980-nm pump lasers for fiber optics applications pass field trial
980-nm pump lasers for fiber optics applications pass field trial
Controversy has intensified about options between long- and short-wavelength pump lasers for erbium-doped fiber amplifiers (EDFA) in fiber optics communications applications. These amplifiers require either a 1480- or 980-nanometer diode-pumped laser. Demand for pump lasers is increasing. For example, approximately 14,000 EDFAs may be sold in 1995, according to Jerry Hobbs, an industry analyst with Newport, RI-based Kessler Marketing Intelligence Corp.
Hobbs explains that the main strength of long-wavelength diode-pumped lasers is proven reliability. Short-wavelength diode-pumped lasers have been hard-pressed to demonstrate similar reliability. This deficiency results in part because of their more recent introduction. However, recently released field data shows a low failure rate of 1500 failures per 1 billion hours for short-wavelength 980-nm devices.
Diode-pumped lasers for EDFAs have logged more than 6.5 million device hours in a long-distance telecommunications carrier`s link between Chicago and Salt Lake City. All the amplifiers along the link are pumped by 980-nm diode laser modules from Lasertron Inc., Burlington, MA. These pumps include an E2 laser chip developed and currently produced by IBM Zurich in Switzerland. The modules are part of EDFAs made by Pirelli Cables North America, Lexington, SC.
IBM`s E2 chip is significant because it reduces the incidence of mirror damage in the diode laser. A short-wavelength diode laser chip tends to fail because of catastrophic damage to the back edge of the chip, which acts as a mirror for the laser. Light is absorbed by the silicon layers in the back facet, which causes localized overheating. This leads to mechanical stress and eventually mechanical damage to the chip. IBM developed a proprietary coating for its single-quantum-well ridge-type laser diode that reduces incidence of the mirror damage. Instead of using silicon as a layer in the laser-cavity mirror, the IBM chip uses titanium dioxide.
Lasertron packages the chips into modules that include a hermetic seal and a fiber pigtail. The company also offers modules that include two pumps per package for better reliability, or one laser with a large-capacity thermo-electric cooler that is more resistant to temperature changes than most modules.
Although the company still buys the chips from IBM, it has also licensed the right to make the chips. Lasertron`s facility for making the chips is expected to be operational by the end of this year.
In the reliability tests, the Chicago to Salt Lake City link contained more than 1200 amplifiers, although the study considers only the 576 diode-pumped lasers installed in November 1993. Until three years ago, only 1480-nm diode lasers were commercially available for pumping the amplifiers.
According to Hobbs, although the 1480-nm diodes are a more-mature technology with proven reliability, the 980-nm diodes couple light into the fiber more efficiently. This results in more power being transferred to the signal from the amplifier, allowing longer spans between amplifiers.
John Hill, marketing services manager at Lasertron, agrees with Hobbs. He says, "In our view, 980-nm diodes for land-based applications have become more accepted by fiber-optic communications users over 1480-nm diodes because 980-nm devices can transmit over longer distances."
Kathy Yanushefski, technical product support manager for pump lasers and EDFAs at AT&T Microelectronics in Brenigsville, PA, agrees that the failure-rate number has significance for the 980-nm devices. (AT&T sells both 980- and 1480-nm pumps for amplifiers. )
She suggests a slightly different outlook on the usage for the pumps. Different technological advantages of the separate wavelength devices indicate which should be used for a specific application, according to Yanushefski.
She notes that amplifiers pumped by 980-nm diode lasers provide better signal-to-noise ratios than the amplifiers pumped by the longer-wavelength lasers. Thus, 980-nm lasers are better suited for amplifying the signal along the fiber and for amplifying the signal immediately before a detector. The 1480-nm pump is better suited for amplifying the signal before it is sent along the fiber, adds Yanushefski.
Laboratory testing has been performed on the 980-nm pumps, but field testing is needed. The 1500 failure-rate number is the first field substantiation of the long-term reliability of the short-wavelength pump laser.
The 576 lasers operating since November 1993 had accumulated 6.5 million device hours when the failure-rate number result was calculated. According to Hill, by late August 1995, the group had accrued approximately 10 million device hours. The only detected failure mechanism is mirror damage in the E2 laser chip. The rate of failures because of mirror damage is less than 0.1% per kilohour (at powers ranging from 150 to 200 milliwatts and temperatures ranging from 30 to 75C), and it does not decrease with time. The failure rate is partially dependent on power and temperature.
Yanushefski was unfamiliar with the Lasertron installation and data-taking criteria, but agreed that the failure rate was reasonable. "Based on our extensive testing and reliability numbers with our partner," she says, "our numbers are comparable to Lasertron`s." Because AT&T Microelectronics does not buy laser chips from IBM, it can provide a second source for customers.
Hill says, "The economics of 980-nm lasers look great, but the question mark for a long time has been the reliability issue." The new 1500 failure-rate number, however, is reasonable for a component in many terrestrial fiber links. Also, the ability to transmit farther is an advantage that may offset the higher price of the 980-nm pumps.
According to Hobbs, "The cost of one thousand 980-nm pumps for EDFAs is approximately $5800, whereas the cost of one thousand 1480-nm pumps is $4000."
In addition to field testing their pump lasers, Lasertron is undergoing corporate changes. Last August, the company planned to be acquired by Waltham, MA-based Oak Industries, which includes business sections specializing in components for cable-TV and telecommunications systems. The purchase price is $112 million with the acquisition scheduled to be completed by the end of the third quarter (see page 3).
Commenting on how significant Lasertron`s 980-nm pump lasers are to Oak Industries, Williams S. Antle III, president and chief executive of Oak Industries, says, "Lasertron has established its line of 980-nm pump lasers used in optical amplifiers. These pump lasers are rapidly being deployed for the next generation of long-distance telephone transmission systems.
"Growth in this field is fueled by increases in data transmission rates, as well as performance advantages over traditional signal regeneration technology." He adds, "It is also expected that 980-nm optical products may be deployed in local access and undersea telephone network applications."
Antle also notes that Lasertron is pursuing growth opportunities for fiber-optic components in the cable-TV and wireless communications markets. Optical amplifier technology is gaining acceptance in cable-TV hybrid fiber/coaxial-cable networks for head-end consolidation and long-distance trunk lines.
He says demand for Lasertron`s products will continue to grow as these networks are upgraded to provide interactive capabilities. "Fiber-optic links are expected to be deployed in cellular and personal communication services applications to extend microcell coverage in areas with poor reception and in dense urban areas with high-capacity requirements."
By purchasing Lasertron, Oak Industries also expects to acquire a growth presence in the People`s Republic of China. This presence results from Lasertron having a 50% interest in a manufacturing joint venture formed in 1989 with Wuhan Optical Communications Technology Co. of China`s Ministry of Posts and Telecommunications.
This joint venture, Wuhan Telecommunications Devices Co., is located in Wuhan, China. The Ministry was an early investor in Lasertron and currently owns approximately 5% of the company. q
Yvonne Carts-Powell writes from Belmont, MA.