Pearson Technologies Site Map

Training Technical Services Benefits Satisfied Clients Free Resources

Gallery Training Products Contact Experience

Fiber Optic Services And Products

On Fiber Optic Communications

Eleven Factors Favor Fiber to the Desk

Presented at

The 22nd Annual Newport Conference of Fiberoptics Markets

Eric R. Pearson, CPC

President

Pearson Technologies, Inc.

Executive Summary

The next two years look like interesting years. During this time, eleven factors are converging to favor the growth of fiber to the desk (FTTD). Each of these factors contributes a reduction in the cost of FTTD. By itself, none of these eleven factors is sufficient to enable significant growth in FTTD. However, when you examine all of these factors, I think you will see that they will create a powerful force favoring significant increases in the FTTD market.

Introduction

These eleven factors, which are somewhat interrelated, are:

Gigabit Ethernet,

VCSELs,

100BaseSX,

50 µm fiber,

Small form factor (SFF) connectors,

Quick cure adhesives,

Mechanical connectors,

Collapsed backbones,

Reduced cost of ferrules

Reduced cable costs,

And pre-terminated cables.

Gigabit Ethernet

The importance of Gigabit Ethernet is mainly in locations other than FTTD. However, Gigabit Ethernet (GBE) reduces the cost of high bit rate transmission to the desk: compare a gigabit Ethernet NIC card at $500[1] to an FDDI DAS at one tenth the throughput and more than double the price ($2000-$2400). Because GBE is based on the ubiquitous Ethernet protocol, it enables piecemeal upgrades of a network without requiring wholesale redesign and replacement of existing components. Ultimately, use of GBE should enable auto negotiation from 10 to 1000 Mbps through switches.

Vertical Cavity, Surface Emitting Lasers (VCSELs)

Vertical cavity, surface emitting lasers (VCSELs), low cost laser diodes, enable the low cost of GBE. VCSELs extend the capacity and life cycle of existing multimode cable plants beyond those possible with LED sources. The low cost of VCSELs will enable wider use of GBE than would otherwise be justifiable.

These devices can be modulated up to 5 Gbps, enabling multimode transmission to these rates, though at reduced distances.

The use of VCSELs has had two major impacts on the industry. The first is the increased awareness of the importance of the structure of the multimode fiber core. The second is the emergence of fiber and cable manufacturer guarantees of transmission distances in excess of those stated in the GBE standard.

The importance of the multimode fiber core structure to data communication networks was not fully realized until the development efforts that led to the GBE standard. Some multimode fibers contain a dip in the refractive index profile at the center of core. This dip did not influence transmission accuracy at the low data rates possible with LED light sources. However, the small spot size and high degree of collimation of the VCSELs can result in splitting of pulses that crossed this dip.[2] The process of developing the GBE standard resulted in fiber manufacturers addressing this issue.

By addressing this issue, fiber manufacturers have been able to provide performance guarantees in excess of the distances specified in the GBE standard. Such guarantees will have the future benefit of extended transmission distances for future protocols operating at data rates higher than the GBE rate. These extended distances will enable increased link lengths in future collapsed backbone, FTTD networks.

100BaseSX

100BaseSX, Fast Ethernet at 850 nm, is expected to become available late in 1999. This low cost fiber Fast Ethernet NIC is expected to be available with auto negotiation in the price range of $120- $160. This price represents a reduction of approximately 50% from the current cost of 1300 nm, 100BaseF NICs ($200-300). This low cost, fiber-based Fast Ethernet solution enables node-by-node upgrades of a full fiber network instead of fork lift upgrades of the entire network. In addition, this 100BaseSX option enables reduced cost, collapsed FTTD backbone networks, as is presented below.

50 µm Fiber

The 50 µm fiber offers a 5.5 % reduction in cable cost. Use of this product has been limited by exclusion from the standards developed prior to Fiber Channel and Gigabit Ethernet.

While this fiber is not the fiber of choice in the early fiber standards, its superior capabilities in bit rate and distance are clearly stated in the Fiber Channel and GBE standards (Table 1). This table indicates that use of the 50 µm fiber results in an increase of about 200 % in the Fiber Channel transmission distances and an increase of about 100 % in the GBE transmission distances. Further education of end users should increase use of this product and reduce the cost of fiber cable plants.

Table 1: Superior Transmission Distance Available with the 50 µm Core Fiber


Core, µm			50	50	62.5	62.5
BWDP, MHz-km[3]			400	500	160	200
Protocol	Wavelength	BR, Mbps		Distances,	m
Fiber Channel	780 nm	1062.5	500		175
Fiber Channel	780 nm	531.25	1000		350
GBE	850 nm	1000	500	550	220	275

The cost advantage of this fiber will improve in the future as volume increases. The current price advantage seems smaller that would be expected from the reduced manufacturing cost and improved yield expect from manufacturing considerations.

Small Form Factor Connectors

Small form factor connectors (SFF) will have a significant influence on reducing the cost of FTTD. The SFF connector will become the King of the Hill. The major questions are of timing and of which designs will dominate.

The small form factor connectors are designs that implement two fibers in a form factor similar to and approximately the same size as the RJ-45 connector that dominates copper cabling systems.

Small Form Factor Connectors, Part 1: The Promise

SFF connectors offer at least four, and possibly five, cost reductions:

Reduced hub/switch cost,

Reduced patch panel/enclosure cost,

Reduced jumper cost, and possibly

Reduced connector installation cost.

Reduced hub and switch cost results from doubling the density of fiber ports. This doubling is possible by the reduced size of the SFF connectors. This reduced size enables closer spacing of the light source and detector. This reduced spacing offers the potential to reduce the hub or switch cost per port by 30-50%. This significant cost reduction has the potential of removing the last major obstacle to FTTD- the increased cost of the electronics.

Reduced size of the connectors results in reduced costs for patch panels and enclosures. The same size panel or enclosure can contain double the number of ports that could be installed with legacy ST-compatible or SC connectors.

Reduced jumper cost is a potential of SFF. This single point comparison demonstrates a 24 % reduction in jumper cost: a 1 m duplex, 62.5 µm cable with ST-compatible connectors can cost as low as $14.42, while the same jumper with SFF Volition connectors can be as low as $11.

Reduced connector installation cost can result from use of anaerobic or quick cure adhesives. Use of these adhesives can result in significantly reduced labor costs. Reduced connector installation cost can also result from use the pre-polished connectors, like the MT-RJ and the pre-polished Opti-Jack. Pre-polished connectors eliminate the single, most time consuming step of polishing.

Small Form Factor Connectors, Part 2: The Difficulties

In spite of the promise of SFF connectors, four difficulties remain. These difficulties are:

Lack of awareness of the benefits,

Competition among six designs,

Lack of consistency among the standards, and most importantly,

An unclear cost analysis benefit.

Lack of Knowledge of the Benefits

The first difficulty is the lack of knowledge of the potential end users. These users are still viewing fiber as an ST-compatible and SC technology. This view is demonstrated by the informal survey I have been conducting during the last year. During this last year, I have trained 600-700 people in design and installation of fiber networks. My informal survey implies that fewer than 10% of potential users of FO data networks know the advantages of the SFF connectors. In addition, fewer than 5% know of the relative advantages of the six designs available.

Competition Among Six Designs

The second difficulty is the competition among the six SFF designs, which are the:

MT-RJ

Opti-Jack

Volition

LX.5

Advantages. These six designs offer different advantages. The MT-RJ, available from AMP and Siecor, offers the most options in optoelectronics, since more hub and switch manufacturers offer this interface than any other. Eventually, this design will be available in the plug and jack configuration, which is favored by the end user. At least 15 manufacturers of hubs and switches offer their products with the MT-RJ interface (Table 2).

Table 2: Partial List of Companies Offering the MT-RJ Interface

Allied Telesyn International Racore Technology Xylan

Canary Communications Inc. EtherCom Inc. XLNT Inc.

Garrett Communications Inc. Nortel Networks 3Com

HP ProCurve Networks Foundary Networks Cabletron Systems

Transition Networks Inc. Extreme Networks Cisco Systems Inc.

The Opti-Jack, available from Panduit Corporation, offers a rugged solution, a complete connector system, all based on well developed and commonly accepted technology. This design is available in the plug and jack configuration. An additional advantage is availability: plugs and jacks for both multimode and singlemode are available from stock. The Opti-Jack appears to be the most rugged of the six designs.

The Volition, available from 3M, offers the lowest cost connector system and the lowest cost electronics. In fact the cost analysis indicates that a Volition FTTD collapsed backbone is less expensive than a UTP, Category 5 link to the wiring closet with a fiber link to a central distribution facility (Table 3).

The LC offers small size, and if tuned, the lowest claimed loss. This product seems well suited to telephone applications in which the fiber density is high.

The LX.5 seems to be a telephone patch panel product in high density applications. The unique advantages of an automatic dust cover and of an APC version for singlemode fiber may make it well suited for CATV applications.

I am at a loss to find performance advantages unique to the MU.

Disadvantages. These six designs have different disadvantages. The MT-RJ is not a complete offering. As of 9/15/99, multimode plugs are available but jacks are not, except with an eight week delivery time. This situation makes implementation of an MT-RJ system contingent upon long delivery times or use of the plug to plug configuration instead of the more popular plug to jack configuration.

Singlemode plugs are available only as factory installed jumpers.

The Opti-Jack has fewer electronic options than the MT-RJ, the Volition or the LC.

The Volition has the disadvantages of being a completely new technology: v grooves are used instead of ferrules.

The LC has the disadvantage of a new ferrule diameter, 1.25 mm. In addition, it is a plug adapter plug solution instead of the plug and jack solution preferred in networks.

The LX.5 has the disadvantages of a new ferrule diameter, 1.25 mm and no known optoelectronics.

The MU has three disadvantages: no US manufacturers; no optoelectronics: no unique advantages over the other five designs.

Inconsistency Among the Standards

The third difficulty is that the standards committees have not adopted a unified view towards fiber connectors in general, and the SFF designs, in particular. The 10BaseF standard recommends the ST-compatible (aka BFOC/2.5); the Fast Ethernet, the SC; the GBE standard, the SC; the Fiber Channel standard, the SC; and TIA/EIA-568 A and B, the SC.

The TR41.1 committee has implied that it will not take a position regarding the recommended SFF connector. It has implied that it will let the market decide.

Unclear Cost Analysis

The third difficulty is that the total installed cost analysis is unclear. In an ideal market we would like to be able to make the following comparisons:

cost of SFF hub/switch < cost of ST-compatible/SC hub/switch

cost of SFF patch panels/enclosures < cost of ST-compatible/SC panels/enclosures

cost of SFF media converters < cost of ST-compatible/SC media converters

cost of SFF NICs < cost of ST-compatible/SC NICs

cost of SFF connectors < cost of ST-compatible/SC connectors

cost of SFF jumpers < cost of ST-compatible/SC jumpers

If the cost of each element in a SFF network were less than that in an ST-compatible/SC network, the decision to use SFF components would be a no brainer. However, the current reality is the following:

cost of SFF hub/switch æ cost of ST-compatible/SC hub/switch

cost of SFF patch panels/enclosures < cost of ST-compatible/SC panels/enclosures

cost of SFF media converters Ñ cost of ST-compatible/SC media converters

cost of SFF NICs > cost of ST-compatible/SC NICs

cost of SFF connectors > or < cost of ST-compatible/SC connectors

cost of SFF jumpers > or < cost of ST-compatible/SC jumpers

At this time, cost comparisons favor the cost of SFF hubs/switches, and patch panels. In one case, the Volition, the costs of SFF NICs, jumpers and connectors are all less than those of legacy ST-compatible/SC products.

However, with other SFF connector solutions, the comparisons are mixed. SFF NICs and media converters are more expensive than ST- compatible/SC NICs and media converters. With the exception of Volition, SFF connectors are more expensive than are ST- compatible/SC connectors (Table 4). Again, with the exception of Volition, SFF jumpers are more expensive than ST-compatible jumpers but less expensive than SC jumpers.

Table 4: Comparison of SFF and Legacy Connector Prices

Product	Jack	Plug	2 Connectors +2 Barrels	1 m jumper	Notes

MT-RJ	$18.85			$27.62
MT-RJ		$18.75			plug + adapter
Volition	$ 1.98			$11.00
OptiJack	$20.88	$14.87		$42.00
LC			$ 17.30		jumper/front panel plugs
LC			$ 14.75		back panel plugs

ST-c			$12.60	$14.42	AMP, distributor selling price
SC			$17.70	$ 32.65	distributor selling price

This unclear cost benefit analysis is probably due to at least three factors: the volume/price (or chicken/egg) conflict; the vested interest in legacy (ST-compatible/SC) products; and the lack of significant competitive pressures.

The volume/price conflict is a classic problem for all new technologies: how does the manufacturer achieve reduced prices? Through increased volumes. How does this same manufacturer achieve increased volumes? Through reduced prices.

The vested interest in legacy technologies may also be retarding reduced pricing of SFF products: a lower gross margin on a higher priced legacy connector product line may be larger than a higher gross margin on a lower priced SFF product line. What motivation does the manufacturer have to price the SFF products competitively? None that I see.

Finally, the SFF marketplace is in the developmental stage. Most end users do not know of the technology. Even fewer know of the competitive benefits of the various products. This lack of knowledge results in little competitive pressure on prices.

Quick Cure Adhesives

Since approximately 1992, a series of quick cure fiber optic connector adhesives have appeared. These adhesives, including anaerobic adhesives, offer the opportunity for reduced labor cost for installation of connectors. These products:

Eliminate the need for heat curing and long cure times,

Enable high installation rates, 12-15/hour and above.

These rates are approximately double those available for the original, and slowest, method, which is epoxy and polish.

In spite of these advantages, use of quick cure adhesives will be limited to the benign environments found in office buildings. These adhesives have reduced resistance to environmental extremes relative that offered by epoxy based connectors.

Mechanical Connectors

Mechanical connectors, such as the LightCrimpÅ from AMP and the UnicamÅ from Siecor, are the fastest to install. Their ease of installation and elimination of polishing time can result in the lowest total installed cost in high labor cost environments. However, the prices for these products tend to be higher and the learning curve for these products can be longer than that for other products, such as the 3M Hot Melt and epoxy connectors.

Collapsed Backbones

The concept of collapsed backbones is not new. However, knowledge of this concept does not appear to be widely spread among network personnel. The concept of a collapsed backbone is that of a direct link from a NIC in a PC (the client) to a centrally located hub or switch. This configuration has two electronic devices in contrast to existing networks, with three devices). This network configuration has six advantages:

Elimination of electronic costs in the wiring closet,

Elimination of costs for power conditioning and environmental control of the wiring closet,

Reduction in real cost/port through increased port utilization,

Reduction in maintenance costs,

Reduction in real estate costs, and

Reduced initial installation costs.

Table 3 contains a cost comparison of a collapsed backbone, FTTD Volition network to that of a Category 5 link to a wiring closet with a fiber link to the central distribution facility. Note that the FTTD solution, without consideration of costs for power conditioning and environmental control, is 21 % less than the cost of the UTP and fiber solutions.

The conclusion from this is a single point comparison is not meant to be expanded beyond its specifics. Our future work will include making this comparison for other combinations of SFF products in the collapsed backbone configuration.

Ferrule and Connector Costs

Ferrule costs continue to fall. ST-compatible connectors with ceramic ferrules are as low as a single piece price of $1.60! Use of liquid polymer ferrules (LCP), which are molded and not machined, also reduce the cost of connectors. Finally, I see no reason LCP ferrules cannot find their way into singlemode connectors, with a significant reduction in singlemode connector cost!

Reduced Cable Costs

I expect to see continued competition in FO cables in the premises and campus marketplace. This competition can only result in a reduction in cable prices. However, this reduction will not be major at this time. The selling price situation is complicated by two shifts: from 62.5 µm to 50 µm; and from standard fibers to GBE guaranteed fibers. These complications may result in stable or slightly increasing cable costs in the near future, with slight reductions as competitive pressures become stronger than the effects of increased pricing due to these two shifts.

Preterminated Cables

Finally, there is the beginning of a trend towards use of pre-terminated cables. This trend, which began in approximately 1994 with the Mod Tap system, can result in complete elimination of field installation of connectors.

Such elimination reduces system cost. Field installation of connectors can be at least three times more expensive than factory installation (Table 5). This increase in cost results from two factors: field labor rates are higher than factory rates and field installation times are longer than factory times due to time spent on field activities other than installation (i.e., set up, clean up and moving to new location). As such, use of pre-terminated cables can reduce the connector installation cost by 75-92 % (Table 5).

Table 5: Comparison of Field to Factory Installation Costs

location of installation=	field	field	field	field	factory
type of labor=	union	union	non union	non union	non union

if total loaded labor rate is=	60	60	30	30	10	$/hr

if installation rate is=	12	12	12	12	12	fibers/hour

if fraction of time spent on installation is=	50%	75%	50%	75%	100%

labor cost for installation is=	$10.00	$6.67	$5.00	$3.33	$ 0.83	$/connector

Savings with factory termination=	$9.17	$ 5.83	$ 4.17	$ 2.50		$/connector

Savings with factory termination=	92%	88%	83%	75%

Savings per 1000 fiber ends	$9,167	$5,833	$4,167	$2,500

This approach was based on the use of MPÅ type connectors, which enabled termination of up to 12 fibers in a connector approximately 0.5 inches x 0.5 inches. The small size of these connectors enabled factory installation without undue problems during cable installation. The ST-compatible and SC styles can allow for factory installation, but can have problems during cable installation due to their large size.

However, the small size of some of the SFF connectors will allow for factory installation. The small size of the Opti-Jack, the LC, and the MU can lead to factory installation. Such a strategy can result in significant reductions in total installed cost (Table 5).

Note that factory installation of MT-RJ and Volition jacks is not practical, since these jacks are relatively large.

I expect to see the increased use of pre-terminated cables in the future.

SUMMARY: A MODEL FOR FUTURE BUILDING NETWORKS

Based on this presentation, I propose the following description of the network design that takes advantage of all these factors. This design assumes a horizontal run of up to 100 m. This design allows for FTTD in a collapsed backbone for buildings of up to 98 stories.[4] This design allows for migration from Ethernet to Fast Ethernet with only replacement of NICs. If migration to GBE becomes necessary, the switch and the NICs requiring the upgrade need replacement. Assuming the switch will auto negotiate from 10 to 1000 Mbps, the NICs that do not need replacement can be used.

The desktop is a 10BaseF or 100BaseSX NIC with SFF. This NIC is connected to a wall plate with a factory built jumper.

All cables are the premises design with 50 µm core fiber rated at 500 MHz-km at both 850 nm and 1300 nm. The outer end of the cable from the face plate to the wiring closet is pre-terminated with the smaller SFF connectors, such as the Opti-Jack, LC, or MU. A strengthened, removable sleeve, similar to that offered originally by Applied Photonic Devices[5] is installed over the end of the cable and the connectors being pulled in. This sleeve protects the connectors from tension during the pull.

The second end of the horizontal cable has no connectors installed and ends at the wiring closet. This horizontal cable is connected to a vertical riser cable. The connection could be done with SFF plug and jack, but will probably be done with splices. Mechanical splices are probably the answer, unless testing demonstrates no BER problems with fusion splicing of multimode fiber with GBE.

The vertical riser cable has no connectors installed on its outer end. This end is pulled into the riser using current methods. The cable manufacturer could form a pulling loop from the aramid yarns and attach a disposable swivel eye for attachment to the pull rope.

The second end of the riser cable has the connectors completely installed. These connectors are fastened to the inside of the core of the reel, perhaps with VelcroÅ fastenings. These connectors are installed into a patch panel in an enclosure at the central distribution facility, often in the basement of building. Factory built jumpers connect this patch panel to the electronics.

The hubs or switches are 100BaseSX with auto-negotiation and with SFF receptacles. The hubs allow for a GBE interface as an uplink, again with an SFF interface.

In the future, as GBE matures, it should be possible to have switches with auto-negotiation from 10BaseF to 1000BaseSX.

Eric R. Pearson, Certified Professional Consultant, is President of Pearson Technologies Inc. and a 29 year veteran of fiber optic communications. Mr. Pearson is a Director of the Fiber Optic Association and Director of Certification of the same association.

Respectfully submitted for your consideration,