Diode lasers are a natural fit for telecommunications applications due to their small physical size and narrow linewidth, which enables worldwide data transmission. That said, a growing number of telecom system architectures are moving steadily toward increasing the number of signals carried on a single fiber. Each of those signals is generated by a single laser of a unique wavelength. Those single lasers can all be unique, fixed wavelength devices; or the lasers can be identical, tunable wavelength devices operated at different wavelengths.
One of many, or many of one?
|Bookham’s InP device has two variable gratings, a phase adjusting and gain region, and a semiconductor optical amplifier all on one chip. The 39-nm tuning range means a single part can cover an entire telecommunications band.|
Today, sophisticated components combine and separate multiple wavelengths for transmission through a single fiber, in a technique called wavelength division multiplexing (WDM). The International Telecommunications Union, a standards body associated with the United Nations, has established a dense WDM wavelength grid with telecommunications channels separated by only 50 GHz, or about 0.5 nm within each of three wavelength bands. The L-band covers from 1570 to 1612 nm, while the C-band runs from 1530 to 1570 nm. The relatively unused S-band spans from 1492 to 1530 nm. But a hundred different wavelength sources make for an expensive and inflexible system. Tunable lasers seem to be the perfect answer; sources that have the desirable spectral and propagation characteristics of fixed wavelength lasers, but have a selectable output wavelength.
In fact, at this year’s Optical Fiber Conference, held in Anaheim, Calif., in March, it became clear that tunable lasers are now pragmatic solutions for telecommunications systems.
More capability, less cost
The telecommunications network links one user to another by routing signals from one node to the next, creating a chain of links. But if each fiber carries 100 distinct wavelengths, that means that each node must have 100 different lasers, and, even worse, spares for each of those 100 distinct laser sources. It’s a logistics nightmare. If a single interchangeable part provides any of the 100 wavelengths, then a reasonable number of spares can be kept on hand at each network service node. In addition to reducing the expense of spares, tunable lasers also make it easier for service providers to modify their networks in response to changing demand using a method called reconfigurable optical add/drop multiplexing.
For service providers to take advantage of the flexibility of tunable lasers, the devices must meet requirements well beyond tunability. Without reliability, maintenance becomes expensive, and the advantages of tunability disappear. The devices must be remotely reconfigurable, and stable at a given wavelength. Another concern is output power variability. Communications links have amplifiers throughout the system and receivers that are optimized for a given power level. Changing the wavelength cannot change the output power too much; too high and the link will be degraded due to nonlinear effects, too low and detection is poor.
The all-in-one approach
|Monolithic integration of Bookham's tunable laser simplifies higher-level assembly. Further system simplification is possible with the addition of an InP modulator.|
Every tunable laser needs to combine components capable of covering a wide wavelength range with some mechanism for selecting just a narrow wavelength window. Theoretically, tunability is as simple as combining a laser gain medium with a variable length cavity. The cavity has resonant modes that are defined by its optical path length so the output is defined by the overlap of the laser gain with the cavity modes, usually resulting in simultaneous lasing in many neighboring modes. For fixed wavelength diode lasers, the most common method of narrowing bandwidth is to incorporate a grating structure along the waveguide, which sets the peak reflectance and defines the laser wavelength.
Bookham, based in San Jose, Calif., is offering a tunable laser that sets the emission wavelength using that same grating method. Their laser is a monolithic device, with a gain section, two distributed Bragg reflectors (DBRs), a phase-modifying section, and a semiconductor optical amplifier (SOA) all included on a 2-mm-long chip. The rear DBR is a relatively long grating, which reflects in a series of narrow-width modes. The index of refraction of the whole region is modified with current injection, which has the effect of shifting the comb-like reflectance peaks. The phase section is a region whose optical path length varies with injected current, which serves as a fine control over the output linewidth. The gain section generates and amplifies photons, as in any laser. The next section is a short Bragg grating with many contacts so current can be injected in different patterns. The pattern of current injection activates certain regions of the grating structure, tuning over a wide wavelength range. Only photons that make it through both the front and rear gratings get amplified. The final SOA section allows the optical power to be balanced without affecting laser tuning.
The device puts out 13.5 dBm in each of the 80 channels at which it operates, covering a 40-nm tunability range.
The external option
Pirelli, of Milan, Italy, is one of the suppliers offering an external cavity laser (ECL). ECLs typically have a gain chip with one facet covered with a high reflectance coating, an intracavity etalon to define a set of evenly spaced, narrow modes, and a wavelength selective mechanism.
Pirelli’s wavelength selector is a tunable liquid crystal on silicon (LCOS) mirror. Liquid crystal is layered on a silicon support inscribed with a periodic structure. The liquid crystal is covered with a glass window and sealed. When voltage is applied to the liquid crystal, its optical path length changes, changing the peak reflection wavelength. The relatively wide reflectance peak of the LCOS mirror only overlaps one of the etalon modes; so as the voltage is scanned the laser output switches from one C-band channel to another in the 35-nm tuning range. Luis Tondi Resta, director of engineering for Pirelli Broadband Solutions N.A., adds, “Independent management of LCOS voltage and laser drive current provides all the necessary control.”
Paxera, a Sunnyvale, Calif., supplier, also offers an ECL. Their wavelength selector is an acousto-optic tunable filter (AOTF). An AOTF is a variable pitch grating. When an RF frequency is input to a transducer on the AOTF, an acoustic wave is established in the AOTF, with the index of refraction varying as a function of location relative to the peaks and valleys of the wave. Changing the frequency of the RF input varies the grating spacing, which modifies the wavelength reflected to the output coupling mirror.
AOTFs have been used to control propagation wavelength for many years, but traditional AO technology is too large and power hungry for telecom lasers. Paxera has a proprietary method for operating the AOTF in a different regime than conventional AO devices, allowing a ten times reduction in size and electrical power consumption. Ben Sitler, Paxera’s CEO, says “our device offers twice the performance at lower cost.” The twice the performance he’s talking about is a tunability range of more than 80 nm, covering both the C and L bands, with tuning speeds of less than 5 msec on any channel to any channel.
A range of solutions
The three approaches outlined here represent the growing maturity of the field. These are not paper studies or laboratory prototypes, but practical devices ready to be deployed.
— Richard Gaughan
Founder and Chief Engineer
Mountain Optical Systems Technology,