Powering of hybrid fiber/coax systems remains a topic of much debate and discussion in the industry today. Operators are faced with the requirements of system reliability, concern about competition, constraints on capital and operating costs and the need to solve the powering dilemma as quickly as possible.

This article describes a unique centralized powering approach that is currently being deployed by Southern New England Telephone (SNET) for its HFC system.

Network requirements

While Connecticut, SNET's service area, was one of the first states to embrace competition in communications, the SNET upgrade has been driven more by cost, quality, growth and service issues rather than as a response to any particular competitive threat. After evaluating all the available technologies, SNET engineers chose hybrid fiber/coax as the solution that most nearly met all of SNET's requirements.

In some ways, Connecticut can be considered a mature state, particularly in its development of useful land. Many Connecticut towns boast fiercely defended historical sections and high-value neighborhoods. This is particularly true in coastal areas. Connecticut's rocky nature limits underground development as well. As a result, more than 80 percent of SNET's plant is aerial.

The widespread deployment of fiber brings with it many challenges, not the least of which is power. Unlike conventional telephony, which provides a metallic conductor from the central office to the home, the HFC system uses optical fiber to carry information from a central location to remote distribution networks. These distribution networks contain optical-to-electrical conversion electronics, signal amplifiers, telephony support electronics and coax. The electronic elements require power-reliable power. SNET's approach to deploying HFC uses nodes that are intended to support about 200 homes. This means that there are powering requirements for about 10,000 locations across the entire state.

In addition to the sheer quantity, SNET engineers faced other challenges.

Powering requirements

Some of these other challenges included:

  • Reliability-power was allocated a reliability requirement that translated into no more than five minutes of down-time per line per year statewide, resulting from power-related outages.
  • Siting-rights-of-way for installing powering systems are simply unavailable in most of Connecticut's towns. When looking for locations to power an early trial, SNET was unable to even obtain rights-of-way along existing train tracks.
  • Cost-powering costs, both deployment and operation, could not be allowed to change the overall economic justifications of HFC deployment.

More traditional distributed powering approaches generally place a power system at each fiber node. Examples of such systems include utility-based cable TV powering augmented with some nominal battery back-up, or larger power nodes using both battery and motor/generator back-up.

While the distributed approach is conceptually simple, makes efficient use of utility power and minimizes failure group size, an analysis of the situation convinced SNET engineers that in fact, neither of these approaches was feasible. Even if siting could be found, neither approach provided the reliability or maintenance characteristics required.

Decision to deploy centralized power

After considerable review of existing powering approaches and in light of the difficulties unique to the Connecticut service area, SNET began to consider a centrally based powering architecture. Such architectures generally place larger power systems at centralized locations and distribute power to many nodes. The advantages of such a system include:

  • Relatively inexpensive reliable power nodes in the distribution area
  • Reduced problems with siting and right-of-way due to smaller housings
  • Simplified maintenance (no batteries and motor/generators in the field)
  • Economies of scale in large UPS systems for the central office
  • Availability of standard components (UPS, generators, etc.).

There are, of course, some difficulties with feeding power from a central location to a large distribution area, including:

  • Higher voltages must be used to achieve reasonable efficiencies when distributing power for longer distances.
  • The power distribution system must be constructed with the power source at the central office, a transmission conductor system and a distribution interface device.
  • On the average, a larger failure group can be expected due to the multiple powering feeds from the CO to the distribution areas. Good engineering can overcome most of these problems. Deployment of redundant UPS and back-up generators at the CO provide high reliability at the central power site.

This centralized approach, if implemented correctly, could prove to provide higher reliability than distributed approaches. Operators are well aware that distributed powering systems will not be practical if no acceptable sites can be found for the power node installations.

SNET network architecture

The new SNET HFC network is intended to be a highly reliable full service network. It is designed to carry traditional telephony along with data, analog and digital video, and special services. The network addresses the telephony needs of both residential and business customers and is expected to meet or exceed the reliability requirements associated with telephony. Telephone reliability is measured in minutes of downtime per line per year, and the normal target for a large system is roughly 53 minutes, or a reliability of 99.99 percent. Power, of course, is only allocated a small portion of this, typically five minutes per line per year, or 99.999 percent.

Currently, HFC equipment requires considerably more power than that consumed by traditional telephony equipment. SNET's coax sub-nets require between 500 watts and three kilowatts of continuous power. This means that a typical central office that supports 30,000 lines is faced with supplying an additional 300 kilowatts of power at the node. The phrase "at the node" is important because there are a number of sources of loss between the utility service at the central office and the power inserter at the fiber node.

Sources of loss include:

  • Power factor of the UPS inputs-today's systems can typically achieve an input efficiency of 93 percent by the use of harmonic filters.
  • Losses in the UPS systems-typically resulting in 90 percent UPS efficiencies.
  • IR losses in the distribution system-limited by design to < 20 percent.
  • Power factor losses in the distribution system-currently not measurable, but anticipated to be no more than a few percent.
  • Losses in the power node-limited to about 6 percent through the use of a reliable and efficient controlled ferroresonant transformer system.

Total efficiency is the product of all the contributing efficiencies. Multiplying the above efficiencies together results in an overall efficiency of about 65 percent, which means that the central office is required to provide the capacity of roughly 450 kilowatts.

Central office powering system

A reliable centralized powering system requires a highly reliable central power supply. Fortunately, within telephone companies, reliable central office powering is a well developed technology. Extending or replicating existing central office systems with redundant parallel UPS systems backed up by batteries and redundant motor generator sets can provide such reliable systems. Many of the components required in these new central office systems are catalog items. Modern high voltage AC UPS units are available from a number of quality suppliers, as are batteries, switchgear and motor/generator sets.

While space considerations, building configurations and total central office service requirements must be taken into account, upgrade of a central office to support a distributed powering system is a fairly easy undertaking with commonly available power equipment. (See Figure 1)

The SNET power and fiber hybrid cable

The SNET communication cable is a hybrid cable combining fiber with the metallic conductors designed to transport the central office power. The main cable is constructed with an annular ring of conductors surrounding a central conduit through which fiber is pulled, completing the construction of the cable. The cable has a steel sheath covered by 110 mils of polyethylene insulation into which is embedded three colored longitudinal stripes.

The primary version of this cable contains nine 1/0 aluminum conductors which can be evenly allocated to the three power phases. When arranged in this way, the cable exhibits a nominal resistance of 0.067 ohms per kilofoot per phase. For branches off the main distribution lines, a similar but smaller cable has been developed that exhibits a resistance of 0.169 ohms per kilofoot. (See Figure 2)

Splice case, termination, distribution

There were a number of splicing issues to be overcome in designing the construction of the power system. With the fiber incorporated into the hybrid cable, provisions needed to be made to allow fiber splicing without exposure to power, and the hybrid cable is designed to allow exactly that. Using conventional splice enclosures, splices can be configured that allow fiber splicing to be done outside the enclosure containing any power splicing.

This is accomplished by allowing the central ducts containing the fiber cable to exit the power splice cases independently. Because slack for the fiber can be pulled when the fiber is installed in the cable, there is no need to modify the fiber splicing procedures. A number of splice configurations have been documented, allowing a variety of branch and through splices to be implemented.

Distribution power node design

The power node (Figure 3) contains a number of specific features, including:

  • Configuration. Pole mount housing with input voltage termination and four output feeders. Integral power transformer, interconnections, service bypass system.
  • Safety. As with all aspects of this system, consideration for safety was paramount in the design of the power node. A dead front design was required and was achieved through the judicious use of barriers. Disconnects for the input power and breakers for the four output circuits were also provided.
    Other safety features include: double isolation of input voltage section, single point ground and neutral bond point, input power disconnect with "lock out-tag out" provision and proper environmental protection.
  • Reliable and efficient design. The power node is based upon a controlled ferroresonant transformer design with multiple input and output taps. The input taps allow the node to be powered from either the central office or the local utility secondaries. The output taps provide 60-, 75- and 90-volt, quasi square-wave, 60-Hz outputs that can be selected based on local conditions such as coax size, distance and voltage drop, etc. The 75-volt tap provides the opportunity to supply large rural coax networks in 60-volt builds. The controlled ferro was selected because of its high efficiency, high reliability and wide range of operating parameters. Because of the limited number of components and lack of any active semiconductor devices, MTBF is calculated to exceed 200,000 hours. (See Figure 4.)
  • Field service and maintenance bypass. The power node allows field maintenance by including provisions for bypassing the internal transformer with no service break. Manual switch gear allows a "make before break" transition after phase synchronization with an auxiliary truck-mounted inverter system. Once this transition is achieved, the transformer module in the power node can be disconnected and repaired or replaced. After re-synchronization, the transition can be reversed with no loss of service.

This is a significant feature that compensates for the non-standby design of the distribution power node and operation with the strict outage limits of a broadband telephony system.

Powering amps, NIUs and telephony gear

Although there has been much discussion around what type of power should be provided down the coax to the RF amplifiers, fiber nodes, network interface units and telephony equipment, SNET selected conventional 60- and 90-volt, 60-Hz, quasi square-wave power.

This approach has a proven history of operating well with most active devices in the coax distribution networks. In addition, there is a proven long-term track record regarding corrosion activity of 60-Hz systems (unlike some of the newer low frequency designs).

By using line frequency powering, power is easily and reliably derived from the higher voltage AC transmission conductors and is provided to the power nodes from the central source. In its simplest form, all that is required in the power node is a ferroresonant transformer.

Powering network stability and margin

There are many issues associated with any powering solution. Irrespective of the source of power, the power system must supply enough current to handle in-rush conditions at start-up and maintain a stable operating environment. In many designs, this is far from trivial.

A Bellcore study, performed at the request of SNET, modeled a typical large distribution network and power node. The actual input circuits of the active devices were included in the model which demonstrated that certain designs can fail under start-up conditions. For this reason, the power nodes are designed with considerable headroom in their operating parameters.

As previously mentioned, there are losses associated with the transmission of power. In order to minimize these loses, the total voltage drop in a feeder network is limited to 80 volts, meaning that the lowest voltage that a power node will see is 400 VAC. The power node will regulate well below this value, so there is adequate headroom built into the basic design.

While up to 20 percent IR loss is allowed for a given distribution run, these runs are typically designed for less than 10 percent loss.

Southern New England Telephone did not want the powering system to be the weak link in the service chain. With redundant UPS systems in the central office, a strong distribution cable construction, a simple, robust power node and the ability to perform preventive maintenance on virtually all components of the system, it is expected that power will be the most reliable of the hybrid fiber/coax system elements.


Unique requirements demand a unique solution. Individually, none of the challenges faced by SNET in the implementation of its HFC system is unique. Still, the combination of requirements, environment, reliability, cost, siting, etc., present a real challenge to system powering. This new system is seen as a serious solution to a unique and challenging problem.

Reliability of HFC powering is very important. While there are a number of other approaches to powering HFC systems, none seem to provide the reliability of this approach. SNET's customers expect and demand a reliable phone system. In addition, the company expects to differentiate its other communications and entertainment offerings with superior reliability. From both a business and public trust point of view, reliability is something on which it cannot compromise.

Could this system be used by other networks? The situation for others may be different. It may be easier in some locations to obtain the sites and permits necessary to install self-contained power nodes. Ambient weather conditions may be such that batteries in the outside plant become less of a maintenance issue.

Natural gas may be widely available to power motor/generator sets. On the other hand, these and other issues may continue to dog fiber installations to the extent that some of the ideas and techniques presented here may be of use to others.

In any case, there is now a new approach available for others to consider. System designers can weigh the advantages of reliability, low maintenance and economies of scale with the disadvantages of higher utility costs and larger average failure group size inherent in this approach. Perhaps variants of the centralized power approach using different voltages and other transmission conductors could be considered.

Author's note

Currently, SNET is running a trial of its HFC telephony technology which is being powered by the approach described in this article. To date, the system has survived the Blizzard of '96 and six months of some of the most "interesting" Connecticut weather in years.

Author Information
About the authors
Duane Elms is a director at SNET responsible for, among other things, developing the powering approach for SNET's HFC system. Tom Osterman is president of Comm/net Systems Inc., a broadband power system integrator, distributor and engineering consulting firm, and CEO of Millennium Power Inc., a power conversion system manufacturer.