactly. For SI fibers a suitable cladding material with low refractive index and an attenuation not too high must be found and applied. In doing so, one must guarantee that the interface is sufficiently smooth and that the cladding has a good wringing fit. For graded index fibers a copolymer or a dopant must be found in order to be able to vary - usually increase - the refractive index. A suitable process is needed in order to distribute this material over the core cross-section so that you have a parabolic refractive index profile.

There are other steps, of course, such as the application of additional protective layers, the production of duplex or ribbon cables and quality control.

2.8.1 Production Processes for POF

Today glass fibers are produced in two different ways. The typically 125 μm thin fibers for telecommunication applications are produced - up to more than 1000 km - from a preform. Light guiding fibers are drawn directly from molten glass.

Even with polymer fibers one differentiates between continuous methods, spinning or extruding, and the drawing out of the preform.

In the preform method a cylinder is produced that already has the index profile of core and cladding while having a much larger diameter. During the drawing process, the diameter is reduced until the desired size has been reached (Fig. 2.191, see e.g. [Wei98]).

Ideally, the index profile should be maintained during this process but at a proportionally reduced scale. The length of the fiber per preform is determined as follows:

Length of fiber = preform length · (preform diameter/fiber diameter)2

2.8 Fiber and Cable Production 181

This method is applied generally for glass fibers. Automated processes are then applied to make several 100 km of fiber out of each preform, as the following example shows:

Length of glass fiber = 2 m preform (5 cm preform diameter/125μm)2 = 320 km

It is easy to see that the large core diameter of common POF is not favorable for this process since only a few km of fiber can be produced from each preform, for example:

Length of POF = 1 m preform · (5 cm preform diameter/1 mm)2 = 2.5 km

Drawing speeds for glass fibers today can attain 10 m/s; with POF about 0.2 to 0.5 m/s.

mounting with feed mechanism

preform

oven

take up drum

diameter control unit

Fig. 2.191: Production of POF from a preform

In addition to being able to draw the complete fiber out of the preform there is also the possibility of producing the core as a polymer cylinder and then applying the cladding by extrusion or enameling. The advantage here is that the polymeri-zation of the core material can proceed under very much better controlled conditions.

This process is used with PCS. A silica glass core is drawn out to 200 μm - or to other thicknesses as well - and is then surrounded by a polymer cladding, typically 15 μm thick. Understandably, the glass and the polymer have to be pro-cessed using different procedures.

Other versions are discontinuous production in which polymerization first takes place in the reactor and then the resulting block is extruded at low temperature, a so-called batch extrusion.


monomer initiator polymerization controller

N2

vacuum-pump

reactor

mixer

heater cooler

cladding polymer

POF with cladding

Fig. 2.192: Batch-extrusion according to [Hess04]

The monomer, the initior and the polymerization controller are first distilled by a vacuum pump. After the polymerization is finished, nitrogen pushes the poly-mers through the nozzle and the cladding is then immediately applied.

In addition, Mitsubishi has developed a method with which the polymerization, described in [Nal04], can take place photochemically.

Figure 2.193 from [Hess04] shows such a method. The core and cladding materials are pushed through a nozzle by a pump and a mixer. The cross-linking then takes place with a UV lamp. This process could prove to be quite suitable, especially for heat-resistant POF.

claddingmaterial

corematerial

mixture

take up drum

spinning

UV-light for crosslinking

nozzle

Fig. 2.193: Polymer crosslinking


When extrusion techniques are applied, the POF is produced in a continuous process directly from monomers. For SI-POF this process is very simple. Figure 2.194 shows such an arrangement (e.g. [Ram99], [Wei98]).

POLYMER filler

conveyor pump

heated vessel

core extruder fiber

cladding extruder

diameter control

Fig. 2.194: Production of SI-POF through extrusion

Such a system is also described in [Hac01]. According to the author the cladding materials used are Poly(3FMA) with n = 1.40 and PVF with n = 1.42. The polymerization takes place at about 150°C. With the drop in pressure when leaving the reactor the remaining monomer is vaporized and can be returned. The cladding is extruded at about +200°C. This temperature lies far above the glass transition temperature for PMMA. Thus is a critical step in the process in which the quick cooling of the fiber must be guaranteed. On the other hand, the cladding is only about 10 μm thick so that the thermal load is limited.

reactor

heating

monomer, initiator, polymerization controller

pump

cladding extruder

extruder

fiber

take up drum

Fig. 2.195: Extrusion of a POF according to [Hac01]


This process is also discussed in [Hess04]. The monomer is polymerized to about 80% in the reactor. The advantage of this standard process for SI-POF lies in the very slight contamination of the polymers caused by the process.

A modification of the process is presented in [Poi06d]. The new components in the process are:

The core material is PMMA granulate which is crushed before extrusion and effectively cleaned. The extrusion head is to be kept free of metal and any impurities whatsoever if possible. The turbo pump used makes a particularly even transport possible.

In addition, two further processes are mentioned in [Wei98]. In the thrust extru-sion technique, polymerization is carried out in a closed heated container from which the fiber is subsequently expelled through a nozzle at high pressure. The cladding is applied directly within the nozzle. This is a non-continuous process just like the preform technique.

In the spin-melt process, a volume of ready-to-use polymer pellets is melted and pressed through a spin head that incorporates many holes. The holes serve to form the core and apply the cladding. This process is very efficient but also very expensive.

2.8.2 Production of Graded Index Profiles

In order to guarantee the optimal functioning of graded index and multi-step index fibers, the best index profile possible should be realized. The developmental goal of the past few years has been to attain as much as possible with minimum effort and to continuously produce GI fibers.

A number of different processes for the manufacture of graded profiles are described in the technical literature:

Interfacial gel polymerization technique Centrifuging Photo-chemical reactions Extrusion of many layers

In most of these techniques the principle is to initially create a preform of up to 50 mm diameter and then to subsequently draw this preform down to the desired fiber size. Some of these methods are described below.

2.8.2.1 Interfacial Gel Polymerization Technique

This method was developed by Prof. Koike of the Keio University (for an example see [Koi92]). In this process a tube is initially manufactured with PMMA. This tube is then filled with a mixture of two different monomers M1 (high refractive index and large molecules) and M2 (smaller refractive index and smaller mole-cules). Initially the inner wall of the PMMA tube is slightly liquefied in an oven


that has been typically heated to 80°C. This results in a layer of gel and accele-rates polymerization. The smaller molecule M1 can more easily diffuse into this layer of gel so that the concentration of M2 increases more and more towards the middle. The index profile is thus formed in accordance with the resulting concen-tration gradient. For manufacturing a PMMA-GI-POF, [Koi92] proposes that MMA (M1) be supplemented with monomers VB, VPAc, BzA, PhMA and BzMA. The material that was finally used is BzA because its reactivity is comparable with that of MMA. The 15 mm - 22 mm thick preform is then drawn at temperatures between 190°C and 280°C to produce fibers ranging from of 0.2 mm - 1.5 mm in diameter. Figure 2.196 illustrates the principle (see also [Ish95]).

PMMA tube filled with a MMA/BzA mix

80°Cmelting of the PMMA tubeformation of a gel layer

the gel layer moves to the center concentration of M2 increases from outer to the the center

Fig. 2.196: GI profile formation by gel polymerization technique

[Koi95] describes this method in more detail. The PMMA tube is produced by rotating a glass reactor at 3,000 min-1 at 70°C that is partially filled with MMA. The polymerization process for the core takes place at a speed of 50 min-1 and a temperature of 95°C and requires approximately 24 hours to complete. [Ish95] describes the production of a PMMA GI-POF with DPS as dopants. For traditional materials such as BB or BBP, one obtains fibers with a NA of 0.17 - 0.21, whereas with DPS a NA of 0.29 is possible. The greater NA improves the bending charac-teristics and makes the launching of light easier.


2.8.2.2 Creating the Index Profiles by Centrifuging

Several publications ([Dui96], [Dui98] and [Chen00]) propose utilizing the den-sity difference of the different monomers to create the index profile through cen-trifugal force in a fast centrifugal process. [Chen00] compares the density and refractive index of different materials for this purpose (Table 2.31).

Table 2.31: Refractive index and density of different polymers ([Chen00])

Molecule Density n Molecule Density n

MMA 0.936 g/cm-3 1.490 BB 1.120 g/cm-3 1.568

DOP 0.981 g/cm-3 1.486 PMMA 1.190 g/cm-3 1.490

BIE 0.982 g/cm-3 1.564 TFPMA 1.254 g/cm-3 1.373

BzMA 1.040 g/cm-3 1.568 PTFPMA 1.496 g/cm-3 1.422

VB 1.070 g/cm-3 1.578 DBME 2.180 g/cm-3 1.538

The production of the preform is carried out in two steps. Once the monomer mixture has been filled into a tube, the GI profile is formed at room temperature. Then the temperature is increased so that polymerization takes place. Rotation continues during this process. Then the fiber is drawn from this preform.

In this process the rotation speeds must be up to 50,000 min-1. Even for a pre-form with 10 mm diameter the centrifugal acceleration (a = 2r) already equals 14,000 times the acceleration due to gravity. At the University of Eindhoven an ultra centrifuge operating at 50,000 min-1 has been constructed for preforms up to 50 mm in diameter which produce a centrifugal acceleration of 70,000 g. In the first trials, GI cylinders were produced from PTFPMA and MMA. The process for forming the GI profile took 24 hours. This was followed by a period of 12 hours during which the polymerization process was carried out at 60°C to 80°C. The refractive index difference achieved was approximately 0.009. No research reports have as yet been published on the production of fibers from such preforms.

2.8.2.3 Combined Diffusion and Rotation

The combination of diffusion and rotation for producing PMMA-GI preforms is described in [Park01]. The monomer is filled into a cylindrical glass reactor in the middle of which a rod made of a material with a high refractive index is located. This material diffuses slowly into the surrounding medium. Both parts can rotate at different speeds: the reactor at 500 to 1000 RPM and the rod at 6 to 60 RPM. The idea for different rotation speeds comes from determining the average of concentration fluctuations so that an ideal rotation-symmetrical profile comes about. After a few hours the preform is thermally polymerized. Figure 2.197 shows the principle and an index profile.

A fiber with a 1 mm core diameter was produced through thermal drawing from the preform described above. The bandwidth-length product amounts to 1.2 GHz · 100 m, measured with a 650 nm InGaAsP laser on a 50 m long fiber.


rotating reactor with monomer, initiator and polymerization controller

solid copolymer with higher refractive index

phase 1 room

tempe-rature

laminar mix of the

phases

n

copolymer is diffused into the monomer mix

phase 2 heated

poly-meri-zation

n n

final GI-preform

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.2 0.4 0.6 0.8 1.0

relative radius

concentration

after 5 hours

Fig. 2.197: Fabrication of GI-POF-preforms according [Park01]

2.8.2.4 Photochemical Generation of the Index Profile

According to [Nal04] the first GI-POFs were also produced by photo-copoly-merization, introduced in 1981 by Koike. A thin glass tube is filled with a mixture of MMA, vinylbenzoate (VB as dopant) and benzoyl peroxide (as initiator). The glass tube rotates during the UV irradiation. Since the UV radiation is higher at the edge a gel phase forms here through faster polymerization. The VB concentra-tion will be greater in the center since MMA has a faster reaction speed. The tube is irradiated from bottom to top and then polymerized out at high temperatures. This procedure did not result in any usable fibers.

[Miy99] proposes a method for the production of index profiles by means of a photo-chemical reaction. In this process, PMMA is doped with DMAPN ((4-N,N-dimethylaminophenyl)-N’-phenylnitrone). During exposure to ultra violet radia-tion (380 nm) the refractive index is reduced by up to 0.028, sufficient for GI-POF. In the experiment, thin films of a few micrometer thickness were used. Fibers have not yet been produced. It is likely that a problem would be the depth of penetration of the radiation which is significantly less than the intended fiber radius. Nevertheless, this process is of great interest since it works fast and makes continuous fiber production possible.

2.8.2.5 Extrusion of Many Layers

This multi-step index POF has hitherto been produced at two institutes (Research-Production Center, RPC Tver) and Mitsubishi Rayon.

The process corresponds to the production of SI-POF or DSI-POF except that several extruders must be combined with one another. Figure 2.198 shows the index profile of an MSI-POF according to [Lev99]. The curve drawn corresponds to that of an ideal parabola. In the core area the deviations of the real structure are relatively small.


0

10

20

30

40

50

60

0 100 200 300 400 500distance to the fiber axis [μm]

refractive index [a.u.]

Fig. 2.198: Index profile of a MSI-POF ([Lev99])

2.8.2.6 Production of Semi-GI-PCS

The production of the preform for semi-GI-PCS does not in effect differ from the manufacturing methods for normal glass fibers. The usual process is MCVD (mo-dified chemical vapor deposition). A mixture of SiCl4 and O2 are introduced into a heated quartz glass tube and SiO2 is formed by the chemical reaction. By adding chlorine, boron, germanium or phosphorus, you can continuously change the re-fractive index (Fig. 2.199). After cooling off, the tube with the inner layer will be collapsed, i.e. the hole disappears, and is drawn into a fiber. As opposed to classic glass fibers the PCS has an optical cladding made of polymers, not of glass, thus making a considerably greater refractive index jump possible.

gas mixing

GeCl4

controller

rail

burner

porous preform

ceramic or graphite rodSiCl4

O2

Fig. 2.199: Fabrication of glass fiber preforms (by OVD)


2.8.2.7 Polymerization in a Centrifuge

A new method for producing PMMA GI-POF ready for production has been deve-loped over the past few years by the South Korean company Optimedia under the direction of Prof. C. W. Park.

The production principle is based on copolymerization. As opposed to doping there is the advantage of the glass transition temperature not dropping as much. The polymer mixture is filled into a rotating tube and polymerized thermally or by UV irradiation. The polymer composition can be changed in steps or continuously. The rotation here does not serve the purpose of separating the materials, but only for achieving rotational symmetry. There are correspondingly fewer demands on the rotation speed. Figure 2.200 shows the set-up. A detailed description can be found in [Park06a].

Fig. 2.200: Rotating cylinder for GI-preform fabrication ([Park06a])

You can see quite well under a microscope that the fiber is built up of many layers. Nevertheless, the index profile is almost ideally parabolic and does not show any steps - see Fig. 2.201 acc. to [Park06a]. An attenuation spectrum of the OM-Giga, 1 mm GI-POF (data provided by the distributor Fiberfin) is shown in Fig. 2.202. At 650 nm the losses are below 200 dB/km.

0.0 0.1 0.2 0.3 0.4 0.5

1.490

1.495

1.500

1.505

1.510

1.515

1.520

1.525 refractive index

normalized radius [mm]

AN: 0.30

Fig. 2.201: Refractive index profile of a PMMA-GI-POF made by Optimedia ([Park06a])


400 500 600 700 800 900100

1000

5000

2000

200

500

attenuation [dB/km]

wavelength [nm]

Fig. 2.202: Attenuation spectrum of a PMMA GI-POF made by Optimedia (Fiberfin)

2.8.2.8 Continuous Production at Chromis Fiberoptics

While there are continuous production processes for SI-POF, PF-GI-POF could only be produced until just recently from preforms. Chromis Fiberoptics - pre-viously Lucent, OFS - has developed a process for the continuous production of such fibers ([Rat03], [Whi03], [Whi04a], [Whi05], [Park05b] and [Pol06a]). First a SI fiber of CYTOP material with a doped core is produced in a double extruder. The fiber is wound around a heated cylinder. Here the dopant diffuses outwardly resulting in the GI profile. The 500 μm PMMA protective layer is then applied and the fiber can be wound up. The fibers almost attain the parameters of POF from Asahi Glass which has had about 10 years of experience in the field.

protectivelayer

extruder

diameter control

to the take up drum

cladding extruder(CYTOP)

core extruder(CYTOP + dopant)

coextrusion head

heated tube

coextru-sion head

capstan

step-index profile

dopantdiffusion

GI-POF indexprofil

-100 -50 0 50 100radius (μm)

index difference

Fig. 2.203: Continuous PF-GI-POF fabrication ([Pol06a])


The insert shows the final index profile with an approximately parabolic curve. The manufacturer indicates the bandwidth-length product of the fiber as being 400 MHz · km.

2.8.2.9 GI-POF with Additional Cladding

As already indicated above, a reduction in the bending losses plays a great role with polymer fibers. For SI fibers a considerable improvement could be achieved by means of a second cladding. Extremely small bending radii can be attained through fiber bundles or multi-core fibers respectively.

For graded index fibers as well, an additional cladding layer with a smaller refractive index evidently offers clear advantages in regard to the bending beha-vior without dramatically reducing the bandwidth. A PF-GI-POF with an addi-tional 6 μm thick cladding layer is introduced in [Oni04] and [Sato05].

Figure 2.204 shows the measured bending losses for three different fibers with different index jumps between the edge of the core and the additional cladding (around n = 0.002, n = 0.005 and n = 0.014). Even with an index jump of 0.005 a bending radius of 10 mm with an attenuation below 0.1 dB can be attained. The bandwidth-length product of the fiber lies between 1,800 MHz · km and 2,700 MHz · km. The fiber attenuation amounts to 30 dB/km at 850 nm, measured with ODTR.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 10 20 30 40 50 60

180°-bending loss [dB]

1.325

1.330

1.335

1.340

1.345

1.350

1.355

0 20 40 60 80 100 120 140 160 180 200

x [μm]

bending radius [mm]

n

Fig. 2.204: Reduction of the bending losses due to a Semi-GI profile ([Sato05])

This method can also be employed for PMMA-GI fibers. The results for a 1 mm thick fiber are presented in [Aru05]. The attainable bending radius drops to below 5 mm with an additional PVDF cladding (polyvinylidene fluoride, n = 1.42). The bandwidth-length product of the fiber is 1,500 MHz · 100 m and re-mains quite constant up to 10 mm. It only drops under full launch and with a 5 mm bending radius to 500 MHz · 100 m. The attenuation at a 90° bend is com-pared to a conventional PMMA GI-POF in Fig. 2.205.


-0.5

0.0

0.5

1.0

1.5

2.0

0 10 20 30 40 50

bend radius [mm]

bend loss [dB]PVDF clad GI-POF NA of the GI core region = 0.17

PMMA based GI-POF NA of the GI core = 0.21

Fig. 2.205: Bend losses in Semi-GI-POF according to [Aru05]

In addition to the extra cladding layer a so-called W-profile for GI fibers has also been developed. Here the goal is to improve the attainable bandwidth. Measurements on PMMA GI-POF with this W-profile and different index expo-nents are presented in [Tak05b]. The W-profile is characterized by a very steep index drop directly at the core-cladding interface. Figure 2.206 shows the index curve.

Fig. 2.206: W-profile for PMMA-GI fibers ([Tak05b])

Furthermore, fibers with a NA of 0.20 and a -parameter (index exponent of the rise outside the core-cladding interface layer) have been produced with index exponents between 1.9 and 5.2. Figure 2.207 shows the theoretically calculated and measured bandwidths.


1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.50.2

1.0

0.5

5.0

2.0

0.3

3.0

3 dB bandwidth [GHz 100 m]

calculated for GI-POF

W-shaped POF

GI-POF

profile index exponent g

Fig. 2.207: Bandwidths of PMMA-GI-POF, improvement by W-profile [Tak05b]

PF-GI-POF with optimized index profiles are presented in [Ebi05]. Their band-width attain that of MM-GOF and in the short-wave range even surpasses it (Table 2.32). The high bandwidth is attained through the approximately ideal index coefficients of 2.05, i.e. in combination with the low chromatic dispersion of the material.

Table 2.32: Bandwidths comparison of GI-GOF and POF according to [Ebi05]

Bandwidth

wavelength 650 nm 780 nm 850 nm

PF GI-POF 8.39 GHz 8.50 GHz 9.54 GHz

SiO2-GI-GOF 5.27 GHz 7.34 GHz 9.31 GHz

Figure 2.208 shows the best attenuation values over time for some of the fibers listed above. PMMA fibers (SI and GI) reached their theoretically maximum possibilities in the mid-80s. Since then, other index profiles (MSI, MC, DSI) have also reached this order of magnitude (approx. 130 dB/km at 650 nm and 80 dB/km at 570 nm). Any differences in measured values and specifications are more likely to result from different measuring conditions than from differences in quality.

The PF fibers have been continually improved, at least as far as the laboratory results are concerned. The best values were attained in 2003 with about 8 dB, almost one magnitude still above the theoretical limits. In the past three years no further progress has been made with the attenuation. On the other hand, there has been some success in attaining a high launch-independent bandwidth with opti-mized refractive index profiles and in reducing the bending sensitivity.


year

attenuation [dB/km]

SI-PMMA at 650 nm

SI-PMMA at 570 nm

PF-GI at 1.300 nm

d8-GI at 688 nm

SI-d8 at 680 nm

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

GI-PMMA

10

100

1,000

200

500

20

50

5

Fig. 2.208: Development of POF attenuation until the year 2005

2.8.3 Cable Manufacturing

This chapter discusses the structure and properties of various cable structures with POF wires. Different applications place different demands on the mechanical shielding of the polymer optical fiber. SI-POF (Step Index Polymer Optical Fiber) is a promising medium for relatively short transmission distances of 100 m. Poly-mer plastics such as polymethylmethacrylate (PMMA) or polycarbonate (PC) are used as the primary core material for manufacturing these fibers. Fluorinated poly-mers, silicone or fluorinated PMMA materials are used as cladding material with a reduced refractive index of ncladding ~ 1.42 as compared with the core material ncore > 1.48 (Fig. 2.209).

Due to the large refractive index difference, numerical apertures of up to 0.50 are attained. Various manufacturer versions of optical fibers are shown in Fig. 2.210, in which glass or plastic are combined for the core and cladding mate-rial. The relatively thin glass fibers are mechanically fragile and must therefore be protected by a multilayer cable construction. The POF is so flexible that a simple jacketing of the optical cladding suffices as a cable construction.


0.98 mmd

1.00 mmD

1.492ncore

1.416ncladding

0.47NA

n

r

ncore

ncladding

D

d

core material: Polymethylmethacrylat (PMMA) cladding material: fluorinated PMMA

Fig. 2.209: Typical SI-POF parameters

Glass fibers with polymer optical cladding represent an intermediate step. They also have a relatively simple construction (two-layer plastic coating around the optical cladding). The large core diameter allows only step-index profiles.

10/ 125/ 250 μm

50/ 125/ 250 μm

singlemode glass fiber

multimode glass fiber

980/ 1000 μm

200/ 230 μm

glass fiber with polymer cladding

polymer fiber

0 mm 0.5 mm 1.0 mm

optical core

optical cladding

primary coating

secondary coating

strength member

outer jacket

Fig. 2.210: Comparison of different kinds of optical fibers


Until recently, step-index profile fibers were manufactured almost exclusively from polymer plastics having a typical outer diameter of 1 mm. These SI-POF exhibit significant transmission ranges with a minimum of attenuation for wave-lengths between 400 nm and 900 nm (Fig. 2.211).

The effective spectral loss windows are at 520 nm, 570 nm, 650 nm, and 760 nm. With improved purity, homogeneity and deuterated or fluorinated poly-mers, it is possible to reduce attenuation to 10 dB/km, as has already been des-cribed in Chapter 2.7.5.

50

100

1,000

10,000

200

500

2,000

5,000

450 500 550 600 650 700 750 800 850

wavelength [nm]

attenuation [dB/km]

PMMA

PC

attenuation minimum

Fig. 2.211: Attenuation spectrum of different POF made from PMMA or PC

Polymer optical fibers that are flexible and break-resistant can be produced with a relatively large diameter (up to 1.5 mm or even more) and are thus easy to handle and to install. The large core diameters in combination with the numerical aperture make simple connection fittings and equipment possible with low de-mands on precision.

2.8.3.1 Cable Construction with SI-POF Elements

SI-POF cables or lines must always be flexible when laid/installed at the end user place. SI-POF must also be flexible for mobile applications.

The flexibility of a line or cable depends on the number and dimensions of the stranding units with the number of the layer changes of the individual stranding elements. The shorter the pitch length is and the larger the number of layer chan-ges, the larger the flexibility of the stranding unit. The pitch length of the indivi-dual POF wires or the stranding elements with the proper diameter has a major influence on the flexibility of the stranding elements. The shorter the pitch length, the more flexible the stranding unit will be (Fig. 2.212).


flexibility

length of lay

Fig. 2.212: Schematic diagram of the relationship between the pitch length and the flexi-bility of the stranding construction

2.8.3.2 Non-Stranded SI-POF Cables

SI-POF Simplex Cable When processed into a cable with the respective strain relief, the SI-POF can be coated with a diffusion lock made from metal over the first cladding, if required.

An absolute diffusion lock can be attained exclusively with a closed tube, for example with laser-welded metal tubing. The metal strip material for laser welding can be made of aluminum, copper or high-grade steel. The foil thickness is typi-cally between 50 μm and 150 μm for welding. For overlapping with or without gluing, the metal foils have a sandwich layer construction, i.e., 9 μm / 20 μm / 9 μm = metal / plastic carrier strip/ metal.

A jacked is extruded onto the traction elements in combination with the metal diffusion locks. This layer is practically always flexible and sturdy; polyurethane or polyethylene are the preferred materials. The next illustration (Fig. 2.213) shows two typical SI-POF simplex cable constructions.

outer sheath

metal band

inner coating

cladding

fiber core

2.2 mm 2.3 mm

Fig. 2.213: Structure of optical fibers with internal cladding


SI-POF Duplex Cable The simplest form of a duplex cable is the combination of two parallel POF wires that are protected by shielding and equipped with traction elements. Various con-struction options for a duplex cable or duplex line are possible. Two very well known cable constructions are shown in Fig. 2.214.

foil tape lapping

POF-element

inner coating

outer sheath

strain relief element/ rip cord 5 mm

2.5 mm5 mm

Fig. 2.214: SI-POF duplex cable in a round cable and flat cable form

With these duplex cable constructions, care must be taken to ensure that the strain-relief elements in the plugs or on the connectors are included in processing. This is necessary because the temperature influence on the SI-POF wires is con-structed in such a way that optimum temperature characteristics are ensured in the temperature range from -40°C through +80°C.

SI-POF and GI-POF Ribbon Cable A ribbon cable with n SI-POF elements can be constructed as an extension to a duplex cable. The SI-POF elements are lined up in parallel as a comb and com-bined in either groups of 5 or 10 elements. A thin protective coating is extruded over this ribbon cable with the respective traction and support elements in one work cycle. Various SI-POF ribbon cable constructions with a modular design are illustrated in Fig. 2.215.

2.3 2.35.2

outer sheath

strain relief element

POF

inner coating

26 mm

13 mm

twin group 5 cables group

10 cables group

Fig. 2.215: SI-POF ribbon cable with traction and support elements


The cross-sections of two POF ribbon cables from [Boc04] are shown in Fig. 2.216. The individual fibers have each been extruded in a joint acrylic cladding.

Fig. 2.216: Ribbon with four 500 μm SI-POF (above) und eight 120 μm/500 μm GI-POF (below, [Boc04])

For the OVAL project (see Chap. 6) of the POF-AC Nürnberg Nexans had pro-duced 8-strand ribbon cables made of SI- and GI-POF each with a 500 μm dia-meter. The cross-section of a prototype with PMMA-GI-POF (Optimedia) is shown in Fig. 2.217.

Fig. 2.217: POF-ribbon cable with eight 500 μm OM-Giga-fibers (dimensioning in μm)

The spacing between the individual fibers deviates only slightly from 500 μm. Only in a vertical position great deviations do arise which can easily be avoided by better guiding of the individual fibers in the extrusion tool.

In order to investigate the influence of the ribbon cable production on the optical parameters, the spectral attenuation and the bandwidth were determined on the SI-POF ribbon cables. The results are shown in Fig. 2.218 and 2.219.


450 500 550 600 650 700 750 800100

1000

200

600

400

300

800attenuation [dB/km]

wavelength [nm]

fiber 1 fiber 2fiber 3 fiber 4fiber 5 fiber 6fiber 7 fiber 8

Fig. 2.218: Single fiber attenuation in the ribbon cable

The attenuations of the 8 fibers agreed within the usual measurement error of ±0.5 dB. There were also no significant deviations in the frequency response in Fig. 2.219.

-35

-30

-25

-20

-15

-10

-5

0

+5rel. level [dB]

frequency [MHz]

1 10 100 1000 3 30 300

Fig. 2.219: Frequency response of the fibers in the ribbon cable

In one last experiment we investigated whether the ribbon cable production increased the mode mixture in the fibers. We determined the far field width of the individual fibers and ribbon cables with under filled launch for different lengths. The experiment on the ribbon cables was repeated after annealing (120 min. at +90°C) and aging (200 hours). As can be seen in Fig. 2.220, it took practically the same length of time in all four cases to achieve equilibrium mode distribution. In other words, the ribbon cables did not influence the mode mixing processes.


12

16

20

24

28

32

36

fiber

ribbon cable

annealed

aged

FWHMeff [°]

lPOF [m]

high NA-fiber NAlaunch = 0.10

0.1 1.0 10.0 100.00.3 3.0 30.0

Fig. 2.220: Effect of mode mixing in fiber ribbons ([Har06])

SI-POF Hybrid Cable Hybrid cables are characterized by the fact that they are constructed from a combi-nation of SI-POF elements with cooper-insulated wires that can be joined together individually or in pairs.

Furthermore, there are hybrid cable combinations in a coaxial construction with a metallic tube, the so-called POF-CMT element (CMT = Corrugated Micro Tube). The illustrations shown in Fig. 2.221 point out possible combinations with SI-POF copper elements or SI-POF aluminum elements in a coaxial construction.

POF CMT

isolation

triaxduplexsingle

basic element hybrid-cable

3 ... 4 mm2 ... 3 mm 7 ... 9 mm 4 ... 5 mm

Fig. 2.221: New design for POF with CMT as electrical conductor

The advantage of such hybrid cable constructions is the possibility of supplying current directly to the transmitter and/or receiver of the individual SI-POF ele-ments ([Ziem99a], [Ziem99b]). The connector combination for hybrid cable con-structions are well known and are used in the automotive field.

Apart from the coaxial hybrid solution, the layer-stranded hybrid solution is also well known (Fig. 2.222 and 2.223).


6.5 mm

copper wire

foil

POF980/ 1000 μm

inner coating

outer sheath

7.5 mm

copper wire

support element

POF 980/1000 μm

inner coating

outer sheath

strain element

Fig. 2.222: Layer-stranded POF-Cu cables (principle)

In these cases, insulated copper wires and POF wires are processed either into a group of four or as stranded layers with several stranding elements. The copper wires are used with diameters of 0.5 mm to 1.5 mm. Thicker copper wires are pro-cessed as braided wires, because the flexibility of the cable usually does not meet customers’ requirements.

Fig. 2.223: Hybrid POF-Copper cable

2.8.3.3 Stranded SI-POF Cables

Introduction SI-POF cables or SI-POF lines are products that must bend easily when they are used and when they are processed. This requirement must be met for the manu-


facturing process or for transport purposes or for winding the cables or lines on production-machine reels or shipping reels or when sold in rings. The individual SI-POF elements are twisted in a screw-like fashion around an imaginary center-line. Twisting is necessary in order for the manufactured products to be flexible and portable.

The advantage of twisting is that the stranding element is stretched and com-pressed alternatively on the inner and outer side of a curved section (Fig. 2.224). If the section in which a SI-POF stranding element is wrapped 360° around a twist axis that is considerably smaller than the curved section, the strain and pressure in a stranded construction are constant and it is possible to bend this SI-POF cable without deformation.

Fig. 2.224: Comparison of cable constructions with short or long lay lengths in terms of the bending characteristics

The flexibility of an SI-POF cable or SI-POF line is a function of the geometric dimension of the stranding elements and of the change in layers present in a cable construction. For example, a large number of layer changes results in a greater flexibility of the SI-POF cable construction.

The SI-POF stranding elements are wrapped spirally around the twist axis in various machine configurations. The foundation for these various machine designs is always the result of a rotary motion with a linear motion. This can be seen schematically in Fig. 2.225.


n1

s

12 3

45

d

DA

n2

1. rotor 2. stranding elements 3. stranding unit 4. capstan gear 5. stranding axis

s: pitch length n1: rotational speed of

the stranding basketDA : diameter of the

stranding basket

d: diameter of the stranding unit

n2: rotational direction and speed of the capstan gear

Fig. 2.225: Schematic diagram of the spiral-shaped strands

The option of being able to twist SI-POF elements together is determined by the following parameters.

Pitch length Lay direction Multiplication factor Number of strands

2.8.3.4 Principles of Stranding

Pitch Length The pitch length is the distance between two points on the twist axis. Within these two points, the SI-POF element has been rotated 360° around the twist axis. The lay length is calculated from the following variables:

1

m

1

2A

n

1000vs[mm]

n

nDs

where DA: Diameter of the capstan gear n1: Rotational speed of the stranded basket n2: Rotational speed of the capstan gear vm: The machine’s pull-off speed

During the manufacture of twisted SI-POF cables or SI-POF lines, the lay length s must be determined very exactly because of the precise geometry in-volved. This means that for stranding machines for SI-POF elements that are twisted via a capstan gear or caterpillar, the diameter of the stranding elements must be taken into account. In practice, a deviating diameter for the SI-POF stran-


ding construction is the result and increases the pitch length manufactured. The geometric assignment is easy to see in the enclosed illustration (Fig. 2.226); the manufacturing pitch length SH is calculated from it.

2

3

1

DA

d

1: capstan gear 2: fiber loop 3: POF

Fig. 2.226: Diagram for explaining the concept of 'manufacturing pitch length'

The manufacturing pitch length is calculated from the following parameters:

A

AH

D

dDss

sH: Manufactured pitch lengths: Pitch length in machinesDA: Diameter of the capstan geard: Diameter of the stranded unit

Lay Direction The rotational direction of the stranding basket determines the lay direction. The following distinction is made depending on the sense of direction of the helix:

Z-lay means a right-handed thread S-lay means a left-handed thread (Fig. 2.227)

Fig. 2.227: Schematic explanation of the lay direction


The following diagram (Fig. 2.228) illustrates how an SZ stranding is to be interpreted. It can be seen that, after a number of rotations, the lay direction is changed. In contrast to classic basket stranding, SZ stranding has the advantage of having a pull-off speed that is 5-20 times faster.

S Z S Z S

Fig. 2.228: Explanation of the lay direction schematically

Economic and engineering stranded cable products are manufactured exclu-sively using the SZ stranding method, i.e. also for POF applications.

In classic production, SI-POF stranding elements constructed from several stranding layers are given alternatively a Z and an S direction. This cable con-struction element - the SZ-stranding method - for SI-POF results in a very com-pact geometric shape of the stranding construction, which allows it to cushion well both traverse and longitudinal forces. The stranding element is to ensure that the optical transmission values are retained during the manufacturing process of the cable product and to ensure that there are no changes after laying the cables and in subsequent operation.

Multiplication Factor The helical SI-POF stranding element (Fig. 2.228) is longer in the stranded unit. The stranding method always leads to an increase in material consumption. The ratio of the laid length L of the SI-POF stranding element to the lay length of the stranded unit results in the well-known multiplication factor f = L/s. The multi-plication factor f is determined from the pitch length and the average diameter Dm

in the stranding layer. The multiplication factor can be easily derived from the triangle shown in

Fig. 2.229.

22m s)D(L and 1

s

D

s

s)D(

s

Lf

2m

22m

with L: Laid length L = s/cos f: Multiplication factor Dm: Average diameter of the stranded layer s: Pitch length of each stranded layer

For relatively large pitch lengths (Dm « s), the calculation can be simplified as follows:

/2s/D1f 2m


Dm

s

d

D

L

s

Dm

Fig. 2.229: Graphical representation of the SI-POF stranding element

Number of Strands To characterize the bending properties of an SI-POF stranding element v, the number of strands is formed from the quotient of the pitch length and the average diameter Dm (v = s/Dm).

s: Pitch length of each stranded layer Dm: average diameter of this stranded layer v: Number of strands

Production developments in stranded cable constructions or SI-POF cable con-structions have lead to the number of strands being v > 8. By using the number of strands v, the multiplication factor f can be easily calculated.

²/2v²1v

v1

vf

222

Layer Structure Standard SI-POF elements have a simple geometric shape but have an exact dia-meter. This makes it easy to calculate SI-POF cables or SI-POF lines. An SI-POF cable in its classic form, i.e. with a core element, has the same diameter as the SI-POF element; it can be constructed in a circular fashion with 6 SI-POF elements in the same layer. The cladding lines are in contact with each other. Two different core layers have been adopted schematically in Fig. 2.230. The other layers are calculated and shown. In Table 2.33 and Table 2.34, the number of elements and the diameters have been compiled for a general case and for the case with d = 2.3 mm respectively, whereby the variables have the following meaning:

n: Layer number z: Number of elements per position

z: Total number of the elements to the layer n d: Diameter of the cable unit Dm: average diameter of the unit D: Diameter of the layer


D2

D3Dm3

Dm2

d = 2.3mm

D2

D3

Dm3

Dm2

D1Dm1

Fig. 2.230: SI-POF cable (layer structure)

Table 2.33: Dimensions of layer-stranded POF cables in general

n z Dm D z n z Dm D z

1. 1 - 1 · d 1 1. 2 1 · d 2 · d 2

2. 6 2 · d 3 · d 7 2. 8 3 · d 4 · d 10

3. 12 4 · d 5 · d 19 3. 14 5 · d 6 · d 24

4. 18 6 · d 7 · d 37 4. 20 7 · d 8 · d 44

5. 24 8 · d 9 · d 61 5. 26 9 · d 10 · d 70

6. 30 10 · d 11 · d 91 6. 32 11 · d 12 · d 102

Table 2.34: Dimensions of layer-stranded POF cables with d = 2.3 mm

n z Dm D z n z Dm D z

1. 1 - 2.3 mm 1 1. 2 2.3 mm 4.6 mm 2

2. 6 4.6 mm 6.9 mm 7 2. 8 6.9 mm 9.2 mm 10

3. 12 9.2 mm 11.5 mm 19 3. 14 11.5 mm 13.8 mm 24

4. 18 13.8 mm 16.1 mm 37 4. 20 16.1 mm 18.4 mm 44

5. 24 18.4 mm 20.7 mm 61 5. 26 20.7 mm 23.0 mm 70

6. 30 23.0 mm 25.3 mm 91 6. 32 25.3 mm 27.6 mm 102

Cable Materials The specification profile for SI-POF cable or SI-POF lines in various fields of applications such as in industry, in office environments or in the automotive field place the highest demands on the material components.

Thermoplastic materials (polymers) are preferred that have been mounted to the cable using an extrusion process. Excellent mechanical properties are needed so that the values listed below are ensured when SI-POF cable or SI-POF lines are installed.


AbrasionRepeating bending characteristics TorsionAccelerationHammer blow Small bending radii

Especially in the automotive field, the material must be highly resistant to the following properties:

Resistance to oil Cooling lubricant resistance Steam Hot gases

The demand for materials that are temperature resistant comes from users. These customers are in the automotive field, in industry or in the cable-installation field for buildings. Special halogen-free material properties are desired in order to provide on-site safety to customers and consumers alike.

Today’s selection of modern plastic insulation and cladding mixtures, which in part can be improved through various methods of crosslinking, should and must protect the SI-POF cables or SI-POF lines in all types of applications.

In case of an accident, special plastic optical fiber cables are to have emergency running properties. SI-POF hybrid cable constructions ensure this reliability to a very high degree.

The mechanical properties of thermoplastic materials such as

HardnessDensityTensile strength Elongation at break Tensile stress value Compression strain Impact resistance Electrical properties

can be found in the relevant data specifications of the standardized norms or the data specifications of the chemical industry. Preferred plastic materials are:

PolyethylenePolypropylene Polyurethane Cross-linked thermoplastics

The properties that have been improved by cross-linking are those of thermal resistance and higher mechanical strength. In addition, the resistance to solvents has also been increased, which can be seen by the fact that less swelling and cracking occur for polymers with residual tensile stress.


The essential physical properties of some of the important materials are listed in section 3.3.6.

A very good alternative is a combination of plastic and metal, for example, with the corrugated micro tube. Metal in the most varied constructions, whether as a steel alloy, in aluminum or in copper keeps the SI-POF in an expanded tempera-ture range protected against mechanical and thermal strain.

2.8.3.5 Corrugated Micro Tube Cables

Corrugated micro tubes have been used to protect cables for quite some time. Nexans was the first company to encase polymer optical fiber wires for manufac-turing resistant cables. Because of the small diameter of the POF, special corruga-ted micro tubes (CMT) were needed. More detailed descriptions of the mechanical and thermal properties are found in [Schei98], [Zam99], [Ziem99a], [Ziem99b] and [Zam00a]. Figure 2.231 illustrates a POF wire with aluminum corrugated tube.

Fig. 2.231: POF wires with corrugated micro tubes

Possible applications for CMT cables will be discussed later in Chapter 8.1.1.7. The manufacturing process for corrugated tubes is described below.

Corrugated Tube Process The UNIWEMA (Universal Corrugated Tube Machine) has become a standard piece of equipment for cable plants worldwide. The origins of the corrugated tube process go back to the 1940’s.

The corrugated tube process as practiced today is a butt-welding process for small dimensions (for example POF wires). A thin metal strip is formed around a cable core and formed into a small metal tube. The strip edges that form a butt joint are welded into a tube cladding by a laser beam under protective gas (argon and/or helium) and then corrugated in a spiral-shaped way or as rings (Fig. 2.232).


Fig. 2.232: Corrugated tube for POF

The UNIWEMA is used to weld copper, aluminum and steel strips or steel alloys or alternative materials. The machine creates smooth and corrugated metal tubing in an economical manner.

The tube welding process is continuous and fast. All weldable metals such as copper, aluminum, steel and their alloys can be processed. The process can be used for manufacturing small metal tubes for core diameters ranging from 1 mm to 500 mm. Strip thickness’ of 0.05 mm to 4.0 mm are welded with a laser using the WIG process. Neither burrs nor bulges are produced at the welding seam (Fig. 2.233).

Fig. 2.233: Welding seams in laser welding


Due to the concentrated thermal effect of the welding source, the welding zone is limited on the metal edges. The heat is quickly dissipated over the tube. Since the welding zone is covered by a protective gas shield, the formation of an oxide layer is prevented.

Corrugated Tube for POF Applications Metal tubes manufacture in compliance with the UNIWEMA procedure (Fig. 2.234) can be used for all POF cables. This applies to metal tubes of steel and welded special steel alloys as well as smooth or corrugated copper or alumi-num tubes that have been welded lengthwise. Corrugated copper tubes are used wherever a particularly high conductivity or large dissipation of heat is required. Due to its comparably small weight when used with thin metal strips, corrugated cable tubes can be easily transported and installed. The corrugated tubing is easy to bend and particularly resistant to external deformation in the radial direction. It is absolutely hermetic. This makes it possible to operate corrugated cable tubes, and POF elements under pressure and in a vacuum.

Fig. 2.234: Laser welding device ([LZH01])

Laser Welding The laser beam is monochromatic and coherent and can be easily focused. As a result, a high power density can be achieved at the processing point - the V-seam between the strip edges (Fig. 2.235).


Fig. 2.235: Welding seam with laser beam ([LZH01])

By applying the auxiliary gases argon and/or helium in such a way that the beam power is absorbed in the capillaries, the coupling properties of the plasma can be controlled. The actual welding joint is produced by the melt converging behind the capillaries (Fig. 2.236).

welding zone (solid)

welding zone (fluid)

laser beam

vapour (plasma) channel

laser induced plasma

metal vapour

laser beam

welding zone (solid)

welding zone (fluid)

direction of welding keyhole weldingconduction limited welding

Fig. 2.236: Principle of laser welding


Keyhole welding causes the process heat to be uniformly distributed at mini-mum levels over the entire welding zone (Fig. 2.237). Typical welding joints are butt-welded or overlapping welding seams, weldable materials steel, special steel, brass, copper, aluminum and special metal alloys. Thin clad metal foils made of aluminum/plastic/aluminum can be used for laser welding. Fluted steel sheets can be welded overlapping or butt-jointed with a YAG laser or a diode laser.

ND:YAG-laser beam source

laser control device

dataacquisation

process computer

controller

quotient pyrometer

partially transparent optic

reflects Nd:YAG radiation transmits heat radiation

detected heat radiation

laser beam

workpiece

laser fiber

laser optics

beam-material interaction zone

modified track

feed direction

Fig. 2.237: Structure of a laser welding system

2.9 Microstructured Fibers 215

2.9 Microstructured Fibers

In addition to classic optical fibers which consist of a core and cladding there are also microstructured fibers in which the wave guiding does not rely on a refractive index profile, but on holes along the entire length of the fiber. Normally, wave guiding in optical fibers is based on the effect of total reflection in the general sense of the term. The core consists of a material with a higher refractive index than the surrounding cladding material. In this fiber configuration special field dis-tributions, so-called modes or eigenmodes, can be guided within the fiber. These modes experience an effective refractive index of the fiber, which lies between the maximum refractive index of the core and that of the cladding material.

In 1996, J. Knight et. al. demonstrated a new kind of optical fiber, the wave guide characteristics of which were no longer based on a rotation-symmetrical re-fractive index. This created a variety of completely new possibilities and novel functions ([Kni96] and [Kni97]). These fibers now only consist of one material, usually silica glass, and have a structure of the cross-section with air holes. The holes in this structure are as a rule considerably smaller than the wavelength of light so that they do not act like objects on which light is reflected or scattered. Instead they change the refraction characteristics of the material.

The material is changed in such a way that it acquires new kinds of charac-teristics. Relatively simple and specific characteristics can be created with these fibers, e.g. for dispersion, dispersion slope, modal field radius and others.

For some years now microstructured fibers have also been made of polymer. These fibers with low temperature processes can be produced on the basis of the low melting point of polymers and other characteristics, thus resulting in possibi-lities for new kinds of fiber geometries and also potentially new applications.

In the following section we would like to deal with the fundamental wave guiding mechanisms. The different types of fibers and their specific characteristics will be introduced and the methods for producing these different types of fibers will then be shown. We would particularly like to take a close look at the diffe-rences between microstructured fibers made of glass and polymers. Applications which are possible with these fibers and are presently the subject of research will then be introduced. Some of these applications can even be obtained commercially now. Finally, the present state of development will be discussed and we will ven-ture a prognosis as to where the limits for such fibers may lie in the future.

2.9.1 Kinds of Wave Guiding

Wave guiding in microstructured fibers is determined by the structure of the cross-section along the entire fiber. Holes which locally vary the refractive index very strongly are normally put into the fiber along its entire length. These areas with noticeably different refractive indices are very small in relation to the wavelength so that they cannot be resolved by light and only have an indirect influence on the propagation characteristics of the light.

216 2.9 Microstructured Fibers

There are two fundamental mechanisms which exercise this influence: holes either act as a kind of doping by changing the effective refractive index of the material in average ([Gho99]) or they are put into a regular, grid-shaped arrange-ment so that they act like a kind of meta-material ([Cre99]. Other materials which have a greatly differing refractive index from that of the core material can also be used). Such fibers can exhibit effects with a great degree of wavelength depen-dence since such arrangements have similar characteristics as e.g. Bragg gratings, in which the light at certain wavelengths can be constructive or destructive over-lapped. The two-dimensional pendant to such a Bragg grating are the Bragg fibers in which concentric areas with greatly differing refractive indices alternate at regular intervals ([Yeh78]). Constructive overlapping waves can come about at certain wavelengths thus resulting in wave guiding. At other wavelengths light is not guided. One can therefore surmise that such fibers are capable of having strong wavelength-dependent characteristics.

A new kind of wave guiding occurs in such fibers with regular structures. This wave guiding is possible in cores made of air as opposed to those fibers based on total internal reflection. For wave guiding with total internal reflection it is essen-tial that the core material has an effective refractive index which is higher than that of the cladding. This is not necessary with fibers having a “photonic band gap”. Because of the regular structure within the fiber band structures are formed analogous to electrical semiconductors in which certain energy states of light waves are allowed and others are rejected resulting in light waves which can remain within the material and others which cannot. When there are light waves which have permissible energy states within the core area, but not in the cladding, then the light must stay in the core and is guided through this band gap since they cannot exit into the cladding.

2.9.1.1 Effective Refractive Index

Fibers based on the effect of an effective refractive index can intuitively be under-stood most easily. They are doped with the material by introducing air or other materials. The holes made must be very small in relation to the wavelength and should be as randomly arranged as possible. The effective refractive index then results from the volume ratio of the two materials (e.g. one talks about air frac-tion). The greater the proportion of air, the smaller is the effective refractive index of the material. Fibers based on an effective refractive index should have holes relatively small in relation to the wavelength of the light so that the holes as such can no longer be resolved. Also, these holes should be introduced into the material in an irregular manner as possible so that the geometry and arrangement of the holes do not have any influence on the characteristics of the material (see Fig. 2.228).

Such fibers are basically not different from traditional fibers in which the core has a higher refractive index than the cladding. Consequently, there is a form of total reflection. Since the effective refractive index can fundamentally only be reduced by doping with air, the cladding area in such fibers is normally structured with holes. The core is mostly undoped glass. Such fibers can be described as


being similar to normal step index glass fibers, whereby the fiber parameter de-pends on the wavelength of the light ([Mor03a ] and [Mor05b]). The reason for this is that the influence of the holes varies greatly depending on the wavelength of the light which also depends on the relationship between the hole diameter and the wavelength and whether light can resolve the holes.

Fig. 2.238: MPOF with effective refractive index according to [Lar02a]

2.9.1.2 Photonic Band Gaps

In addition to the fibers whose refractive index profile arises from the effective refractive index resulting from the holes there is also wave guiding on the basis of a photonic band gap ([Cre99]). Fibers based on the principle of a photonic band gap behave fundamentally differently from the fibers with an effective refractive index just discussed.

These fibers must have holes introduced in a specific periodic arrangement so that a kind of meta-crystal comes about. According to the Bloch theorem the neighboring holes act like elementary cells which are repeated regularly in several dimensions resulting in new kinds of characteristics for the meta-crystal. Just as with semiconductors, energy bands can be formed which originate from the perio-dic structure of the material. In semiconductors these are the periodically arranged atoms of the semiconductor material; in fibers with a photonic band gap it is the periodically arranged holes.

In such fibers the light guiding comes about when the light of a certain wave-length, and thus photons with a specific energy, possess allowed energy states in the core area while the same energy states are not permitted in the cladding area. Thus, the photons in this energy state can only stay in the core area of the fiber (see Fig. 2.239).


eff

ect

ive

ind

ex

neff

0 1 2 3 4 5 6 7normalized frequency

nSiO2

nair

PBGF-mode1.0

1.2

1.4

1.1

1.3

Fig. 2.239: Distribution of intensity in a large-mode-area-laser fiber according to [Lim03] (left). Effective refractive index of the radiation modes of the cladding (grey) with position of the bound defect mode in the band gap (center) and the mag-netic field strength of the linear polarized fundamental mode for = 2.27 μm, d = 1.993 μm, D = 4.54 μm at = 1,55 μm with neff = 0.977 (right).

The form of the energy bands, i.e. the energy areas, which correspond to the permissible energy states is greatly dependent on the arrangement of the indivi-dual holes. Even small deviations can lead to great changes in the energy bands so that with this kind of fiber only slight tolerances are allowed in the arrangement of the holes. Nevertheless, these fibers permit greater possibilities for structuring ([Arg06]). As a consequence, propagation characteristics such as dispersion, dis-persion slope, effective area, etc. can have relatively large dimensions. Especially for very narrow-band applications, e.g. sharp-edged filters, fibers with photonic band gaps can be employed quite well. This is also true for high-performance applications in which the linear characteristics of the hole core are used ([Lim03], [Mat05b] and [Nie06]).

Fig. 2.240: Air-hole - MPOF with 220 μm outer diameter/5 μm hole distance, [Eij03a]


2.9.1.3 Bragg Fibers

Bragg fibers consist of concentric rings with different refractive indices. These rings act like a Bragg grating in radius direction so that they reflect certain wave-lengths which are adapted to the spacing between the rings while letting other wavelengths through. This results in wave guiding in only those wavelengths which the Bragg rings reflect. With all other wavelengths no wave guiding takes place. The fibers thus act like a filter and only let light through with very specific wavelengths ([Yeh78]).

Fig. 2.241: Cross section of a Bragg-fiber according to [Arg06]

The rings can be produced in a variety of different ways. Refractive index pro-files can be produced which have higher or lower refractive indices with specific radii. Microstructured fibers, however, are also possible where the rings with dif-ferent refractive indices are realized by hole structures. In this case rings with holes are arranged at regular distances from the fiber axis which, because of the effective index of this layer, acts like a layer with reduced refractive index.

Bragg fibers behave similarly to fibers with photonic band gap. They are also based on the exact arrangement of the holes or the layers with different refractive indices respectively. If the geometry is followed exactly very sharp-edged filters can be produced or fibers which are very selective in regard to the wavelength.

2.9.1.4 Hole-Assisted Fibers

In addition to these new kinds of fibers whose waveguide characteristics are based solely on the structures introduced, hybrid fibers have also been introduced which represent a cross between conventional fibers with refractive index profiles and microstructured fibers ([Has01]). These fibers have the same wave guiding as with conventional fibers. However, the additional holes change the propagation cha-racteristics so that you get other degrees of freedom in fiber design. In particular


ring-shaped hole structures are arranged around the core in order to reduce the bending sensitivity of the fibers ([Guan04] and [Nak03b]). The outer structure acts like an additional step in the refractive index profile which should hold part of the output emitted in the bend in the cladding area. This measure is supposed to increase the wave guidance without having to make compromises concerning the propagation characteristics of the fiber.

Fig. 2.242: Cross section of a hole-assisted fiber according to [Guan04]

2.9.2 Production Methods

Microstructured fibers can be produced in very different ways. Various production methods are possible with glass and polymer fibers.

2.9.2.1 Microstructured Glass Fibers

The first microstructured fibers were made of glass ([Kni96] and [Kni97]). Since glass has a very high melting point the production possibilities are limited. The fibers are mostly produced using the so-called stack-and-draw technique in which small glass tubes with different diameters - the number depends on the desired hole diameter - are put together in a bundle. Depending on the type of fiber, either a filled glass rod (effective index) or another small glass tube (photonic band gap) is used for the core. These small tubes combined then form the preform. They are melted and drawn into a fiber. A fiber cladding is generally drawn over the entire preform which then forms the outer area of the fiber. This only serves the purpose of stabilizing the fiber.

The fact that the small round glass tubes are combined into a preform generally only allows a few arrangements: rectangular, hexagonal or so-called honey comb structures. Even if you decide on hexagonal structures when arranging the holes, the hole spacing and the hole sizes can be put into a rather large range which can lead to diverse design possibilities.


Fig. 2.243: Cross section of a microstructured glass fiber, fabricated by stack-and-draw-technology ([Ort04])

Glass melting at low temperatures can also be extruded, whereby the glass is either melted or liquefied. The ensuing viscous fluid can then be pressed through specifically arranged nozzles which have the structure of the desired preform. This preform can then be used to immediately draw the fiber or to make a preform. This method of making preforms in effect allows the production of as many hole geometries as one likes. In principle, round holes and any kind of arrangement can be produced in this way. However, this production method is limited to glass with a low melting point. Consequently, silica glass, for example, cannot be processed.

The production engineering of microstructured fibers has improved tremen-dously in the past few years. Whereas the first fibers still had attenuations of seve-ral 100 dB/km, today fibers based on an effective index with attenuations per unit length can be produced below 0.3 dB/km at a wavelength of 1.55 μm ([Taj03]). Photonic band gap fibers permit attenuations per unit length up to 13 dB/km ([Smi03]).

2.9.2.2 Microstructured Polymer Fibers (MPOF)

Fibers made of plastics can be produced in a variety of different ways, especially since they can be processed at much lower temperatures. Whereas glass fibers can be drawn at temperatures around 2000°C, MPOF can already be drawn at 200°C ([Lyy04]). This not only allows simpler production techniques, but also permits the introduction of other materials into the fiber which would otherwise decom-pose, e.g. dyes ([Lar04]). However, there are also some disadvantages in regard to increased attenuation, lower operating temperatures, other operating wavelengths, etc. ([Lar06a]).


Microstructured polymer fibers can also be extruded and then drawn into fibers. The same limitations regarding geometry and production tolerances are valid for them as for glass fibers.

Researchers at the University of Sydney ([Bar04c] and [Lar01b]) have developed a particular kind of preform production in which a massive cylinder made of polymer is structured using drills with different diameters. At present, preforms up 65 mm in length can be structured with this method, otherwise the drills would be too long. As many geometries as one may wish can be produced in which both the arrangement and the hole diameter can be freely chosen. Present-day production processes have hole diameters between 1 mm and 10 mm with minimum spacing in between of about 100 μm which then shrink to their original size through drawing.

New kinds of process techniques can even produce elliptical holes which give the fiber an intrinsic double refraction. Preforms can either be poured into molds or around capillary tubes and then drawn into fibers ([Zha06]).

Fig. 2.244: Preforms of MPOF ([Lwin06], [Poi06e])

Other materials can be introduced into the fiber in addition to the holes. Fibers with metal wires for the poling of the material have been demonstrated as well as fibers with liquids in the capillary for controlling the propagation characteristics and doping materials for changing the optical and electrical characteristics ([Cox03b] and [Cox06]).

After the first MPOF was introduced at the end of 2001 ([Lar01b]), the tech-nology has continued to develop at an amazing pace. The fibers introduced back then still had an attenuation of 30,000 dB/km. In the course of time the individual process parameters have been continuously improved so that the attenuation could be steadily reduced. The process parameters optimized include conditions when drilling the preforms, rinsing and cleaning steps as well as drawing parameters. The best microstructured polymer fibers today have an attenuation of 200 dB/km and are thus not very far away from conventional polymer fibers which have an attenuation of about 120 dB/km at a wavelength of 650 nm.


0.1

1

10

100

0 5 10 15 20 25 30 35 40 45

achieved attenuation [dB/m]

months for the first publications

Sept. 2001

April 2005

Fig. 2.245: Development of the attenuation of the MPOF 2001-2005 ([Lwin05])

2.9.2.3 End Surface Preparation

Microtome cutting has proved to be a useful method for working on the end sur-faces of conventional polymer fibers. This method of work only produces unsatis-factory results with microstructured fibers since the fine, step-like structure at the end of the fiber in the holes can lead to defects and irregularities (see Fig. 2.246). These structures can be seen at the ends of all such fibers and on conventional polymer fibers, too. Nevertheless, it can be seen that the mechanical characteris-tics in particular of the MPOF intensify the step-like effect. These filigree struc-tures absorb the lateral forces and give in again after each thrust.

Fig. 2.246: Singlemode-MPOF cut by microtome, 1000-fold magnification

The direct cutting of the fiber with conventional cutting pliers can destroy the fine structures because of these lateral forces.


Fig. 2.247: Singlemode-MPOF cut by MOST®-tongs, 100-fold magnification

Other processing methods such as hot plate or subsequent polishing have also been investigated, but did not deliver any good results. The hot plate technique leads to inclusions at the end surfaces so that the original geometry can no longer be recognized. On the other hand, polishing leads to the deposition of rubbed off shavings and their removal into the holes. A reproducible coupling is therefore not possible since the influence of these inclusions or that of the deposited foreign matter in the structure’s holes is not controllable.

Better processing characteristics are shown by those MPOF which are surroun-ded by another, so-called buffer layer made of hard polyester. This layer absorbs a large part of the mechanical forces when cutting and prevents the breaking of the fine webs within the structure. Since such fibers consist almost exclusively of polymer they can almost be worked on like polymer fibers. Figure 2.248 shows the end surface of such an embedded fiber with a buffer layer. You can see that the fiber is not embedded centrically which leads in practical use to a lateral mis-alignment of the plugs and thus to plug losses and power redistribution. In the future you can expect, however, that the dimensions of the fibers will become greater and that the fibers can be better centered with new drawing techniques.

Fig. 2.248: End face of an embedded fiber with buffer layer; 100-fold magnification


No practical solution exists yet which can provide for good reproducibility and a high degree of reliability. Processing methods still have to be found for both practical and laboratory use which can meet the necessary requirements. In the case of termination in the field the end faces must allow acceptable losses; in the laboratory, preparation with high reproducibility is necessary. Both kinds of pre-paration still have to be developed.

2.9.3 Applications for Microstructured Fibers

Microstructured fibers allow a number of applications since their characteristics can be adapted to wide areas as desired because of the additional degree of free-dom in design and production. For example, waveguide characteristics such as chromatic dispersion and its slope can be adjusted as well as the mode field dia-meter. Other materials or fluids can be introduced into the fiber through the holes running along the fiber. These materials can change the propagation charac-teristics through which tunable components or sensors are made possible. Some possible applications for microstructured fibers are subsequently described. This list does not make any claim to being complete, but is solely intended to present the best-known applications as well as the commercial applications available today.

2.9.3.1 Dispersion Compensation

The first applications for microstructured glass fibers were the compensation for dispersion or its slope respectively. Because of the additional possibilities for fiber design, selection of the number of holes, their size and distance from one another, wavelength-dependent effects in particular such as chromatic dispersion can be adjusted very well. As already described above, the holes have weaker wave gui-ding at small wavelengths because the light can enter the bridges between the holes. This causes a different kind of wave guiding so that the fiber behaves as if it had another fiber parameter. By skillfully selecting the diameter of the holes and their spacing, the dispersion and higher orders can be adjusted very well. Disper-sion-compensating microstructured glass fibers are commercially available today.

2.9.3.2 Endlessly Singlemode

Microstructured fibers also allow applications which are not possible with conven-tional fibers. Such an application are the so-called endlessly singlemode fibers which have one mode in the entire wavelength spectrum and do not have a cut-off frequency. This characteristic can come about when the wave guiding changes with the wavelength.

In step index fibers the existence of one mode is clearly determined by the fiber parameter V which is proportional to the core diameter, the numerical aperture and the reciprocal value of the wavelength used. Thick fibers with large numerical apertures are characterized by a large fiber parameter V. Fibers are only guide


only one mode for V < 2.405, the first zero of the Bessel function of zeroth order. If the wavelength selected is large enough then V will become small enough at some point so that the fiber becomes singlemode. In microstructured fibers the fiber parameter is not simply anti-proportional to the wavelength since the holes in the cladding area act differently with large wavelengths than with small ones leading to a wavelength-dependent numerical aperture so that fibers can be pro-duced which are singlemode for all wavelengths ([Bir97], [Mor03b] and [Zag04]).

2.9.3.3 Birefringence

Since microstructured fibers are not rotation-symmetrical such as conventional fi-bers with a refractive index profile, for example, they tend to be birefringent. Typical hexagonal structures do not exhibit any birefringence. However, when this symmetry is disrupted, e.g. through production tolerances, then these fibers are birefringent.

This effect is used positively in some fibers, whereby the high birefringence causes the fibers to retain their polarization ([Ort04]). In the case of very great differences between the propagation constants of both polarizations they can then only very weakly interact with each other and exchange power. When only one polarization is launched into the fiber, then the power in this polarization is retained and is propagated in this way to the end of the fiber.

Fig. 2.249: High birefringent MPOF by incorporated asymmetry ([Issa04b])

The effect of birefringence can be generated in microstructured fibers in two ways: either the holes are arranged asymmetrically so that a geometric birefrin-


gence occurs which can be created in a controlled and thermally stable manner, or the holes are elliptical and not round which contributes to the birefringence ([Issa04b]). It is more difficult to control this kind of birefringence, but it does allow complete freedom of fiber design because the arrangement of the holes and their size can be freely chosen.

2.9.3.4 Highly Nonlinear Fibers

The nonlinear characteristics of fibers are influenced on the one hand by the nonlinearity of the material and on the other by the level of confinement which is described by the so-called effective mode area. With very strong wave guiding, light is guided into the center of the core and the optical power can propagate in the area of the core-cladding interface layer or even in the cladding. Here the light is strongly concentrated in a small area of the core which results in very high intensities with the same power which can lead to nonlinear behavior within the fiber. Such strong wave guiding can only be attained by means of big differences in the refractive index between the core and cladding.

In conventional fibers the differences in refractive index are in the range of a few percentage points. Microstructured fibers on the other hand consist of areas of glass with a refractive index of about nglass 1.5 and holes, which are generally of air (nair 1). The very high confinement can be achieved by this very high contrast in refractive index.

In fibers based on this effective refractive index, the cladding has to have a very high air fraction. The proportion of air in relation to the entire volume has to be so high that the effective refractive index lies near the value for air. Fibers with effec-tive area up to Aeff 2.85 μm2 have been realized using this process ([Lee02]).

In addition, other materials such as Bi2O3 can be used which have a highly non-linear susceptibility 3. With such materials nonlinear parameters of = 1100 W-1km-1 can be produced ([Lee06c]).

2.9.3.5 Control of the Effective Area

Fibers with a particularly high nonlinearity are needed for all optical signal pro-cessing. There are, however, a number of applications in which the nonlinear effects should be particularly weak so that the light propagation in such fibers is not disrupted. In such fibers the opposite path is taken as with highly nonlinear fibers: the material used should be as slightly nonlinear as possible and the effec-tive area of the fiber should be as large as possible so that the intensity within the fiber remains low at the given luminous efficiency. Even if the difference in refractive index between the core material and the holes continues to remain large you can still see to it through skillful fiber design that the light is guided relatively weakly and the mode field takes up as large an area as possible.

In general, these fibers have a very low air-fill factor so that the effective refrac-tive index in the cladding area lies only slightly below that of the core. Fibers with effective areas of Aeff 100 μm2 have been presented by [Kim06c] and [Sai06].


This technique can also be used for controlling the form of the mode field in order to adapt it to other types of fibers and thus minimize coupling losses at the connector. For example, Furukawa introduced such fibers at the ECOC in 2004 ([Guan04]) the mode fields of which are adapted to standard singlemode fibers.

2.9.3.6 Filters

Microstructured fibers can show very strong wavelength-dependent effects. As described above, the dispersion can be adapted to a wide area, but other wave-length-dependent characteristics can be specifically designed, e.g. group velocity or even the attenuation per unit length of the fiber.

Fibers with an effective refractive index permit the relatively simple adaptation of the group velocity with which one can generate all-pass filters with specific phase responses.

Fibers based on a photonic band gap can have very sharply delimited wave-length ranges with which light is guided. Thus, filters with specific amplitude response and sharp edges can be produced ([Vill03], [Kim05c], [Kim06d] and [Sai05]).

2.9.3.7 Sensor Technology, Tunable Elements

The characteristics of microstructured fibers can be manipulated in many ways. In particular materials can be introduced into the holes along the fiber which can change the characteristics of the microstructured fiber through their different re-fractive indices. These materials can be gases or liquids which are guided through the fiber and can alter the characteristics when the composition is changed ([Car06b]).

With such methods you can also analyze liquids such as blood in the human body. Polymer fibers are especially attractive for this kind of application because glass can split and would thus be considered too dangerous in the human body.

You can also intentionally change the characteristics by means of the controlled introduction of liquids. Thus, sensors have been introduced which are based exact-ly on this phenomenon, e.g. a liquid is pushed into the capillaries in the cladding area when the temperature rises, thereby changing the propagation characteristics of the fiber ([Jen05]).

Consequently, the dispersion ([Gun06]) or the band gap ([Sun06]) can be adjusted to a lesser or greater extent by introducing liquids.

Pressure sensors represent another application. Since the geometry of the holes has a great influence on the fiber’s propagation characteristics, lateral pressures can have a very noticeable effect on its behavior ([Eij03b]).

Especially fibers based on a photonic band gap react very sensitively to changes in the geometry. As a result, microstructured fibers can be produced which work like filters, the passband of which is changed when pressure is exerted.


2.9.3.8 Double-Core and Multi-Core Fibers

Most microstructured fibers consist of a cladding area in which the holes are arranged symmetrically or asymmetrically. In fibers based on an effective refrac-tive index the core consists of an area in which a hole has been left out of the arrangement. The core is thus a kind of imperfection within the photonic crystal. In this way two or more cores can be produced by introducing two or more imper-fections within the cladding area in which the light can propagate instead of having just one hole in the middle. Each individual location where a hole has been left out and the core material exists can be viewed as a separate fiber in which light can be propagated. If the individual cores are placed far enough apart, they either do not influence each other at all or only slightly.

Such fibers with several cores can be used for parallel data transmission ([Eij06a]). The arrangement of the individual cores is retained and so these fibers can be used like a well-ordered fiber bundle. However, these fibers have a consi-derably smaller diameter and can be laid like individual fibers ([Eij03b] and [Pad04]).

Fig. 2.250: Double core-MPOF with 9.6 μm spacing between the cores ([Eij03b])

2.9.3.9 Imaging

As we have seen above, microstructured fibers can be produced with more than one core for parallel data transmission. If you continue to increase the number of cores, you can use the same method to produce image guides in which every indi-vidual core transmits a part of the image (a pixel). As mentioned above, the arrangement of the holes stays the same and the cores along the fiber are retained. Each individual pixel reaches the end of the fiber in its definite position so that the image is retained ([Eij04c]).


Fig. 2.251: Image guide-MPOF ([Eij04c])

2.9.3.10 Multimode Graded Index Fibers

The fibers introduced so far are relatively thin singlemode fibers. In addition to these fibers, graded index multimode fibers made of polymer have also been deve-loped, so-called GI-MPOF ([Kle03b] and [Eij04d]), which have the large core dia-meter of a polymer fiber and the effective graded index profile of a multimode glass fiber (see Fig. 2.252).

Fig. 2.252: Schematic cross section of a GI-MPOF ([Kle04b] and [Lwin06])


Polymer fibers offer a number of advantages, especially with fibers having large core diameters compared to glass fibers (these advantages are also valid for other types of fibers). These considerably larger core diameters are possible with-out the fiber becoming inflexible. For this reason graded index polymer fibers have been produced for some years now which attain core diameters into the milli-meter range. However, these fibers have a refractive index profile in the core which has been adjusted through doping and they are quite difficult to produce when the core diameters are very large. Another advantage of these microstruc-tured fibers is the lack of doping material which results in these fibers having very good thermal and aging stability of the profile. Graded index profile polymer fibers which already exist are not particularly thermo-stable. With aging and especially in combination with increased temperatures they exhibit a flattening of the profile through diffusion of the doping material. This leads to an alignment of the concentrations of the doping materials resulting in a leveling out of the profile.

Fig. 2.253: Cross section of a graded index profile multimode polymer fiber (GI-MPOF) with 135 μm core- and 520 μm outer diameter ([Eij04d]) and of a MPOF according to [Lwin06]

Figure 2.253 shows a multimode fiber in which the effective refractive index continuously decreases with increasing distance to the fiber axis. If you take an average of the entire circumference of the refractive index, then you have a para-bolic refractive index profile in the radius direction. Measurements have shown that these fibers have a similar propagation behavior as a conventional multimode fiber. However, the differences lie in the detail. If you stimulate the GI-MPOF with a small spot, for example, the fiber behaves differently, depending on whether or not the light hits a hole or the core material; something that cannot happen in conventional fibers. For this reason greater research and development in measurement techniques and characterization are necessary before the GI-MPOF is widely used in commercial applications.

Date post:	04-Apr-2015
Category:	Documents
Upload:	binarystars
View:	552 times
Download:	5 times

Fundamentals of Optical Fibre

Documents