Emerging 1310-nm amplifiers reopen a legacy window

Emerging 1310-nm amplifiers reopen a legacy window

With bandwidth at a premium, dwdm and cable-TV networks could benefit from adding 1310-nm transmission to the typical 1550-nm pathway.

By Michael Shimazu Molecular OptoElectronics Corp.

The erbium-doped fiber amplifier (edfa) has launched a revolution in fiber-optic systems, enabling such high-value applications as dense wavelength-division multiplexing (dwdm) and cable-TV supertrunking. The edfa also has forced a shift among telecommunications carriers and cable-TV operators from the 1310-nm spectrum, where most systems had been functioning, to the edfa gain band around 1550 nm. Recent developments in 1310-nm amplifiers now promise to reopen this window for advanced systems, with the benefits of increased fiber-optic capacity and enhanced performance.

edfa background

Standard singlemode fiber supports transmission at both 1310 and 1550 nm. The chromatic dispersion is at a minimum near 1310 nm and the attenuation is at a minimum near 1550 nm where the longer wavelengths are less susceptible to scattering. The graph in Figure 1 shows attenuation by wavelength and indicates that these two low-loss windows are separated by an O-H (oxygen-hydrogen bond) absorption peak in the 1400-nm region.

The first singlemode systems deployed in the 1980s used 1310-nm lasers to take advantage of their relative maturity versus those at 1550 nm and the larger bandwidth at 1310 nm. But with the advent of the edfa, new system designs gave carriers compelling economic reasons to shift to 1550 nm.

Amplifiers were first deployed in cable-TV trunking systems to consolidate headends. Instead of having to deploy a satellite downlink for every headend, a cable-TV multiple system operator could distribute the signals from a single downlink to several systems over optical fiber. The more extensive architectures of this type involved long distances and therefore optical amplification.

The benefits of this application include substantial cost savings from the reduction in the number of headends and the ability to locate satellite downlinks where real estate costs may be lower. These advantages far outweigh the additional costs incurred by having to contend with the distortion in analog am-vsb signals caused by the chromatic dispersion in the edfa window.

dwdm systems, recently adopted as a response to capacity constraints on long-haul backbones, represent another amplifer-enabled capability. The edfa simultaneously amplifies signals carried by several wavelengths within the gain band, thus allowing several optical channels to share the same fiber. The savings, not only in optoelectronic regenerators but also in avoided fiber construction, much more than justify the cost of the equipment.

The bandwidth limitation imposed by chromatic dispersion at 1550 nm has been addressed by the use of multiple channels. Thus, rather than a single OC-192 (10-Gbit/sec) channel, which would normally require dispersion compensation of some sort, four OC-48 channels, each on a separate wavelength, may be operated.

Benefits of 1310 nm

Notwithstanding these impressive accomplishments in the 1550-nm window, there remain substantial benefits to be gained by operating near the 1310-nm zero-dispersion point. While we compare these windows it`s important to note that 1550- and 1310-nm systems are not necessarily competitive; they can be deployed together on the same fiber.

The advantage of low dispersion can perhaps best be visualized in the context of digital systems. Chromatic dispersion is a property of the fiber that causes different wavelengths to propagate at different speeds. Over distance, some wavelength components of an optical pulse will outrun others, causing the "bit" to spread and eventually cause unacceptable bit errors. This problem is exacerbated at higher data rates, as these pulses are more closely spaced in time. So the lower the dispersion, the higher the bit-rate ceiling.

Consequently, operation at 1310 nm allows much higher bit rates per optical channel than at 1550 nm (see Fig. 2). Note that Figure 2 presumes a 40-GHz laser linewidth and 60-km link distance. Modern laser transmitters have far narrower linewidths and therefore can attain much higher data rates than indicated for the given distance. But the data rate ceiling at 1310 nm will still be orders of magnitude higher than at 1550 nm.

In analog systems such as am-vsb cable-TV systems, the chromatic dispersion at 1550 nm contributes to signal distortion and raises the noise floor. Conventional operation at 1310 nm, which is standard in the distribution portion of a hybrid fiber/coaxial-cable deployment, reduces this distortion, even though directly modulated distributed feedback lasers are used.

As chromatic dispersion at 1550 nm is a well-characterized property of singlemode fiber, compensation is possible. The use of narrow-line stabilized lasers, for example, mitigates some of the problems. Dispersion compensation modules employing negative dispersion fiber or Bragg gratings are being incorporated into OC-192 class equipment as well.

These solutions are not perfect, however. The narrow-line lasers can`t completely eliminate dispersion over long distances. And in high-power analog systems they can contribute to stimulated Brillouin scattering. Negative dispersion fiber can increase the attenuation of a link by tens of decibels and promotes nonlinear frequency mixing in multiwavelength systems. Bragg grating-based compensators have limited optical bandwidth and must be made quite large to cover the entire edfa band.

The advantages of operating in the region of intrinsically low dispersion is amply demonstrated by the growing sales of dispersion-shifted fiber, in which the zero-dispersion point is moved near the 1550-nm edfa band. But the vast majority of the installed fiber base is not of the dispersion-shifted type. Moreover, the standard singlemode fiber (in marketing parlance, "non-dispersion-shifted" fiber) with zero dispersion near 1310-nm still accounts for most fiber sold today.

1310-nm amplifiers and singlemode fiber

One alternative to using new dispersion-shifted fiber with edfas is to use new 1310-nm amplifiers with standard singlemode fiber, including the installed legacy fiber-optic infrastructure. A number of alternative approaches to such amplifiers are in various stages of commercial availability and development.

Semiconductor optical amplifiers are commercially available in the 1310-nm window. They are current-injected devices using the semiconductor as a gain medium and can be fabricated at a variety of wavelengths. However, the commercial versions of these amplifiers exhibit low saturated powers and high noise figures compared to edfas.

Praseodymium-doped fluoride fiber is the basis for the praseodymium-doped fiber amplifier (pdfa), which has been developed over several years at Nippon Telegraph & Telephone (NTT--Tokyo) and other laboratories and is now commercially available. A number of materials-related challenges have had to be overcome to make the pdfa feasible. The fluoride glass is brittle and water soluble, for example. Also, once a stable singlemode fiber is drawn from it, the flouride glass still must be spliced to standard silica fiber. The development of coating technologies and suitable tapers to enable splicing has given rise to today`s commercial units.

The remaining challenge with respect to pdfas is the pump laser, which ideally should be a high-power source around 1020 nm. Cladding-pumped ytterbium-glass fiber lasers offer high powers at wavelengths as short as 1030 nm, and along with developmental diode lasers, are candidate pump sources. All pump alternatives are currently far costlier than the 1480- and 980-nm pumps used by edfas.

Raman amplifiers use a seeded third-order nonlinear effect to give rise to optical amplification. In these units, a high-power pump excites Stokes-shifted wavelengths in standard silica optical fiber. These new Stokes lines excite still more lines, and so on, until the amplifier`s operating window is covered. The data signal seeds the nonlinear conversion. A pump at around 1240-nm, which is attainable from a Stokes-shifted 1064-nm Nd:yag (neodymium-doped yttrium-aluminum-garnet) laser, enables amplification in the 1310-nm window.

Raman amplifiers may be implemented at either 1310 or 1550 nm, and in fact they have been tested with the object of extending the edfa range within the 1550-nm window. Such amplifiers are nearing commercial introduction, with the advent of pump lasers having the necessary powers to excite the Raman scattering.

Waveguide devices using doped glasses and crystals are another class of developmental amplifier aimed at 1310 nm. These waveguides may be combined with a side-polished fiber architecture to form the amplifier gain block (see Fig. 3). The optical signal couples from the side-polished fiber into the overlay waveguide and back into the fiber. The gain material in the overlay can be either end-fire or top-down pumped.

In contrast to fiber amplifiers, this architecture enables the use of several different gain materials that cannot be easily made into fiber. A variety of neodymium-, praseodymium-, or dysprosium-doped hosts, among others, are feasible here. Of particular interest are neodymium-doped materials that offer high gains in the 1310-nm window. These materials are pumped with well-developed and inexpensive lasers in the 800-nm region. Furthermore, the spontaneous emission, which gives rise to noise, largely occurs out of band at around 1060 nm. In the side-polished fiber architecture, this noise source does not even couple back into the fiber.

Working together

In principle, the advanced dwdm and cable-TV trunking systems described earlier can be implemented in the 1310-nm window on standard singlemode fiber using such amplifiers. They would have the added benefits of higher data rates per channel and better carrier-to-noise performance that accrue from operation near the dispersion minimum. But because 1310-nm systems can share the same fiber with 1550-nm systems, the combination of these capabilities offers the most interesting possibilities.

One example is dual-window dwdm, in which a plurality of signal wavelengths at 1310 nm is combined with others at 1550 nm. A unidirectional link based on this concept is illustrated in Figure 4. On legacy singlemode fiber, the 1310-nm channels can each support data rates of 10 Gbits/sec and higher without dispersion compensation. Such a link must be carefully engineered. The 1310-nm channels must be separated from the 1550-nm channels before any dispersion compensation or amplification is performed. In addition, the 1310-nm amplifiers must offer higher gains than the 1550-nm blocks with comparable noise performance to ensure equivalent spacing on the line.

Yet the benefits of such a system are clear. Burgeoning data-communications traffic is forcing carriers and their suppliers to find new sources of capacity. The ability both to add new optical channels and in so doing to boost data rates makes 1310-nm functionality a potentially attractive en hancement to standard dwdm offerings using standard fiber.

In the cable-TV trunking application, operation in the 1310-nm window with suitable optical amplifiers would obviate the costs involved with conditioning a signal for low-distortion transmission at 1550 nm. But the combination of both windows may offer a smooth transition step from trunked am-vsb analog video to multiwavelength digital services for cable-TV multiple-system operators.

In this dual-window application, the analog signal can be carried in the 1310-nm window, for example, at 1319-nm where Nd:yag-based trunking lasers have been deployed. The other digital signals, which form the basis of the data services offered by cable companies, can be optically multiplexed in the 1550-nm window. The separation of the analog from the digital signals is important. At -3 to 0 dBm at the receiver, the standard analog signal is about 20 to 30 dB stronger than the typical digital signal. Passing all the digital and analog signals through the same amplifier would cause gain nonuniformity and possibly intolerable gain saturation-induced crosstalk.

There are also emerging applications that can benefit from the new 1310-nm amplifiers. For example, Gigabit Ethernet transmitters are available using Fabry-Perot lasers at 1310-nm, in accordance with the recently published ieee 802.3z standards. Using this new technology, gigabit metropolitan area optical networks may be configured using low-cost local area network-class equipment. And there is also the potential to use multiple channels in this window for higher data rates or to support multiple networks on the same fiber base.

Just as the edfa sparked new capabilities in optical networking, the emerging 1310-nm amplifiers promise to bring additional capacity and performance to fiber-optic systems, especially using the legacy fiber plant based on standard singlemode fiber. Instead of competing with 1550-nm systems, however, amplifier-enabled 1310-nm networks can provide complementary enhancements to existing applications and fulfill future needs for bandwidth and functionality. u

Michael Shimazu is marketing and business-development vice president at Molecular OptoElectronics Corp. (Watervliet, NY).