Dense WDM mandates advanced passive component requirements
Achieving the full potential of dense WDM networks calls for improvements in both the control of laser center wavelengths and in multiplexer and demultiplexer performance
dicon fiberoptics inc.
Increasing demand for bandwidth in installed fiber-optic backbones is driving the development of wavelength-division multiplexing, or WDM, technologies in which optical carriers at different wavelengths are combined in the same fiber. Generally, these technologies are used in applications where the cost to upgrade existing fibers with WDM components is less than the cost of either installing additional fibers or integrating time-division multiplexing, or TDM, transmission equipment (see Lightwave, December 1995, page 42).
The reliability of WDM for synchronous optical network, or Sonet, 2.5-gigabit-per-second OC-48 systems with four-wavelength channels has been field-demonstrated, providing a combined transmission rate of 10 Gbits/sec. Undersea and terrestrial fiber-optic systems with more than four channels at 10-Gbit/sec OC-192 and higher rates are under development and are expected to reach the market in late 1996.
Most of these new systems employ erbium-doped fiber amplifiers, or EDFAs, to take advantage of their high gain and bit-rate independence performance. In these systems, maximum combined transmission rates can be achieved by closely spacing several WDM channels (1 to 7 nm apart) so as to operate in the erbium gain window from 1530 to 1560 nm. Such systems are said to use dense WDM technology.
At the heart of dense WDM networks are the optically passive components that combine or decombine two or more optical carriers at the transmission system terminal equipment and within the EDFAs (for bidirectional systems). These devices are known as multiplexers or demultiplexers, depending on the direction that the signals pass through them. If only one carrier is being added or removed from the line, the devices are known as add/drop multiplexers, or ADMs. In systems with more than two channels--and if all channels are being added or removed simultaneously--the devices are referred to as multichannel multiplexers or demultiplexers.
In general, transmission system designers prefer to select carrier channels that are as densely spaced as possible to preserve the possibility of upgrading capacity with additional channels in the future and to minimize differences in the signal levels associated with the wavelength-dependent amplifier gain. Minimum channel spacings currently achievable are limited by the performance of transmission system lasers and multiplexing components.
Laser manufacturers are delivering devices with predetermined ranges for center wavelength. For bidirectional systems operating at 1533 and 1557 nm, the two laser types have a ۭ.5-nm variation corresponding to a channel passband of 7.0 nm and a nominal channel spacing of 24 nm. Commercial lasers with center wavelength variations less than ۫.0 nm are expected to be available soon, which corresponds to 2-nm passbands for 3.5- to 5.0-nm spaced systems.
Industry standard groups, such as the International Telecommunication Union, are calling for commercial lasers with center wavelength specifications tight enough to achieve 100- or 200-gigahertz spacings (0.8 or 1.6 nm); however, these parts are not expected to be available for some time.
On the multiplexer side, the primary technical challenges deal with achieving low losses in a narrow passband region and, at the same time, obtaining high isolation for both adjacent and nonadjacent channels. Losses in the passband region directly affect the power budget. Channel isolation affects the signal-to-noise ratio and bit-error rates at the terminal equipment receiver end and, therefore, limits the channel spacing.
Many component technologies, including dielectric thin-film coatings, bulk optical-fiber gratings, embedded-fiber gratings and planar waveguide phase arrays, have been used to meet these advanced goals. Thin-film technology appears best suited to meet the low-loss, high-isolation objectives for 3- to 24-nm wavelength-spaced systems. Such devices are currently available from a number of component manufacturers. For a more narrowly spaced system, the best approach is yet to be determined and, therefore, commercial availability of advanced parts is limited.
The performance requirements for passive components used in dense WDM systems are more stringent than those in the older broad WDM (1310- and 1550-nm) dual-wavelength networks. Consequently, several new parameters have evolved. The following parameter specifications are used to describe multiplexer performance in dense WDM networks. The definitions for some of these parameters are also covered in Bell Communications Research GR-2883-CORE, Issue 1.
Channel Passband: This parameter is the wavelength range for which a given multiplexer port has low losses, corresponding to the center wavelength variation for a particular model laser. In some systems, the multiplexer channel passband might be wider than the nominal laser center wavelength range to compensate for temperature and aging effects in both the laser and the multiplexer. Common passbands in current dense WDM systems range between 1.0 and 8.0 nm.
0.5-dB Bandwidth: This is the spectral width over which the difference between the peak transmitted power and transmitted power at every other point is less than 0.5 dB. By convention, the 0.5-dB bandwidth is nominally equal to the difference between the maximum and minimum wavelengths in each channel passband.
Passband Ripple: This parameter is the largest peak-to-peak variation in insertion loss over the channel passband. For systems in which the 0.5-dB bandwidth of each multiplexer and the channel passbands are the same, passband ripple is by convention equal to 0.5 dB maximum.
Maximum Insertion Loss over Passband: This worst-case insertion loss is measured over all wavelengths in the channel passband/0.5 dB-bandwidth region and over the full operating temperature range. Normally, its value is equal to the insertion loss at the peak power point plus 0.5 dB.
Center-wavelength Tolerance: Because the optical spectrum of each multiplexer port is generally displaced from the ideal center wavelength for each laser channel, center wavelength tolerance specifies the maximum allowable variation. Center-wavelength tolerance typically needs to be an order of magnitude smaller than the channel passbands, and, therefore, ranges between ۪.1 and ۪.5 nm.
Adjacent Channel Isolation: Channel isolation measures the degree to which the unwanted channels are attenuated on each add/drop channel and is directly related to the signal-to-noise ratio and the bit-error-rate degradation resulting from the presence of multiple optical carriers. Because the ideal spectral shape of each add or drop channel is similar to a Gaussian curve, the channel(s) closest to the add/drop channel has the worst isolation value. Adjacent channel isolation is, therefore, the worst-case isolation measured over all adjacent channel passband(s).
Channel Spacing: This distance is the wavelength space between the nominal center wavelengths of adjacent channels. It is a system parameter as opposed to a component specification.
30-dB Bandwidth: The parameter represents the width of the spectral range outside of which any signal is attenuated by more than 30 dB with respect to the peak power wavelength. The 30-dB bandwidth is a commonly used parameter because the adjacent-channel isolation specification for many systems is 30 dB. Note that 0.5-dB bandwidth, 30-dB bandwidth and channel-spacing parameters are related to each other by the following: The channel spacing value is greater than or equal to the 0.5-dB bandwidth value added to the 30-dB bandwidth value, with the added value divided by 2.
30-dB Figure of Merit, or FOM: The greatest challenge in multiplexer design is in achieving narrow passband regions with low loss while at the same time providing high isolation for closely spaced adjacent channels. The 30-dB FOM value is the ratio of the 0.5- and 30-dB bandwidths and is a measure of the ability of a given multiplexer to meet this objective.
The next three parameters characterize the stability of each multiplexer relative to system and environmental factors such as temperature and state of polarization. If these effects are large, then the multiplexer channel passband might have to be made wider than the center-wavelength variation of the lasers, and the channel spacing might have to be increased.
Polarization-dependent Loss: Because the spectral performance of each multiplexer port depends on the input polarization state, and the polarization state is indeterminate in laser transmission systems, this loss value measures the largest variation in insertion loss over the passband of each multiplexer port.
Center-wavelength Temperature Stability: The spectral performance of each multiplexer port also depends on temperature. Center-wavelength stability measures the maximum change in center wavelength of each port over the intended operating temperature range.
Thermal Stability: This parameter is the maximum variation of insertion loss with operating temperature measured over the passband of each port.
The need to factory-test these parameters has evolved with the advent of dense WDM systems and, because the measurements require high accuracy for power and wavelength, test systems for multiplexer characterization are complex, expensive and slow. These systems generally require the use of a state-of-the-art power meter, polarization controller, optical spectrum analyzer, wavemeter and broadband or tunable laser source in addition to custom software. The total test system cost can exceed $200,000. Test equipment manufacturers are expected to view dense WDM as an opportunity to develop new types of lightwave multimeters, which focus on multiplexing applications, and offer lower-cost, higher-speed and easy-to-use instruments. u
David Polinsky is vice president of commercial relations at Dicon Fiberoptics Inc., Berkeley, CA.