The increasing density of optical connections puts a new focus on the performance-affecting shape of the connector’s end.
BY ERIC QUINBY
As the need to maximize the use of precious real estate in the data center becomes increasingly important, the outside-plant telecommunications environment requires quick, convenient, multi-fiber connections. Both of these communications-cabling trends have driven the need for ever-denser fiber-optic connection points.
Meeting this need is the array-style MTP connector. Though many array-type connectors are available, the MTP is the considered the primary industry-accepted device. The MTP has become ubiquitous in the data center because of its economic use of space, and because it provides a clean migration path from today’s serial digital applications to tomorrow’s high-speed applications, which will be parallel transmissions over optical fiber.
As fiber deployments closer to the customer in various outside-plant customer-owned local area network (LAN) spaces grow, so too does the need for dense connectivity solutions in a relatively small package.
 FIGURE 1. A pinned version of the MTP connector, the primary industry-wide accepted array connector in use today. |
Since MTP connectors are increasingly used to replace the splicing of fiber-optic backbones, the environment in which they are expected to perform changes from the relatively benign setting of the data center (where temperatures are generally expected to be maintained between 0° and 60° C and most other mechanical and environmental exposures are limited) to the harsher environment found at the customer premises (where temperature extremes could swing rapidly from -40° to 70° C and the connectors may be exposed to mechanical challenges).
The rigors of using MTP connectors, brought on by environmental conditions expected in the outside plant and data center, serve to amplify the importance of understanding and controlling the endface geometry conditions necessary to achieve low loss and low reflectance for every fiber pair that gets mated in the connector.
Ethernet’s upward march
Data rates of 1 Gbit/second are common over fiber, and increasingly, so are 10-Gigabit/second systems. Applications of 100 Gbits/sec are targeted by the Institute of Electrical and Electronics Engineers (IEEE; www.ieee.org) and will be common in the future. Because fiberoptic cabling systems are capable of supporting applications in excess of 15 to 20 years after installation, it is reasonable to anticipate that the systems will also need to remain functional for that time span.
 FIGURE 2. Radius of curvature refers to the radius of the connector’s hemispheric-shaped 250-μm center. For singlemode connectors, the value should be between 10 and 30 mm. |
Though there’s no way to accurately predict what will be required of fiberoptic connectors in the future, we can look to recent trends and requirements imposed by other high-speed optical systems.
Looking at how Ethernet has progressed from 10 Mbits/sec to 10 Gbits/sec, the trend that stands out in the Layer 1 optical backbone supporting these applications is the decrease in link-loss budget allowed by the transceiver pairs. Coupled with the fact that networks are becoming increasingly connector-rich, the trend has driven a need in the industry for lower-loss connectors. Therefore, it is reasonable to deduce that lower loss is a requirement for connectors installed today if they are to remain functional with the next generation (or two) of high-bandwidth applications.
 FIGURE 3. Fiber height refers to either the protrusion or the undercut of the optical fiber in relation to the tip of the ferrule. |
In addition, connectors installed today will need to have low reflectance at each fiber-to-fiber mated joint to ensure functionality with the next two application generations. Though low reflectance is not a functional requirement for 1 or 10 Gbit/sec Ethernet, it is for such high-speed digital applications as Synchronous Optical Network (SONET) and most analog video.
Endface geometry’s place
Fortunately, the parameters required to achieve low loss and low reflectance in MTP connectors are the same: superior endface geometry of the fibers and ferrule endface.
Understanding the endface geometry parameters that affect single-fiber connectors provides a good background to understanding factors that affect MTP connectors:
- Radius of curvature (ROC) of the fiber’s and ferrule’s endface. The fiber tip is polished to a parabaloid dome. Around the region of interest of the fiber, i.e., the center 250 μm, the shape is mathematically fit to that of a hemisphere (see Figure 2). This endface geometry parameter refers to the radius of that hemisphere; the tighter the number, the smaller the dome. For single-fiber connectors, the value should fall between 10 to 30 mm to ensure that low loss and low reflectance will come from fiber-to-fiber contact.
- Fiber height, i.e., undercut or protrusion of both fibers. With single-fiber connectors, the fiber can either protrude above or be undercut relative to the ferrule so that its high point, or apex, actually lies below the polished surface of the ferrule (see Figure 3). The accepted industry standard for this undercut is ± 50 nm across a parabaloid-shaped endface (think of a flattened bullet tip) that is either 1.25 mm (for connectors like LC and MU) or 2.5 mm (for connectors like SC, FC, or ST-compatible). It’s an amazingly tight tolerance. Consider this analogy: 50 nm of height variance across a 2.5-mm surface is equivalent to having a ¼-inch height variance across the length of a football field. Having the fibers physically touch is critical, because even a sub-nanometer gap between them causes an abrupt change in the index of refraction (IR) of the light’s path. The abrupt change in the IR causes an unacceptable point of reflectance of the transmitted light.
- Apex offset. Ideally, every polished connector will have the high point on its dome coincide with the center of the fiber’s core (see Figure 4). But due to polishing errors, such as having the connector wobble in the polishing fixture, the high point may not always be in this spot. The difference in where the high point is and where it should be is known as apex offset or dome offset. The actual high point should be no farther than 50 μm from its theoretical location. Deviations will yield a connector that does not properly mate (i.e., fiber cores do not physically touch), has higher insertion loss, and has high reflectance.
Common characteristics
Proper endface geometry for MTP connectors has several characteristics in common with that of single-fiber connectors. First, as with conventional cylindrical ferruled ceramic connectors, each fiber must protrude (microscopically) from the ferrule, come together, and mate (physically touch) with a corresponding fiber in the corresponding connector.
 FIGURE 4. Apex offset describes the extent to which the high point of a polished connector is away from the exact center of the fiber’s core. |
With MTP connectors, as with single-fiber connectors, fiber height is critical; however, unlike single-fiber devices, MTP connectors are polished so that the fibers protrude from the ferrules to ensure physical contact. This is done because there is more mating surface on the endface of an MTP connector (see Figure 5).
When mated to another MTP connector, the large mating surface will evenly distribute the spring force in each connector that is pushing the ferrules together. If one or more of the fibers in either of the MTP connectors should be undercut (i.e., the polished fiber apex lies below the MTP connector ferrule endface surface), the spring force inside the connector cannot deflect the ferrule material enough to force the fibers to mate. MTP connectors must always have a positive controlled fiber protrusion when properly polished.
Depending on the MTP connector manufacturer and its given philosophy for achieving low loss and low reflectance, these fiber heights can range anywhere from several dozen to several thousand nanometers. There has yet to be an industry standard on exactly how high these polished fibers must protrude above the ferrule, but work is underway and it is reasonable to expect either a de facto or published standard to emerge in the next three to five years.
MTP connectors must also maintain a fiber-height range, defined by the difference of the maximum fiber height from the planar surface of the connector ferrule to the minimum fiber height from the same plane (see Figure 6). The difference between the maximum and minimum fiber heights must be minimal, so that all the fibers touch.
 FIGURE 6. This figure shows the fiber height differential that can exist between fibers. The difference between the maximum and minimum heights must be minimal. |
The complexity is compounded by mating arrays that may have fiber heights that are not in a linear configuration. Though the goal is always to polish all the fibers to the same height, thus giving you a theoretical line of fiber heights, some deviation is inevitable. Even though every fiber is not exactly the same height, each must still physically touch its mating fiber in the connector plugged in from the opposite side.
Peer-to-peer height is also significant, particularly the differential in height from the planar surface of the ferrule’s monolithic endface of one fiber relative to its adjacent fiber’s height. Each fiber in a row, when mated to a complementary connector, will be able to deflect when subjected to force (think of the higher fibers compressing). The deflection will be resisted by the height differential of the fibers adjacent to as well as others that make up the planar surface of fiber endfaces.
The concept of coplanarity
Recently, “coplanarity,” a new concept that captures and explains both fiber-height range (maximum to minimum) and the peer-to-peer height differential, was submitted to the Telecommunications Industry Association (TIA; www.tiaonline.org) as a ballot for the revision of the Fiber Optic Test Procedure that will become TIA/EIA-455-219-A; Examinations and Measurements-Polish Angle and Fiber Position on Single Ferrule, Multi-fiber Connectors (FOTP-219).
The concept of “coplanarity” (see Figure 7), was first floated as a submittal as an informative annex to the document. Subsequent meetings have yielded such strong industry agreement that its adoption as a nominative (considered a prescriptive part of the standard) definition seems likely.
 FIGURE 7. The concept of coplanarity, illustrated here, is described in technical detail in the drafted version of FOTP, PN-3-002-RV1. |
A complete technical description of coplanarity can be found in the drafted version of the FOTP, PN-3-002-RV1.
There are two similar concepts of coplanarity:
- The first, used for a single array-style connector and accounting for the fiber heights relative to the ferrule, uses a least-squares-fit (two-dimensional) line. The coplanarity of any given fiber is then a measure of the deviation from this theoretical line.
- The second is used for multiple array-style connectors and uses a least-squares-fit (three-dimensional) plane. In this model, the coplanarity of a fiber is the deviation from the theoretical plane.
Coplanarity is a valuable tool in that it also accounts for the rounding of the planar surface that can be experienced by physically polishing the ferrule. If care is not taken, the planar surface can have the edges rounded in either or both of the X and Y axes. Were this to occur, the fibers on the outer edges would be polished lower (closer to the ferrule material itself).
Coplanarity, however, would account for these “shorter” fibers, either linearly in the X-axis only as in the case of a single-row array connector, or planarily as seen in both the X- and Y-axes as in the case of a multiple- array connector.
The next parameter in array-style MTP connectors that must be controlled and limited to eliminate the possibility of fiber-to-fiber air gaps is X and Y tilt (see Figure 8).
 FIGURE 8. For multimode connectors, the ferrule endface ideally is perpendicular to the guide pin bores; in reality, it can be tilted in either the X or Y axis. |
Ideally, the ferrule endface is perpendicular (for multimode connectors) to the guide pin bores, but in reality, it can be tilted in either the X or Y axis. This concept is based on one connector having guide pins, which are mated to a complementary connector with guide pin bores. Ideally, the guide pins are straight, perfectly cylindrical, and are sized and spaced exactly the same as the guide pine bores.
If there is any interference fit between the guide pins and guide pin bores, it should be minimal enough to be overcome by the ferrules’ material elasticity, and is sufficiently over-ridden by the spring force in the connectors. A minimal set of requirements to mandate inter-company mateability and functionality for this set of parameters is defined and maintained by the Fiber Optic Connector Intermateability Standard for MPO-style connectors, TIA/EIA-604-5 (FOCIS-5).
The standard establishes the definition for adequate guide pin force by setting guidelines for ultra-low-tolerance, reference-grade, guide pin gauge artifacts that can be used to determine go/no-go status with ferrules that have been made. Presumably, manufacturers will couple such artifacts with statistical manufacturing means to achieve 100%-compliant parts in the field.
The practical realities
Singlemode parts are always made with an endface and fiber that are polished with an 8° angle in the X-axis, which achieves ultra-low reflectance needed for most high-data-rate systems. Singlemode MTP connectors always mate together with the keys in the opposite orientation so that the 8° angled endfaces line up in a complementary fashion.
Even with complementary polished endfaces, however, some amount of offset of the connector’s planar endface relative to the desirable angle of 8° will be more problematic if it exists in the Y-axis. Some amount of angular offset in this plane is acceptable, but should ideally be met with a complementary angle from the other (mating) connector.
Again, the concept of coplanarity goes a long way to account for this parameter, because it assumes the fiber heights tend to be polished planar rather than at totally random heights. Some amount of offset in the fiber hole spacings, as well as some amount of “tilt,” can be overcome because the ferrules are forced together and the endfaces float and form complementary angles to one another.
All these factors are important in ensuring that each fiber makes a physical contact connection that pushes fiber-onto-fiber in the connector’s joint, so that the index of refraction between the two glass cores is uninterrupted by even a nanometer of air or debris.
As array connectors become deployed more often in the outside plant, endface geometry parameters become even more critical because the environment and mechanical forces will subject them to work to counteract the fibers from making physical contact. The ferrules and other connector parts will shrink at a rate different from that of the glass used in the optical fibers. Some amount of shrinkage of one material relative to another will be acceptable, so long as every fiber is able to maintain physical contact in the process.
FOCIS-5 offers all the physical criteria of the moving parts that come into contact when an MTP connector, adapter, and second connector are mated together. While compliance with the document ensures what is known as “intermateability,” the document does not stipulate the endface geometry parameters and values needed to ensure fiber-to-fiber contact for all fibers in the array.
Telcordia has published GR-1435, which defines a battery of tests that are intended to provide assurance of fiber-to-fiber contact and low loss during most conceivable environments for the MTP connector. The tests are only valid on installed, polished connectors, and while they are an indicator of a finished assembly’s performance in the field, they do not specify all the endface geometry parameters or values needed to achieve passing results. Instead, the Telcordia test regimen assumes the connector assemblies being tested have the geometries needed to make them pass the various tests.
A starting point
MTP connector manufacturers are only now beginning to formulate comprehensive lists of the endface geometries needed for successful deployment of MTP connectors in the data center and outside plant. State-of-the-art MTP polishing and enface measurement equipment is large, bulky, and expensive, suitable only for a factory environment. Consequently, the designer/installer/user community cannot expect to achieve adequate MTP connector endface polishing results, and has no way of knowing if they have done so.
The only options currently available are either factory-made pre-connectorized assemblies (either splicing single-ended pigtails or assemblies pre-terminated on both sides), or factory-pre-polished connectors that contain a field splice inside.
ERIC QUINBY is an engineer with Corning Cable Systems (www.corningcablesystems.com).
