In recent years, news about communications networks technology always seems to involve some pronouncement on the urgent need for more bandwidth. The facts still bear repeating: an ever-growing number of people with telephones, faxes, modems, and computers are, through the exchange of terabytes of digitized information-videos, images, modeling procedures, as well as data and voice-demanding a larger share of the carrier spectrum. In response, high-tech communications companies who thrive on growth are competing to feed this appetite for bandwidth. Over the past decade, major resources have gone into developing fiber-optic networks in which light waves transport information at rates of gigabits per second through optical fibers finer than the human hair.
The stakes are very high. In a May 2000 news release, The Aberdeen Group (Boston, MA), an IT consulting firm, predicted that "The optical network market, excluding SONET elements, will grow to $17.7 billion by 2003. The suppliers that can deliver the technologies that solve the problems that carriers face will be the ones to succeed."
The use of the plural in "suppliers" and "technologies" highlights a key issue in this article.
The burgeoning communications network is highly complex. While a few large companies tend to dominate the global deployment of the optical network, behind the scenes there is a fusion of technologies developed by multiple companies, each with specialized technological expertise. Dallas Semiconductor falls into this latter category; we have designed a family of variable resistors especially for optical transceiver modules (see figure 1). A look at where Dallas' humble (but sophisticated) resistors fit into the grand scheme of communications networks reveals something about the way the communications industry develops solutions.
Identifying the big picture
Optical transceiver modules are designed and built by a variety of manufacturers. Applications for the modules include Synchronous Optical NETwork (SONET) and Synchronous Digital Hierarchy (SDH), Asynchronous Transfer Mode (ATM), Fiber Distributed Data Interface (FDDI), Fibre Channel, Fast Ethernet and Gigabit Ethernet. The names of these systems reflect the range of internationally defined transmission protocols and standards. On the other hand, the modules themselves were initially developed without definitive physical characteristics.
Recognizing the need for conformity if their products were to succeed, in 1998 a group of manufacturers banded together and developed a Multi-Source Agreement (MSA) for transceiver modules. The group consisted of AMP Incorporated, Hewlett-Packard Company, Lucent Technologies Microelectronics Group, Nortel (Northern Telecom), Siemens AG - Fiber Optics, and Sumitomo Electric Lightwave Corp. These parties agreed to cut the size of their modules in half (to 0.535 inches in width) and specified a set of module packages and pin-outs that would be interchangeable among the variety of RJ-45-style (including duplex LC, MT-RJ and SC/DC optical connectors) used in high-speed fiber-channel applications.
Currently, a new consortium is drafting a new MSA for transceiver modules, reflecting a larger contingent of manufacturers and a new generation of module. These multi-source manufacturers now include Agilent Technologies, Glaze Network Products, E2O Communications, Finisar, Fujikura Technology America, Hitachi Cable, Infineon Technologies, IBM, Lucent Technologies, Molex, OCP, Picolight, Stratos Lightwave, Sumitomo Electric Lightwave, and Tyco Electronics. The module specification is now called Small Form-factor Pluggable (SFP) and covers expected transmission rates of up to 5.0 Gb/s. The specifications reflect the industry's drive for high-density signal transmission in hot-pluggable modules of smaller size and higher speed.
The optical transceiver module To find where our resistors come into the picture, it helps to understand some basics about the transceiver module. The module converts incoming light waves to electrical signals and outgoing electrical signals back to light. Of fundamental significance, the optical transceiver is based on semiconductor laser technology. The module is a printed circuit board (PCB), and the optical source for the coveted bandwidth is a tiny semiconductor chip: a light-emitting or laser diode. At frequencies in the near-infrared spectrum, the laser's output can be modulated in tens of GHz, a capacious bandwidth.
The following briefly summarizes a signal path through the transceiver module: the receiving port connects to incoming light fibers; a photodetector diode converts the light to electrical signals, which are next amplified so that clock and data signals can be recovered, de-multiplexed, and sent out through the electrical interface; the photodetector requires an automatically power-controlled bias circuit to provide a constant operating voltage (see figure 2); meanwhile on the transmitting side of the module, electrical clock and data-bit signals are synthesized and latched and sent to the laser driver; finally, the laser driver sends the signal as electrical current to the laser diode, which converts electron energy to light.
In some designs that use laser diodes, a photodetector monitors the laser diode output and, in a feedback loop, reconverts the light back to electrical circuits that measure the laser's actual output power. This feedback stabilizes the laser output power. The optical feedback is a complicating drawback to this design. However, the latest laser technology, Vertical Cavity Surface Emitting Lasers (VCSELs), often do not require a photodetector because of exceedingly low current.
The laser driver must do two things: It must maintain a consistent DC-bias current to set the laser operating point, and it must maintain a modulation current to carry the signal. As manufacturers strive to increase signal throughput in transceivers, the laser source must be carefully characterized for operating constants in order to control the light output.
The laser diode and VCSEL
The Fabry-Perot type of laser diode emits a coherent light beam from the narrow, beveled edge of the chip, with reflecting mirrors incorporated at the edges or stationed outside the chip. For the future of the communications industry, however, a more promising laser source is the VCSEL. As its name suggests, the VCSEL vertically emits the laser beam from a circular cavity 5 to 25 micrometers in diameter at the top (the bottom is a future possibility) of the chip. The mirrors are incorporated as an integrated array on both ends of the cavity-a design known as a "distributed Bragg reflector."
In the future, parallel optical interconnects using multi-element VCSEL arrays could enable terabyte throughput.
Academic and corporate institutions are vigorously developing VCSEL designs for more widespread deployment. Compared to edge-emitters, the VCSEL requires less current and has a lower lasing threshold (1 or 2 mA versus 30 mA). At this level, simple current control is often sufficient without the extra photodetector to monitor output. The VCSEL's emitting aperture is measurably larger, which means that the output beam's angle of divergence (a measure of dispersion) is significantly smaller. There are several manufacturing and processing advantages, as well. The die is much smaller, allowing more VCSELs to be packed on a wafer with more interconnects; and all the VCSELs on an entire wafer can be tested at once. Lastly, the VCSEL is more robust in operation than a laser diode, with a longer life expectancy and lower failure rates.
Whether laser diode or VCSEL, the laser emitter in any optical transceiver is a semiconductor whose photoelectric effects depend on the interplay of current, voltage, and resistance. Some of the following factors affect safety and performance:
- Laser output is exceedingly sensitive to temperature.