140 Mbit/s – 622 Mbit/s
Approximately 30 years ago, optical transmitters and receivers were designed and built using “discrete” devices. These initial laser devices and pin-FET receiver modules were typically packaged in their own housing. The complete optical transmitter or receiver function, requiring more than just the laser or receiver, was realized by using a Printed Circuit Board (PCB), upon which the laser driver, receiver amplifier, and decision circuitry were placed using non-integrated parts.
The laser package typically contained a laser chip, an integrated backface monitor photodiode to control the laser output power, in some cases (depending on the package style) a Thermo-Electric Cooler (TEC) and a fiber pigtail with coupling lens. Initially, two package styles were available: bulky coaxial packages and the 14-pin Dual-In-Line (DIL) style. The coaxial packages were uncooled and there were many (non-standardized) versions, each requiring a vendor-specific bracket and electrical interface, and therefore vendor-specific PCB mountings and wiring. In general, and these early packages were difficult to handle in a manufacturing environment. The DIL packge became a de facto packaging standard for optics, as it was easier to implement in a general manufacturing environment. Both cooled and uncooled DIL versions existed. In early implementations TECs were used to control the laser temperature (mostly at room temperature), thus improving the reliability of the laser. Unfortunately, the TEC devices were also the most unreliable component in a laser package. Initially lasers were used in 850 nm and 1310 nm applications for bit rates up to around 1 Gbit/s. The first laser chips were Multi-Longitudinal-Mode (MLM) or Fabry Perot types. Later, 1550 nm Distributed Feedback (DFB) laser chips became available, exhibiting narrow spectral width and being useful for distances up to 80 km. The 1550 nm lasers were initially packaged in DIL housing in order to get improved laser stability and high-power operation through the use of the TEC for laser chip cooling.
Early lasers for optical transmission operated at fiber-coupled optical output powers of approximately 1 mW or above. Newer low-power lasers, operating in the 1310 nm window around 0.1 mW, were later introduced, allowing the coupling between the laser chip and the fiber pigtail to be less critical. In this case, one could work with mechanical tolerances much higher than those for laser devices with high fiber-coupled powers. It was recognized that, when lasers are manufactured at high volume, the majority of the cost of a laser is in the optical coupling between the laser and the fiber as well as the package and not in the laser chip itself. As such, anything that could be done to increase the yield of optical coupling or simplify the manufacturing resulted in significant cost savings. The trend towards using lower-power lasers to save cost is reflected in the definition of intra-office/short-reach and short-haul/intermediate-reach interface specifications in ITU Recommendation G.957.
At the beginning of the 1990s a de facto standard for smaller-sized coaxial pigtail packages arose. This new package, together with an improved performance of uncooled laser devices, allowed a migration towards using uncooled transmitters for almost all applications with distances up to 80 km for bit rates up to 622 Mbit/s.
For discrete receiver modules, the packaging evolution has processed at a slower pace. In most cases the receiver devices required additional semiconductor parts, like GaAs FETs or preamplifier ICs, in order to obtain the standardized receiver sensitivity performance. For receivers, both DIL and coaxial housing have been used. Initially only pin receivers were available to the market. However, due to the need to achieve improved receiver sensitivities, APD (Avalance Photo-Diode) receiver devices were introduced to the market and the standards. The APD receivers were initially used for 622 Mbit/s 120 km applications. The lower cost PIN receiver technology has been appropriate for all applications at bit rates up 622 Mbit/s for distances up to 80 km.
Approximately 25 years ago, the first 2.5 Gbit/s optical interfaces were designed, and with that, several new challenges related to High Frequency (HF) performance and signal integrity through the package were introduced.
At this speed, optical isolators were necessary inside the laser packages to minimize the influence of back-reflection from the outside plant towards the laser chip, since this could cause transmission errors with the laser. The traditionally used DIL packages were not suitable at 2.5 Gbit/s because of their limited HF performance. As such, “butterfly” packages were introduced, incorporating impedance-matching circuitry for optimized HF performance. For all 2.5 Gbit/s applications, SLM or DFB laser chip technology was required to meet the dispersion requirements of the various applications. Moreover, TECs were initially also used to obtain stable laser performance.
At the end of the 1990s, DFB laser performance had improved such that uncooled operation was becoming possible and miniature version of the butterfly package was introduced, since the space for a TEC was no longer needed. The mini-butterfly packages initially served 1310 nm applications and later were used also for 1550 nm applications.
When the first 2.5 Gbit/s DWDM applications were being introduced at the beginning of the 1990s, direct laser modulation was no longer suitable for transmission over distances beyond 80 km. Instead, external modulation was required to minimize the effect of laser chirp, permitting transmission distances well above 80 km. The laser and external modulator were ultimately integrated into a single package, known as an Externally Modulately Laser (EML), and later the device was integrated on to a single chip. The single-chip EML device consists of a laser section, operating in Continuous Wave (CW) mode, and an electro-absorption modulator section, in which the light emitted by laser section is modulated. A new range of EML driver ICs had to be developed, because EMLs work in “reverse” mode: no light is emitted when current is applied, which contrasts with conventional laser chips, where light emits when current is applied to the laser. For the first 2.5 Gbit/s DWDM applications, with channel spacings of 100 GHz (0.8 nm) or more, the drift of the EML frequency was not significant enough to result in optical link problems. However, for narrower channel spacing, wavelength lockers were necessary to achieve the required frequency stability and minimize the drift of the laser’s wavelength.
2.5 Gbit/s were initially deployed in 40 – 80 km long-haul/long-reach spans. The link budget of these spans required the use of a very sensitive receiver, hence APD receivers were initially deployed. In order to maintain high-frequency signal integrity, the electrical leads coming out of the receiver package needed to have a minimum numbei of discontinuities. Because of this, the electrical leads were typically oriented parallel to the axis of the cylindrically shaped package. At 2.5 Gbits/s, integration of (pre)amplifier electronics within the receiver package was required in order to achieve the required performance. Later, once 2.5 Gbit/s systems were deplyed in short-haul/intermediate-reach and intra-office/short-reach applications, the lower-cost pin-type receiver devices were introduced, employing the same packaging concepts as for APS-type receivers.
Welcom to buy the 140 Mbit/s – 2.5 Gbit/s SFP modules from Fiberstore!