OM4: The next generation of multimode for the enterprise

While 40- and 100-Gbit/sec speeds are still in the future, there are many reasons to consider installing this type of fiber today.

Th 321397

While 40- and 100-Gbit/sec speeds are still in the future, there are many reasons to consider installing this type of fiber today.

The explosion in demand for bandwidth in enterprise networks is driving an urgent need for higher Ethernet network speeds. Several contributing factors include broadband penetration fueled by video-rich content, data center demands, and exponential growth in supercomputer and R&D computing activities.

Laser-optimized multimode fiber is recognized as the medium of choice to support these high-speed data networks. With next-generation 40- and 100-Gigabit Ethernet speeds on the horizon, the industry is developing a new type of multimode fiber, called OM4, which will offer an effective minimum modal bandwidth of 4700 MHz•km at 850 nm, compared with 2000 MHz•km for OM3.

What is OM4 fiber? How can you certify that the fiber you select has the bandwidth you need for these demanding applications? Are there benefits to installing this fiber today? This article seeks to answer those and other important questions.

The evolution of multimode fiber

The development of OM4 fiber can be traced from the introduction of the first multimode fibers more than 20 years ago. Multimode fiber systems traditionally have provided the most cost-effective solutions for meeting bandwidth demands in LANs, storage area networks, central offices, and data centers. Compared to single- mode fiber, multimode systems enable lower transceiver, connector, and connector installation costs at data rates from 10 and 100 Mbits/sec up to today's 1 and 10 Gbits/sec applications.

As demand for bandwidth and transmission speed continues to increase, the fiber industry has responded by developing evermore capable multimode product types, which are identified by the OM (“optical mode”) designation as outlined in the ISO/IEC 11801 standard:

  • OM1, for fiber with 200/500 MHz•km overfilled launch (OFL) bandwidth at 850/1300 nm (this is typically 62.5/125-µm fiber);
  • OM2, for fiber with 500/500 MHz•km OFL bandwidth at 850/1300 nm (typically, 50/125-µm fiber);
  • OM3, for laser-optimized 50-µm fiber having 2000 MHz•km effective modal bandwidth (laser bandwidth), designed for 10 Gbits/sec transmission. (See the table on page 22 for more details.)

Th 321397
Research indicates more than 10% of today's LANs could benefit from an OM4 fiber that extends beyond 100 meters.
Click here to enlarge image

Today, this evolution continues as the industry prepares for speeds of 40 and 100 Gbits/sec. In response to this emerging need, fiber manufacturers are developing OM4 fiber, a 50-µm media with extended bandwidth, which will be used to enhance the system cost benefits enabled by 850 nm VCSELs.

OM4 fiber will support Ethernet, Fibre Channel, and OIF applications, allowing extended reach upwards of 550 meters for ultra long building backbones and medium length campus backbones. With an Effective Modal Bandwidth (EMB, also known as laser bandwidth) of 4700 MHz•km (more than double the IEEE requirement for 10 Gbits/sec 300 meter support), OM4 fiber is also well suited for shorter reach data center and high performance computing applications. The reason is that optical loss budgets for these applications are rather tight at 10 Gbits/sec (and expected to get even tighter at 40 and 100 Gbits/sec), and the high 4700 MHz•km bandwidth can actually provide extra channel insertion loss headroom when OM4 fiber is deployed at less than its rated distance (explained in more detail later in this article).

What will the standards say?

There are a number of standards under development that will define the use of OM4 fiber for high-speed transmission. Within the TIA, work is progressing on TIA-492AAAD, which will contain the OM4 fiber performance specifications. Similarly, IEC is working in parallel to adopt equivalent specs that will be documented in the international fiber standard IEC 60793-2-10 as fiber type A1a.3.

As of this writing, there is general agreement among the standards committees that OM4 will have a significantly higher bandwidth (EMB of 4700 MHz•km with VCSEL launch at 850 nm) than OM3. It will also be backward compatible with applications calling for OFL bandwidth of at least 500 MHz•km at 1300 nm (e.g., FDDI, IEEE 100Base-FX, 1000Base-LX, 10GBase-LX4, and 10GBase-LRM).

Th 321398
Click here to enlarge image

There continues to be discussion and debate within the standards groups about a minimum OFL bandwidth requirement at 850 nm. Although current applications primarily use 850 nm VCSEL lasers with fibers that are specified to a minimum EMB, there is good reason to also establish a minimum 850 nm OFL bandwidth specification. It has been shown that fibers with higher OFL bandwidth will perform better with VCSELs that launch more power into outer modes. That is why the existing OM3 fiber standards require a minimum 1500 MHz•km OFL bandwidth at 850 nm.

For OM4, OFS and others in the standard group recommend at least 3500 MHz•km OFL bandwidth to ensure top perform- ance and reliability. Consensus within the standards group should be reached soon, if not by this printing, and the OM4 specification should be published by the end of this year.

Meanwhile, IEEE continues to work on standards for next-generation speeds, where OM4 fiber is likely to play a large role. For short reach 40- and 100-Gbits/sec applications on multimode fiber, it appears the IEEE 802.3ba Task Force will define a Physical Medium Dependent (PMD) solution involving already-proven parallel optics technology. This will help preserve the low-cost advantage of today's 850 nm VCSEL light sources. These parallel systems will transmit one 10 Gbits/sec signal on each of 4 or 10 fibers (for 40- and 100-Gbits/sec, respectively). Each 10 Gbits/sec signal will be aggregated in an arrayed transceiver containing 4 (or 10) VCSELs and detectors.

For these parallel systems, IEEE set an objective of a minimum reach of 100 meters, specifically on OM3 fiber (OM1 and OM2 will not be supported in the 40- and 100-Gbits/sec standard). There continues to be discussion about including OM4 in the standard, which would provide extended reach beyond 100 meters. The longer reach is expected to support the remaining 10 to 15% of access-to-distribution and distribution-to-core links that exceed 100 meters in larger data centers.

It is important to carefully study the performance characteristics of enhanced multimode fiber. Performance of OM4 fiber will be verified using the same criteria as OM3, but to tighter specifications. The IEEE 802.3 10GbE link model recommends an EMB of 4700 MHz•km for 10 Gbits/sec operation to 550 meters. Differential Mode Delay (DMD) mask specifications will be tightened proportionately.

There has been much discussion about the use of DMD, EMB, and OFL bandwidth measurement methods. It's important to remember that these parameters exist for one reason—to determine whether a link will operate when inserted into a system. OFS has been a strong proponent of the DMD mask method for characterizing high performance multimode fiber, and the results of a recent study show that the DMD masks provide the most rigorous screening of high performance multimode fiber.

For this study, commercially available OM4 fibers and cables from various vendors were tested to determine their ability to support claims of extended link distances for 10 Gbits/sec transmission. The fibers were measured using the DMD mask technique as well as the calculated EMB (EMBc) method, and they were then subjected to BER systems testing.

Results of the study indicated that the DMD mask method showed excellent correlation with system performance, and reliably identified poor performing and even failing fibers when matched up with marginal transceivers. The EMBc technique does not always find such fibers. Some fibers that failed the DMD masks—yet passed EMBc requirements—showed significantly poorer performance, and in several cases actually failed system testing.

It should also be noted that fibers with relatively low OFL bandwidth tended to perform poorly in systems testing when matched with a transceiver that launched higher power into outer modes. Similarly, fibers with good control of inner modes (as evidenced by low DMD in the center region of the fiber) performed better with transceivers that launched more power into this region. The EMBc method does not probe the center portion of a fiber, thus allowing marginal fibers to be deployed.

Going for additional power margin

As mentioned, some fibers in the market already meet or exceed the emerging OM4 standard. While 40- and 100-Gbits/sec speeds are still in the future, there are many reasons to install this type of fiber today. For example, you may want extra channel insertion loss “headroom” in your network to accommodate additional connections and higher loss connectors, and to improve overall reliability. This is especially critical in 10-Gigabit Ethernet applications at 850 nm, since loss budgets for these systems are significantly lower than previous applications.

There are two ways to achieve greater power headroom (also known as power margin):

  • Reducing Channel Insertion Loss (CIL), the end-to-end loss resulting from all connections and splices in the link, plus the attenuation of the cable itself;
  • Use a higher bandwidth fiber to reduce Inter-symbol Interference (ISI), which occurs when bits of data run together.

For greater flexibility in network design and—ultimately—greater reliability, follow these strategies:

  • Specify lower loss cables and connectors to reduce CIL.
  • Specify a fiber with bandwidth that is rated for a longer distance than what it will be used for. This frees up ISI pen- alty that can be added to the CIL budget. For example, you can get an extra 2 dB of margin in a 10GbE link by using an OM4 fiber deployed to 300 meters, or an extra 2.5 dB or more at a distance of about 150 meters or less.
  • Don't assume that all products that meet a particular standard are equal. It's possible to find higher performing products that exceed the standards.

Fibers with lower loss and tighter geometry specs provide extra loss margin, and those that are measured for bandwidth using the DMD mask method and have tight DMD specs all the way to the center of the core provide greater systems performance and reliability.

TONY IRUJO is sales manager for OFS (www.ofsoptics.com), Sturbridge, MA.

More in Fiber