A new method combines fiber measurements with that of transceivers to enhance the performance of multimode 50-mm fiber in 10-Gigabit Ethernet networks.
By Doug Coleman and Phillip Bell
Enterprise local area networks (LANs) should be designed to support legacy applications as well as emerging high-data-rate applications. Information technology (IT) managers typically include optical fiber as the primary media type in structured wiring systems to support such requirements. Until recently, 62.5/125-µm multimode fiber has been the dominant fiber type deployed in the LAN. The emergence of high-data-rate systems such as 10-Gigabit Ethernet (GbE) now warrants a migration to 50/125-µm laser-optimized multimode fiber in the LAN.
Fiber bandwidth measurement techniques have evolved and been enhanced to ensure 50/125-µm laser-optimized multimode fiber will reliably support 10-GbE transmission. Differential-mode delay (DMD) measurements on 50/125-µm fibers can help predict effective modal bandwidth (EMB). The most recently adopted method, called minimum calculated EMB (minEMBc), further enhances the 10-GbE reliability of laser-optimized multimode 50-µm fiber.
High-data-rate systems
The Institute of Electrical and Electronics Engineers (IEEE) approved the 802-3ae 10-GbE specification in June 2002. The spec offers guidance for only one serial multimode fiber physical media dependent (PMD) solution, 10GBASE-SR.
The 850-nm serial PMD includes distances from 2 to 300 m. The 850-nm wavelength is used for multimode fiber in response to the economic feasibility criteria of the IEEE 802.3ae task force. Simply, vertical-cavity surface-emitting lasers (VCSELs) render the 850-nm serial solution more economical than a 1,300-nm solution using long-wavelength lasers.
The high data rate in conjunction with the desired application distances support 50/125-µm laser-optimized multimode fiber as the default fiber choice for application distances up to 300 m. The relatively smaller fiber core size of 50/125-µm yields an inherently higher bandwidth capability than 62.5/125-µm fiber. The 300-m maximum distance requires use of 50/125-µm fiber with a 2,000 MHz•km EMB.
As early as 1998, the TIA FO 4.2.1 working group discovered that high-performance laser-based multimode systems needed transceiver specifications alongside 50/125-µm fiber measurements. For 10-GbE development, they were tasked by IEEE 802.3ae to define the optimal encircled flux launch conditions to correspond with the 50/125-µm DMD specifications, such that a 300-m distance at 10 Gbit/s could be obtained. Encircled flux is the percent power within a given radius launched by a transmitter into a multimode fiber core. For 10GBASE-SR, the encircled flux power distributions became ≥ 30% at a 4.5-µm radius and ≥ 86% at a 19-µm radius (see Fig. 1).
 FIGURE 1. Power launched by the transceiver into a multimode fiber is optimized between 4.5 µm and 19 µm in radius. |
In addition to specifying the encircle flux of the transceiver, TIA FO 4.2.1 used a DMD measurement procedure to ensure the required EMB of 2,000 MHz•km. DMD is a fiber manufacturing measurement described in FOTP-220, where a single-mode pulse with a spot size of approximately 5 µm is scanned across the 50/125-µm laser-optimized multimode fiber core in at most 2-µm increments. The resulting data captures the differential mode delays and mode coupling power of a single fiber as a function of radial position (see Fig. 2).
 FIGURE 2. From this DMD measurement, two methods of predicting EMB were standardized. |
null
DMD masks
The first method for translating DMD data into an EMB prediction is commonly referred to as the DMD mask approach, where the leading and trailing edges of each pulse are recorded and normalized in power relative to each other. This normalization approach reduces the raw DMD data to focus exclusively on time delay, where the overall fiber delay can be calculated by subtracting the slowest trailing edge from the fastest leading edge in units of ps/m.
To meet a nominal EMB target of 2,000 MHz•km, a fiber must first conform to one of six DMD templates, or masks. In addition to meeting one of these mask sets, the fiber DMD must also not exceed 0.25 ps/m for any of the radial offset intervals.
According to standards, a fiber meeting any of these mask templates and meeting all of the sliding masks will meet a nominal 2,000 MHz•km EMB. Note that the DMD mask method provides only a pass or fail estimation around 2,000 MHz•km.
The newer method for predicting EMB from DMD is called calculated effective modal bandwidth (EMBc). EMBc takes advantage of additional DMD data that is neglected by the DMD mask approach. As mentioned, the DMD measurement characterizes a single fiber's modal performance in high detail, including both modal time delay and coupling as a function of radial position. With EMBc, this fiber's performance is then characterized by a set of representative sources, which are chosen to span across a range of more than 10,000 standards-compliantVCSELs.
Conceptually, this is done by first weighting the individual DMD launches to represent any desired VCSEL. Those weightings are then combined with the richer DMD bench output (the raw DMD data) to build an output pulse for that fiber/laser combination. The output pulse may then be used to calculate EMB in units of MHz•km. Combining the source and fiber DMD measurements yields a synergistic method that accurately calculates the effective modal bandwidth of a 10-GbE system. The concept described previously, however, is only an example of how system performance can be predicted between a given fiber and a single VCSEL.
To ensure field performance, EMB is then calculated for the entire range of standards-compliant VCSELs. In other words, the FO 4.2.1 task force noted that within their defined encircled flux requirements, the source manufacturers could supply a wide range of potential power/intensity distributions for 10-GbE VCSELs. In a characterization study of 10,000 10-Gbit/s VCSELs conceptually created for the TIA modeling work, along with a sample of 10 actual lasers used for anchoring the standard, only two extreme laser sources had power concentrated nearest the core and the edge of the core, respectively (see Fig. 3).
A system that satisfactorily works at 10 Gbit/s must also operate without failure in all cases. Some source manufacturers may tend to produce large-spot lasers, while others may produce small-spot ones, and the intent of the standards is to guarantee functionality for as broad a distribution of lasers and fibers as possible. One could test each fiber with dozens of lasers, verifying that it worked with each, but such an approach is impractical. The EMBc method uses a specific fiber measurement (the DMD) and a specific laser measurement (encircled flux) to predict how any fiber will work with a set of lasers, or how any laser will work with a set of fibers.
An important feature of the EMBc concept, as described briefly, is that any source's modal coupling to a fiber can be accurately represented mathematically using a weighted summation of offset pulses, so the system performance with any DMD-measured fiber and any light source can be tackled mathematically.
With regard to the distribution of VCSELs in the characterization study, any fiber will have ten different EMB values for each of these ten sources. TIA-455-220A Annex D provides a procedure for simulating ten different sources spanning the entire range of permissible sources and calculating the corresponding EMB value for each. Selecting the minimum calculated EMB value in turn guarantees that fiber's performance in the field with any acceptable VCSEL. For this reason, the second method of predicting EMB from DMD is often called minimum EMBc, orminEMBc.
In summary, the main purpose of the EMBc calculation is to ensure that the EMB of a fiber will meet the 10 Gbit/s requirement of 2,000 MHz•km with any conforming laser. Further, the method provides a bandwidth value in units of MHz•km, which can in turn be used to design systems supporting 10-Gbit/s performance beyond 300 m.
Advantages of EMBc
The EMBc method combines the properties of both the source and fiber, and more importantly, their interactions. It has many advantages compared to other bandwidth measurements adopted to date for guaranteeing a system's performance.
The first is a sound physics base and experimental verification. The EMBc process predicts source-fiber performance by integrating the fundamental properties of light sources with the multimode fiber's modal structure, which has been measured using a standardized DMD measurement.
Secondly, the EMBc method ensures worst-case compliance. The minimum EMBc used to specify the fiber performance ensures that the multimode fiber will work for all types of qualified sources, including, for example, the extreme hot centered and hot outside lasers. It is therefore a conservative and robust system performance metric.
Thirdly, it offers standards compliance and multi-vendor support. Many fiber, component, and system vendors widely support the EMBc method. Its adoption into the TIA and 10-GbE standards obtained broad consensus. This method is also in the process of being adopted by IEC as an international standard.
Finally, this technique allows measurement scalability. Because the EMBc method predicts fiber performance in scalable units (MHz•km), it can therefore be scaled to predict other bit rates and/or link lengths. Conversely, the DMD mask approach provides a pass or fail estimation around a nominal 2,000 MHz•km, so it does not easily lend itself to predicting other EMB values.
DOUG COLEMAN is manager of Technology and Standards in the Private Networks division of Corning Cable Systems, 800 17th Street NW, Hickory, NC 28601. Tel: (828) 901-5000; email: doug.coleman@corning.com. PHILLIP BELL is global product line management in the Premises Fibers division of Corning Optical Fiber, 1 Riverfront Plaza, Corning, NY 14831. Tel: (607) 974-5515; email: bellpd@corning.com.
