Polymer Optical Fiber

This is a continuation from the previous tutorial - Liquid-Core Optical Fibers

 

 

1. INTRODUCTION

This tutorial describes the polymer optical fiber \(\text{(POF)}\), probably one of the fiber types with the highest loss and the smallest bandwidth. Nevertheless, it is the only optical fiber that can be installed by everyone without any special tool. That is why the potential of \(\text{POF}\) systems is very high.

 

 

2. POF BASICS 

The first \(\text{POFs}\) were manufactured by DuPont as early as the late 1960s. Because of the incomplete purification of the source materials used, optical attenuation values remained in the vicinity of 1000 dB/km. During the 1970s, 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, glass fibers with losses significantly below 1 dB/km at 1300 nm/1550 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 computer networks was dominated by copper cables (either twisted-pair or coaxial) that were completely satisfactory for the typical data rates of up to 10 megabits per second (Mbps) commonly used then.

There was hardly any demand for an optical medium for high data rates and small distances, so the development of the \(\text{POF}\) was slowed down for many years. A significant indicator for this is that at the beginning of the 1990s, the company Hoechst stopped manufacturing polymer fibers. 

During the 1990s—after data communication for long-haul transmission had become completely digital—the development of digital systems for private users was started on a massive scale. In many areas of life, we are being increasingly confronted with digital end-user equipment.

The \(\text{CD}\) player has largely replaced analog sound carriers (vinyl records and cassettes). The \(\text{MP3}\) format is leading to a revolution in music recording and distribution. The \(\text{DVD}\) (Digital Versatile Disc) replaces the analog video recorder. Even today more digital television programs are available than analog programs. Decoder boxes have become standardized \(\text{(MPEG2}\) format) and will be integrated into television sets.

More and more households are using powerful \(\text{PC}\) and digital telephone connections \(\text{(ISDN)}\) or triple-play services. With offers such as \(\text{T}\)-\(\text{DSL}\) (\(\text{ADSL}\) and \(\text{DSL}\) technology provided by Deutsche Telekom \(\text{AG})\), as well as fast internet access via satellite or broadband digital services on the broadband cable network, private users were 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. \(\text{CD}\) changers, navigation systems, distance-keeping radar, and complex control 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—and 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. \(\text{POFs}\) can meet many of these requirements to an optimum degree and are, therefore, increasingly of interest.

 

Materials for POF

The majority of all used \(POFs\) are made from polymethylmethacrylate \(\text{(PMMA)}\) as the core material. Due to the Rayleigh scattering and the strong absorption of the \(\text{C}\)-\(\text{H}\)-bonds, the smallest attenuation is approximately 100 dB/km (Fig.1). 

The absorption peak at 620 nm with a typical loss of 440 dB/km is related to the sixth overtone of the \(\text{C}\)-\(\text{H}\)-bond vibration as an example. Loss minima of \(\text{PMMA}\) are at 520 nm, 570 nm, and 650 nm.

The only way to reduce the loss of the material is the substitution of hydrogen by heavier atoms, like fluorine. Figure 2 shows the molecule structure of \(\text{PMMA}\) and \(\text{CYTOP}\) (a completely fluorinated polymer by Asahi Glass Co.). 

The lowest loss ever reported for \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) is less than 10 dB/km (see Fig. 3 for the attenuation spectrum). 

Other options for polymer fibers are polycarbonate \(\text{(PC)}\) or elastomers. The main reason for the search for other materials is the limitation of the operating temperature of \(\text{PMMA}\)-based \(\text{POF}\). Most of the \(\text{POFs}\) available are

 

 

 

 

 

 

specified for a maximum operation temperature of \(+70\) to \(+85^\circ C\). The temperature range can be increased by cross-linking of the \(\text{PMMA}\) up to \(130^\circ C\).

Much higher temperatures (up to \(+170^\circ C\)) can be realized by using elastomers, but there are no commercial products available. 

 

Light Propagation Effects in POF 

Because of the large fiber diameter, the high numerical aperture \((NA)\), and the short operation wavelengths, the \(\text{POF}\) can guide more modes than every other kind of fiber.

The standard 1-mm \(\text{POF}\) with an \(\text{NA}\) of 0.50 owns about \(2^{1⁄2}\) million modes. On the other hand, the mode-dependent effects in \(\text{POF}\) are more important than in other fibers. The reasons are the high loss in the cladding material (several 10,000 dB/km) and the strong mode mixing in the polymer material and mainly at the core–cladding interface. 

The effect of the cladding material absorption can be seen clearly by measuring the mode-dependent loss. One result for a 1-mm \(\text{PMMA}\) \(\text{POF}\) is shown in Fig. 4, measured at 650-nm wavelength.

The consequence of the strong mode-dependent loss is that the far-field width under equilibrium mode distribution \(\text{(EMD)}\) conditions is much smaller than calculated from the \(\text{NA}\).

The measured bandwidth of most \(\text{POFs}\) is higher than determined by the \(\text{NA}\) (under Uniform Mode Distribution \(\text{[UMD]}\) assumption) and the bend loss is smaller. The influence of mode-dependent loss is shown in Fig. 5.

 

 

 

 

 

The second effect is the strong mode mixing (strongly dependent on the fiber type and the cable manufacturing). Figure 6 shows the length-dependent measured attenuation for a 1-mm \(\text{POF}\). The equilibrium state can be seen after a length of about 100 m. 

 

 

 

Bandwidth of POF

The parameter with the biggest influence of the mode-dependent effects is the bandwidth. Results of length and launch-dependent bandwidth for a 1-mm \(\text{PMMA}\) \(\text{POF}\), measured at 650 nm wavelength, are presented in Fig. 7. 

The differences of the measured bandwidth for short lengths are more than one order of magnitude and still a factor of 2 at 100 m. The same behavior can be seen for other multimode fibers as 200 \(\mu m\) plastic clad silica \(\text{(PCS)}\). An example is shown in Fig. 8.

 

 

3.TYPES OF POF

\(\text{POFs}\) are available with different index profiles. The aim of the modified profiles is an increased bandwidth and a reduced bending sensitivity. As was the case with silica glass fibers, the first \(\text{POFs}\) were pure step-index profile fibers \(\text{(SI}\) \(\text{POF)}\). This means that a simple optical cladding surrounds a homogenous core. For this reason, a protective material is always included in the cable. Figure 9 schematically represents the refractive index curve. 

Glass multimode fibers usually have an \(\text{NA}\) of approximately 0.20. Glass fibers with polymer cladding have an NA in the range of 0.30–0.50. The large refractive index difference between the materials that are used for the core and the cladding of polymer fibers allows significantly higher \(\text{NA}\) values.

Most of the 

 

 

 

 

 

initially produced \(\text{SI}\) \(\text{POFs}\) had an \(\text{NA}\) of 0.50. \(\text{SI}\) \(\text{POF}\) with an \(\text{NA}\) around this value is nowadays generally called standard \(\text{NA}\) \(\text{POF}\), or standard \(\text{POF}\) for short.

The bandwidth of such fibers is approximately 40 MHz for a 100-m long link (quoted as the bandwidth-length product 40 MHz \(\cdot\) 100 m). For many years, this was a completely satisfactory solution for most applications. However, when it became necessary to replace copper cables with \(\text{POF}\) to accomplish the transmission of \(\text{ATM}\) (i.e., asynchronous transfer mode) data rates of 155 Mbps over a distance of 50 m, a higher bandwidth was required for the \(\text{POF}\). In the mid-1990s, all three important manufacturers developed the so-called low-\(\text{NA}\) \(\text{POF}\). 

\(\text{POFs}\) with a reduced NA (low-\(\text{NA}\) \(\text{POFs}\)) feature a bandwidth increased to approximately 100 MHz/100 m because the NA has been reduced to approximately 0.30. The first low-\(\text{NA}\) \(\text{POF}\) was presented in 1995 by Mitsubishi Rayon. Usually the same core material as for \(\text{SI}\) \(\text{POF}\) is used, but the cladding material has an altered composition. 

 

 

Unfortunately, practical testing showed that although this fiber met the requirements of the \(\text{ATM}\) forum with respect to bandwidth, it did not meet the requirements with respect to bending sensitivity. These requirements specify that for a 50-m long \(\text{POF}\) link, the losses resulting from a maximum of ten 90-degree bends having a minimum bending radius of 25 mm should not exceed 0.5 dB. To meet both these requirements simultaneously, it became necessary to find a new structure. 

The double \(\text{SI}\) \(\text{(DSI)}\) \(\text{POF}\) features two claddings around the core, each with a decreasing refractive index (Fig.10). In the case of straight installed links, light conduction is achieved essentially through the total reflection at the boundary surface between the core and the inner cladding.

This index difference results in an \(\text{NA}\) of around 0.30, similar to the value of the original low-\(\text{NA}\) \(\text{POF}\).

When fibers are bent, part of the light will no longer be conducted by this inner boundary surface. However, it is possible to reflect back part of the decoupled light in the direction of the core at the second boundary surface between the inner and the outer cladding.

At further bends, this light can again be redirected so that it enters the area of acceptance 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. 11. All low-\(\text{NA}\) \(\text{POFs}\) offered today are \(\text{DSI}\) \(\text{POFs}\) in reality.

As described earlier, the requirements of high bandwidth and low sensitivity to bending are difficult to accomplish together within one fiber having a diameter of 1 mm. Fibers with a smaller core diameter can solve this problem because 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. As a compromise, Asahi developed a multicore fiber \(\text{(MC}\) \(\text{POF)}\).

In this fiber, many cores (19 to >200) are put together in production in such a way that they together fill a round cross-section of 1-mm diameter. Figure 12 shows the parameters for the percentage of covered area. 

 

 

For 37 cores with \(d_m=5\mu m\), the partition of core area is only 65.3% and the value is 51.7% for 217 cores. Practical experience shows that a better utilization of the area can be achieved. 

During the manufacturing process, the fibers are placed together at a higher temperature, which means that they change their shape and, thus, reduce the gaps between 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.

 

 

 

 

 

Figure 13.  shows the refractive index curve of a MC POF, shown as a cross-section through the diameter of the fiber. The index steps correspond to those of a standard \(\text{POF}\). 

Because the bandwidth only depends on the \(\text{NA}\) for \(\text{SI}\) fibers, it should be possible to measure values comparable to the standard \(\text{POF}\). However, the fact is that the measured values are actually significantly higher, which has been explained with the aforementioned mode-selective attenuation mechanisms.

In the \(\text{MC}\) \(\text{POF}\), too, an increase in bandwidth was achieved by reducing the index difference. Because of 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. 14. The principle is the same as in the \(\text{DSI}\) \(\text{POF}\) with an individual core. In this case, a bundle with single cladding is completely surrounded by a second cladding material (‘‘sea/islands’’ structure). 

The \(\text{MC}\) \(\text{POF}\) features a noticeably reduced sensitivity to bending and only insignificantly increased attenuation, as well as a significantly increased bandwidth compared to single-core fibers, possibility because of smaller \(\text{NAs}\). Whether these fibers can be produced at the same price is still an open question. Should this be possible, data rates of 1000 Mbps over 50 m can easily be achieved. Figure 15 shows a cross-section of different \(\text{MC}\) \(\text{POFs}\). 

When using graded-index \(\text{(GI)}\) profiles, an even greater bandwidth becomes possible. In these profiles, the refractive index continually diminishes (as a gradient), starting from the fiber axis and moving outwards to the cladding. Of particular interest are profiles that follow a power law.

 

 

 

 

\[\tag{1}\text{refractive}\;\text{index}\; n^2=n^2_{\text{fiber}\;\text{axis}}\cdot\left[1-\left(\frac{\text{distance}\;\text{to}\;\text{fiber}\;\text{axis}}{\text{core}\;\text{radius}}\right)^g\cdot\Delta\right]\] 

Parameter g is referred to as the index coefficient. When \(g=2\), we speak of a parabolic profile. The limiting case of step-index profile fibers is described by \(g=\infty\).

Parameter \(\Delta\) signifies the complete index difference between the fiber axis and the edge of the core. Figure 16 shows a parabolic index profile. 

Because of the continuously changing refractive index, the light beams in a \(\text{GI}\) fiber do not propagate in a straight line but are constantly refracted towards the fiber axis. Light beams 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 boundary surface.

This behavior is illustrated schematically in Fig. 17. The geometric path of the beams running on a parallel axis is still significantly smaller than the path of beams that are introduced 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 parameters, 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. However, it is possible to increase bandwidths by two to three orders of magnitude compared with the \(\text{SI}\) fiber.

 

 

 

 

When considering not only the pure mode dispersion but also chromatic dispersion (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 Y. Koike.

The significance of this effect is particularly pronounced. Because of the smaller chromatic dispersion of fluorinated polymer compared with silica, the bandwidth of \(\text{GI}\) \(\text{POF}\) theoretically achievable is significantly higher than that of multimode \(\text{GI}\) glass fibers. In particular, this bandwidth can be realized over a significantly greater range of wavelengths.

This makes the \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) 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 yet been provided.

Another factor involved in the bandwidth of \(\text{GI}\) \(\text{POF}\) is the high level of mode-dependent attenuation compared to glass fibers. In this case, modes with a large propagation angle are suppressed resulting in a greater bandwidth. An example is the simulation that was carried out by Yabre: The bandwidth of a 200-m long \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) increases from 1 GHz to more than 4 GHz, taking into account the attenuation of higher modes.

This was also shown in practical trials. Mode coupling is less significant for \(\text{GI}\) fibers than it is for SI fibers because the reflections at the core–cladding boundary do not occur.

Following the many technological problems experienced in the production of \(\text{GI}\) fibers having an optimum index profile that remains stable for the duration of its service life, an attempt was made to approach the desired characteristics with the multistep-index \(\text{(MSI)}\) profile fiber.

In this case, the core consists of many layers (e.g., four to seven) that approach the required parabolic curve in a series 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. 18. 

In this case, light beams do not propagate along continually curved paths as in the \(\text{GI}\) \(\text{POF}\), but on multiply diffracted paths as demonstrated in Fig. 19. However, given a sufficient number of steps, the difference to the ideal \(\text{GI}\) profile is relatively small so that large bandwidths can nevertheless be achieved. \(\text{MSI}\)

 

 

 POFs were presented in 1999 by a Russian institute.

The manufacturing of \(\text{PMMA}\)-based \(\text{POF}\) was very difficult over a number of years. A number of groups tried to generate the \(\text{GI}\) profile by dopants. The disadvantage is that the doping reduces the glass-transition temperature of the core. This leads to diffusion of the dopant molecules at higher temperature, ending with a destroyed index profile.

A number of products have been announced over the last years but had never become available fibers. Table 1 gives an overview of published \(\text{GI}\) and \(\text{MSI}\) \(\text{POFs}\).

The most interesting development in the field of \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) was made by Optimedia in South Korea in the last years. The manufacturing process is shown in Fig. 20.

A rotating PMMA tube is filled with liquid monomers. The composition of the reactants can be changed continuously or stepwise. The polymerization process is induced by heat and/or \(\text{UV}\) radiation. The result is a preform with an MSI or parabolic profile. Because of the use of co-polymers instead of dopants, temperature stability is comparable to \(\text{PMMA}\) fibers.

The measured refractive index profile is shown in Fig. 21. The profile is very close to the ideal parabolic shape. At present, two fibers with \(900\mu m/1000\mu m\) or \(500mm/750\mu m\) core/outer diameter are available.

 

 

 

Table 1. Parameters of MSI- and GI-POF 

 

 

 

 

 

We have measured the long-term stability of the 1-mm \(\text{GI}\) \(\text{POF}\). After 5000 hours of aging at \(80^\circ C\) (dry atmosphere), no degradation of the bandwidth could be observed, indicating a stable index profile. The maximum transmitted bit rate was 2 Gbps over 100 m using a 655-nm edge emitting laser diode and a 800-mm \(\text{Si}\)-pin photo diode. The loss spectrum of the \(\text{OM}\)-\(\text{Giga}\) is shown in Fig. 22 (www.fiberfin.com). 

 

 

 

4.  POF STANDARDS 

The different kinds of \(\text{POF}\) are specified in the \(\text{IEC}\) 60793 as the fiber classes A4. Table 20.2 lists the main parameters of the different categories (‘‘\(\text{IEC}\) 60793- 2-40 Ed. 2.0: Optical Fibres; Part 2-40: Product specifications—Sectional specification for category A4 multimode fibres’’). 

• The classes A4a–A4c describe \(\text{SI}\)-\(\text{PMMA}\) \(\text{POFs}\) as used is mobile networks, home applications, and automation.
• The fiber class \(\text{A4d}\) is a \(\text{DSI}\) fiber for Fast Ethernet and \(\text{IEEE}\)1394 applications.
• \(\text{GI}\) and \(\text{MSI}\) fibers mainly for high speed home networks are categorized in class \(\text{A4e}\).
• The final classes \(\text{A4f}\)–\(\text{A4h}\) are \(\text{GI}\) \(\text{POFs}\) made of perfluorinated materials for building backbones and local area networks with data rates up to 10 Gbps.

Table 2. Parameters of different POFs 

 

 

5.  POF TRANSMISSION SYSTEMS 

The following sections present some selected \(\text{POF}\) transmission systems with different fibers. A more detailed description can be found elsewhere (second edition planned for end of 2006). The following examples are the best values for bit rate and/or transmission distance for several fibers. 

 

SI-PMMA POF 

The typical bandwidth of a \(\text{PMMA}\) \(\text{POF}\) with a standard \(\text{NA}\) of 0.50 is about 40 MHz/100 m. Therefore, the maximum bit rate of a 100-m link should be around 100 Mbps, but there are a number of options for higher capacity. The mode dispersion can be dramatically reduced by low \(\text{NA}\) launch and detection. Postcompensation and precompensation by high-pass filtering can give further improvements. Finally, if there is sufficient signal-to-noise ratio \(\text{(SNR)}\), some penalty can be accepted. 

High data rate transmission experiments on standard \(\text{SI}\) \(\text{POFs}\) were introduced in a series of publications spanning the years 1992–1994. With data rates of 265 and 531 Mbps (1994), 100-m \(\text{POF}\) was covered. Figure 23 illustrates the principle of the test setup.

The Mitsubishi \(\text{ESKA}\) \(\text{EXTRA}\) \(\text{EH4001}\) was used as the fiber medium. It has 139 dB/km of attenuation at 652 nm. A Philips laser diode \(\text{CQL82}\) with a wavelength of 652 nm served as the light source.

The laser was operated at \(290\;\text{K}\;(17^\circ C)\) with 36-mA bias current. To increase the bit rate, a first order high-pass filter was preconnected as the peaking filter. With the help of input optics, 2:7\(mP_{p-p}\) of power was achieved at launch of \(\text{NA}=0:11\). During

 

 

 

modulation, the average power was  1.7 dBm (0.68 mW); with the peaking filter, the average power fell to  6.7 dBm (0.21 mW). An \(\text{AEG}\)-Telefunken \(\text{BPW89}\) photodiode with 4.9-pF capacity at 20 V of reverse voltage was used as a receiver. The responsivity is 0.4 \(\text{A/W}\) at 650 nm (76% external efficiency).

The coupling to the \(\text{POF}\) is done with a ball lens. A second high-pass filter was connected behind the receiver as a compensation filter for the mode dispersion. The receiver achieved \(-22.1\) dBm sensitivity at \(\text{BER}=10^{-9}\). As a result, a data rate of 265 Mbps was achieved.

The newest result is the transmission of 580 Mbps over 100 m of standard \(\text{POF}\) (Mitsubishi MH4001) at the \(\text{POF}\)-\(\text{AC}\) in 2006. We used a 650-nm laser with \(+6\) \(\text{dBm}\) optical power, an 800-\(\mu m\) \(\text{Si}\)-\(\text{pin}\)-\(\text{PD}\), directly coupling at the receiver and transmitter side, and a passive compensation filter behind the receiver.

 

PMMA-GI POF

The highest bit rate ever transmitted over \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) is described elsewhere. The transmission wavelength was 645 nm (5-mW optical power), the fiber diameter was about 500 \(\mu m\), and the receiver based on an \(\text{Si}\)-\(\text{APD}\) with \(-29\) \(\text{dBm}\) sensitivity (Fig. 24). 

The \(\text{POF}\)-\(\text{AC}\) has demonstrated the transmission of 2 Gbps over 100 m of a 1-mm \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) \(\text{(OM}\)-\(\text{Giga}\) of Optimedia) in 2005, using a 650-nm laser diode and an \(\text{Si}\)-\(\text{pin}\)-\(\text{PD}\) receiver.

The transmission of the complete coaxial cable \(\text{TV}\) signal (862 MHz) over 50 m of \(\text{OM}\)-\(\text{Giga}\) \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) has been demonstrated by the Fraunhofer Institute

 

PF-GI POF

The perfluorinated material \(\text{CYTOP}\) (made by Asahi Glass) offers the lowest attenuation of all \(\text{POFs}\). Because of the parabolic index profile and the low chromatic dispersion, the bandwidth of \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) can be very high.

In contrast to the \(\text{SiO}_2\) \(\text{GI}\)-\(\text{GOF}\), the bandwidth is high over a wide wavelength range from 

 

 

600 to 1300 nm. Some of the best results ever reported with the \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) include the following: 

• 1.25 \(\text{Gbps}\) over 1006 m at 1300 nm
• 2.5 \(\text{Gbps}\) over 550 m at 1310 nm
• Three-channel \(\text{WDM}\) 2.5 \(\text{Gbps}\) over 200 m
• 12.5 \(\text{Gbps}\) over 100 m at 850 nm \(\text{(Nexans)}\)

The bit rate time’s length world record was realized by the group around D. Khoe in 1999. The transmission length was 550 m and the bit rate was 2.5 \(\text{Gbps}\). This was made possible by providing a 550-m \(\text{GI}\) \(\text{POF}\) piece with a core diameter of 170 \(\mu m\) without any connector (Fig. 25). 

Experiments with various sources were carried out. The measured attenuation for the wavelengths was as follows: 

• 110 dB/km at 650 nm \(\text{(LD}\) as source)
• 43.6 dB/km at 840 nm \(\text{(VCSEL}\) as source)
• 31 dB/km at 1310 nm \(\text{(LD}\) as source) 

An \(\text{Si}\)-\(\text{APD}\) with 230-mm diameter was used for the receiver at 840 nm. It reached  28.6 dBm sensitivity with a \(\text{BER}=10^{-9}\) , whereby a budget of 29.9 dB was available.

In addition, a 1310-nm \(\text{DFB}\) laser with a modulation bandwidth of 5 \(\text{GHz}\), a spectral width of 0.1 nm, and maximum 0.4 dBm of optical output power (1.1 mW) was used. The laser is a standard transmitter element for single-mode fiber systems and is equipped with a corresponding fiber pigtail for single-mode fiber systems.

The single-mode fiber was also used for direct coupling to the \(\text{GI}\) \(\text{POF}\) (<0.1-dB loss). With this method, only a small part of the mode field is excited, which increases the bandwidth considerably. The highest transmission length with \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\), ever reported, was 1006 m (Fig. 26). 

 

 

 

 

 

 

6.  APPLICATIONS OF POF 

There is a wide area of \(\text{POF}\) applications, including the following examples: 

• \(\text{POFs}\) are used in car networks for entertainment systems (Digital Domestic Bus \(\text{[D2B]}\), Media Oriented System Transport \(\text{[MOST]}\), \(\text{IDB}\) 1394 systems) and for control networks (Byteflight). The use of \(\text{PCS}\) will be specified in the next \(\text{MOST}\) generation.
• \(\text{POFs}\) and \(\text{PCS}\) are widely used in automation (field busses like Profinet, Sercos, and Fast Ethernet).
• A wide area of application exists for illumination and design (side emitting fibers, textiles).
• \(\text{POFs}\) can be used as sensors (e.g., pedestrian protection, demonstrated by Siemens \(\text{VDO}\) 2004).

 

POF in Automobile Networks 

DaimlerChrysler introduced the use of \(\text{POF}\) in automotive networks in 1998. The \(\text{D2B}\) was designed for the transmission of entertainment data with up to 5.6 Mbps. The next generation of optical car busses was the \(\text{MOST}\) system, developed under the participation of manufacturers like DaimlerChrysler, Audi, and \(\text{BMW}\). The bit rate of the \(\text{MOST}\) bus is about 25 Mbps, sufficient for \(\text{DVD}\) data as well. Some of the technical data of the \(\text{MOST}\) system are as follows 

• Transmitter: 650 nm \(\text{LED}\) or \(\text{RC}\)-\(\text{LED}\)
• Wavelength range (over temperature): 630–685 nm
• Spectral width: \(<30\) nm
• Fiber coupled power: \(-1.5\) to \(-10\) dBm
• Optical power budget: 16.5 dB 
• Temperature range: \(-40\) to \(+85^\circ\text{C}\) 
• Receiver dynamic range: 25 \(\text{dB}\)
• Fiber type: 980 \(\mu m\) \(\text{PMMA}\)-\(\text{SI}\) \(\text{POF}\), \(\text{NA}\) 0.50
• Fiber bandwidth; \(>30\) \(\text{MHz}\)/100 m
• Minimum fiber bend radius: 25 mm

The architecture of the \(\text{MOST}\) system is a unidirectional active ring. Every device works as a repeater. The maximum link length between two devices is about 8 m. Power consumption is reduced by a sleep mode. The disadvantage of the architecture is the complete loss of operation if one device fails. The first \(\text{MOST}\) car was the \(\text{BMW}\) 7 series.

In September 2005, the number of cars with \(\text{MOST}\) equipment was increased to 34, representing 10 \(\text{Mio}\) devices per year. Cars in alphabetical order include the following: 

• Audi \(\text{A6}\), \(\text{A8}\), \(\text{Q7}\)
• \(\text{BWM}\) 1, 3, 5, 6, 7 Series
• Citroen \(\text{C8}\)
• DaimlerChrysler \(\text{A}\), \(\text{B}\), \(\text{C\), \(\text{CLS}\), \(\text{E}\), \(\text{M}\), \(\text{S}\), \(\text{SLK}\) classes
• Dodge \(\text{RAM}\)
• Lancia Phedra
• Landrover Discovery, Range Rover, Freelander
• Maybach
• Mitsubishi Colt
• Peugot 807
• Porsche Boxter, 911, Cayenne
• Rolls-Royce Phantom
• Saab 9-3
• Smart Forfour
• Volvo \(\text{S40}\), \(\text{V50}\), \(\text{XC90}\)

Active components for \(\text{MOST}\) and Byteflight (Infineon Technologies Regensburg, Germany) and a typical hybrid optical/electrical \(\text{MOST}\) connector (Komax, Switzerland) is shown in Fig. 27. 

 

 

 

POF Sensors 

Fiber optic sensors using \(\text{POF}\) have been described in a general overview. \(\text{POF}\) sensors are mostly restricted to multimode fibers, are of intrinsic and extrinsic type, and the change of intensity will be measured in most cases. 

In these application areas, additional benefits of \(\text{POFs}\) are exploited such as ease of handling, robustness, flexibility, and low weight. These new systems benefit from the positive experience gained with more than 4 million cars equipped with optical data bus systems using \(\text{MOST}\), \(\text{D2B}\), or Byteflight.

One possible sensor is the Pedestrian Protection System \(\text{(PPS)}\), developed and published by Siemens \(\text{VDO}\). Due to European Union directive \(\text{2003/102/}\text{EC}\), pedestrian protection has to be provided for every new car from 2007 on. This can be fulfilled by either of the following: 

• Passive means: structural measures such as ‘‘soft’’ front ends and sufficient deformation room between the hood and the engine .
• Active means: sensors that identify a pedestrian impact and then trigger protective means such as lifting the hood by means of actuators as shown in Fig. 28. 

 

 

The basic principle behind this approach is a cladding surface treatment of the fiber at discrete zones along the fiber. As shown in Fig. 29, bending the fiber in one direction leads to a better transmission, whereas bending in the other direction leads to a lower transmission, compared to the straight position. 

The spatial resolution is achieved by using several fiber strands in parallel and numerical signal analysis of the treated zones. Because of its principle, the sensor is able to distinguish between positive and negative bends, and because of its high bandwidth, the sensor can determine the impact with a high temporal resolution. This is necessary to distinguish not only between a human leg and a lamp-pole but also for identifying a collision with an animal. 

A different approach for the same and other applications has been published called Kinotex cavity sensor. The principle behind it is light scattering dependent on the compression of the scattering medium (e.g., rubber foam).

The transducer operates by detecting a change in energy intensity in and around an illuminated integrating cavity. Deformation of the integrating cavity by an external influence such as pressure results in a localized change to the illumination energy intensity.

This change is measured. The information can be used to measure the state of deformation. 

The Tactile Sensor (described in reference) uses the so-called ‘‘evanescent field.’’ Although it is a known fact from theory, it used to be often forgotten that 

 

 

the optical rays do not reflect exactly on the boundary between two optical media. In fact, the rays (electromagnetic field) break into the adjacent medium and do not suddenly drop to zero at the core–cladding boundary.

They decay instead exponentially within the cladding. The penetration depth \(d_p\) depends on the difference between the respective refractive indices of core and cladding material (the higher the difference, the lower \(d_p\)), the angle \(\varepsilon\) of incidence (the steeper  the higher \(d_p\)) and the wavelength \(\lambda\) of the incident light (the higher \(\lambda\), the higher \(d_p\)).

In ray optics, this phenomenon is known as ‘‘Goos-Hänchen- Shift,’’ indicating that the reentry of the ray into the core region is shifted along the geometrical core–cladding interface. 

This type of sensor gives a very high sensitivity without the necessity of actual physical contact and the first coming applications are envisaged as safety protection sensor for car windows or garage doors (to prevent shutting) or as tactile sensors for robot gripper tools, but also applications as described in the previous sections are conceivable.

 

POF in Home Networks 

The biggest application area is the use of \(\text{POF}\) in home networks as an extension of broadband access networks, first of all for \(\text{FTTB}\) systems. One possible example for a future network is shown in Fig. 30. The building is connected by a broadband access (at the beginning \(\text{ADSL2}+\), later WiMax, or glass fiber for higher capacity). The building network is based on 470-nm duplex 

 

 

\(\text{POF}\) transceivers. All apartments are equipped with 650-nm simplex transceivers with reduced reach but very simple installation. 

Wireless systems are installed in the rooms. The active node realizes the handover. Because of the small required radio power in the rooms, the interferences between the radio cells are very small and the full \(\text{WLAN}\) capacity can be used.

One of the major reasons for the simplicity of installation is the use of connector-less systems (Fig. 31). The customer can install all \(\text{POFs}\) without special tools in seconds. Connector-less solutions are developed by DieMount and Ratioplast/Infineon, for example. 

At this time (2006), only very few companies offer end-user devices and components with \(\text{POF}\) interfaces. That is why the customer requires media converters for the installation of \(\text{POF}\) links. Two examples of commercially available products are shown in Fig.32 \(\text{(PC}\) Card by DieMount and Fast Ethernet media converter by Ratioplast). With increasing number of installations, devices with direct \(\text{POF}\) interfaces will become available, first of all offered by broadband access suppliers. 

 

 

7. POF FABRICATION METHODS

Initially, \(\text{POF}\) fabrication methods focused on techniques for making \(\text{SI}\) fibers such as the preform and extrusion techniques. Later on, as the proliferation and 

 

 

 

 

When extrusion techniques are applied, the \(\text{POF}\) is produced in a continuous process directly from monomers. For \(\text{SI}\) \(\text{POF}\), this process is very simple. Figure 34 shows such an arrangement. 

In addition, two further processes are mentioned. In the thrust extrusion 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 

 

 

noncontinuous 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. 

 

Production of Graded-Index Profiles 

A number of processes for the manufacture of \(\text{GI}\) profiles are described in the technical literature: 

• Surface gel polymerization technique
• Centrifuging
• Photochemical 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. 

 

Interfacial Gel Polymerization Technique

The interfacial gel polymerization technique was developed by Y. Koike of Keio University. In this process, a tube is initially manufactured with \(\text{PMMA}\).

This tube is then filled with a mixture of two monomers M1 (high refractive index and large molecules) and \(M_2\) (smaller refractive index and smaller molecules). Initially, the inner wall of the PMMA tube is slightly liquefied in a stove that has been typically heated to \(80^\circ\text{C}\).

This results in a layer of gel and accelerates polymerization. The smaller molecule \(M_2\) can more easily diffuse into this layer of gel so that the concentration of \(M_2\) increases more and more towards the middle. The index profile is, thus, formed in accordance with the resulting concentration gradient. 

For manufacturing a \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\), Koike proposes that \(\text{MMA}\) \(\text{(M}_1)\) be supplemented with monomers \(\text{VB}\), \(\text{VPAc}\), \(\text{BzA}\), \(\text{PhMA}\), and \(\text{BzMA}\). The material that was finally used is \(\text{BzA}\) because its reactivity is comparable with that of \(\text{MMA}\). The 15- to 22-mm thick preform is then drawn at temperatures between 190 and \(280^\circ\text{C}\) to produce fibers ranging from 0.2 to 1.5 mm in diameter. Figure 35 illustrates the principle.

Koike et al. describes this method in more detail. The \(\text{PMMA}\) tube is produced by rotating a glass reactor at 3000 \(\text{min}^{-1}\) at \(70^\circ\text{C}\) that is partially filled with \(\text{MMA}\). The polymerization process for the core takes place at a speed of 50 \(\text{min}^{-1}\) and at a temperature of \(95^\circ\text{C}\) and requires approximately 24 hours to

 

 

be completed. Shi et al. describes the production of a \(\text{PMMA}\)-\(\text{GI}\) \(\text{POF}\) with \(\text{DPS}\) as dopants. For traditional materials such as \(\text{BB}\) or \(\text{BBP}\), one obtains fibers with an \(\text{NA}\) of 0.17–0.21, whereas with \(\text{DPS}\), an \(\text{NA}\) of 0.29 is possible. The greater \(\text{NA}\) improves the bending characteristics and makes the introduction of light easier. 

In principle, \(\text{GI}\) \(\text{POFs}\) 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-polymerization.

In the case of silica glass, the index variation can be easily achieved by replacing the silicon atoms with germanium because these two substances behave identically within the glass structure.

However, the components used for 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 refractive 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, although the sequence may be irregular, it is important that no long chains of one 

 

 

monomer are formed because otherwise the losses due to scattering increase considerably. This means that the bonding force of monomers among one another must not be greater than the bonding force to the respective other monomer. Of course, both monomers must have sufficient transparency. Figures 36. and 37. show schematic illustrations of the principles. 

For perfluorinated materials, the dopant process is used mainly. A lot of experimental fibers have been developed with doped \(\text{PMMA}\) as well. In theory, the dopant method allows lower loss compared with the co-polymerization (due to increased Rayleigh scattering).

Unfortunately, the reduction of the glass-transition temperature makes the fibers not usable for higher temperatures. The fiber with the best parameters on the market  is made by co-polymerization. The attenuation is increased by about 50 dB/km, but the temperature stability is quite good.

 

 

 

GI POF Extrusion 

A very new method for continuous production of \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) was developed by Chromis-Fiberoptics (formerly, OFS). The process in shown in Fig. 38. An \(\text{SI}\)-type \(\text{POF}\) is extruded conventionally.

The \(\text{GI}\) profile is made by diffusion inside a heated tube (wound around a big cylinder). The best results (loss of 20 dB/km at 1300 nm and about 10 GHz/100 m bandwidth) are close to the values of the \(\text{PF}\)-\(\text{GI}\) \(\text{POF}\) made from a preform.