Cable Interference and Noise Issues

Electromagnetic flux is a potential problem that can disrupt network communications wherever there are active electrical and electronic devices. The selection of the right cabling and its network routing design is important to reduce communications interference problems. All network components, including the connectors and patch panels, must be designed to satisfactorily perform in the presence of external noise. Cable routing should conform to the manufacturer’s recommendations and always avoid potential interference sources. Likely office building sources of EMI are lift motors (elevators), automatic doors, and air-conditioning units. The older the equipment, the more likely it will produce EMI. Closed metal conduits and ducting for the cabling system will provide extra protection against EMI sources that cannot be corrected or avoided. Balanced transmission over UTP cable offers strong protection against external noise. In EMI-sensitive or hostile environments, the only solution may be optical fiber cable that is immune to external noise.

There are regulations specified by the FCC (part 68 and part 15 subpart) that cover telecommunications network electromagnetic compatibility (EMC) with other electronic devices. Network system installers and users are responsible for conforming to EMC guidelines. Installers must ensure that cable specifications for routing and ducting eliminate interference problems. Some manufacturers provide warranties on the EMC performance of certified installations using their cabling.

In addition to the potential for interference from external electrical and electronic source devices, the active pairs in a multipair cable can interfere with each other. Interference between cable pairs is known as crosstalk. Crosstalk measurements may be performed with two methods: pair-to-pair and PowerSum. The pair-to-pair method measures only the maximum interference caused by any other single active cable pair. Near end cross talk (NEXT), the pair-to-pair measurement metric, is defined as the signal coupled from one pair to another in a UTP cable. It is called NEXT because it measures the crosstalk at the end where one pair is transmitting (and the transmitted signal is largest and, hence, causes the most crosstalk). Crosstalk is minimized by the twists in the cable, with different twist rates causing each pair to act as antennas sensitive to different frequencies so that signals are not picked up from neighboring pairs. Keeping the twists as close as possible to the terminations minimizes crosstalk. Far end crosstalk (FEXT) measures the effect of signal coupling from one pair to another over the entire length of the cable, and it is measured at the far end.

Another frequently cited measurement associated with crosstalk is the attenuation to crosstalk (ACR) ratio. Attenuation is the reduction in signal strength due to loss in the cable. ACR measures how much “headroom” the signal should have at the receiver. It is important that the signal strength at the receiver end be high enough for reception by the network hub/switch to pass through to workstation nodes or other hubs/switches. Ethernet LANs send very high-speed signals through the cable, and the attenuation varies with the frequency of the signal. Attenuation tests are performed at several wavelengths, as specified in the 568 standards. The test requires a tester at each cable end, one to send and one to receive. The loss between the ends is calculated, recorded, and compared with pass/fail criteria for UTP cable at Category 3, 4, and 5 frequencies.

Performance losses can be greater than indicated by pair-to-pair measurement if there are several active pairs in a multipair cable strand. For this reason, the preferred method of measuring crosstalk is known as PowerSum. It is based on the measurements taken when all pairs in a multipair cable are active. This is the more realistic crosstalk measurement for Fast Ethernet and Gigabit Ethernet LANs, where all pairs are used to carry signals, often simultaneously. PowerSum is the recommended method to use for cables with more than four wires.

Cabling System Fundamentals

A structured cabling architecture design is intended to accommodate telecommunications technology changes with minimal impact on any of the other cabling subsystems, such as, electrical cabling. The target life cycle of an average cabling installation is up to 20 years. It is expected that a few generations of telecommunications systems will be installed and replaced or upgraded. Another planning assumption is that networking and bandwidth requirements will certainly increase during the life cycle of the cabling system. The following are key factors used to specify networks and cabling, as identified by Avaya in its SYSTIMAX CSC guidebook:

• Usage patterns, including combined size and duration of peak loads for all applications

• Expected increase in bandwidth demands

• The number of users and anticipated changes in that number

• Location of users and maximum distances between them

• The likely rate of change in users’ locations (churn)

• Connectivity with current and future devices and software

• Space available for cable runs

• Total cost of ownership

• Regulations and safety requirements

• Importance of protection against loss of service and data theft

PBX systems traditionally have been based on a star network topology. A star network topology includes many point-to-point links radiating from central equipment. The early LAN topologies were based on ring-and-bus network designs. A ring network topology has a continuous transmission loop that interconnects every device. The most familiar example of a ring network topology is the IBM token ring LAN. A bus network topology is a communications link that connects devices along the length of a cable. The original Ethernet LAN was based on a bus network topology.

Today’s dominant LAN technology is based on Ethernet standards. The logical topology of an Ethernet LAN is a bus topology, but the physical topology of the network is a star. Ethernet workstations that connect to an Ethernet hub or switch communicate over a high-speed bus housed in a hub or switch, but these network nodes are connected in a clustered star network topology. The star topology favored by PBX systems and adapted by Ethernet LANs is now the accepted communications system network topology.

The first Ethernet LAN installations were based on coaxial cable used for the transmission medium. During the mid-1980s the cabling used by PBX systems, known as unshielded twisted pair (UTP), was adapted for Ethernet LANs. Telephony UTP cabling was classified by IBM’s cabling system specifications as Category 3 and was used for 10Base-T Ethernet LANs operating at 10 Mbps. A 10Base-T Ethernet LAN used two pairs of Category 3 UTP cabling. A 100Base-T4 Ethernet LAN used four-pair Category 3 UTP cabling. A 100-Mbps Fast Ethernet, also known as 100Base-TX, used two-pair Category 5 UTP cabling. The 1000-Mbps (1 Gbps) Ethernet, 1000BASE-T, uses four-pair Category 5 UTP cabling. The 1000Base-TX, a lower cost alternative to 1000Base-T, uses the recently introduced Category 6 UTP cabling. PBX system telephony requirements can be satisfied with any of these UTP cabling types, making possible a single network cabling system infrastructure for voice and data communications applications.

In the SYSTIMAX SCS guidebook, Avaya lists the following considerations for choosing the type of customer network cabling:

• Maximum distance between network hubs and nodes

• Space available in ducting and floor/ceiling cavities

• The levels of electromagnetic interference (EMI)

• Likely changes in equipment served by the system and the way it is used

• Level of resilience required

• The required life span of the network

• Restrictions on cable routing that dictate cable bend radius

• Existing cable installations with potential for reuse

For the past two decades, most customers have used or installed two different cabling systems for telephony and data LAN applications. The evolution of the PBX system to an IP telephony platform will allow the large installed base of customers with installed circuit switched PBX systems to slowly phase out an infrastructure with two cabling systems and allow customers who are designing an entirely new converged voice/data network the opportunity to install a single cabling system. PBX systems installed before 1990 were implemented with Category 3 UTP, but more recent installations may have been based on Category 5 UTP, the same wiring used for data LANs. A newly installed communications system installation likely would be based on a generic cabling infrastructure using Category 5 UTP to provide for future needs.

A generic cabling system is a structured telecommunications cabling system capable of supporting a wide range of customer applications. Generic cabling can be installed before the definition of required applications because application-specific hardware (telephones, computers, etc.) is not part of the structured cabling design. Generic cabling can be enhanced through the use of flood wiring, which is the installation of sufficient cabling and telecommunications outlets in a work area to maximize the flexibility of the location for devices connected to the network. Many customers are currently installing four or six telecommunications outlets per work area, although the recommended minimum is two.

PBX Cabling Guidelines

Telephony wiring dates back 125 years ago, to the days when Alexander Graham Bell was tinkering with the first telephone. Telephones traditionally have used loop current for voice communications and signaling transmission. For many years single-pair (two-wire) cabling had supported telephones working behind a PBX system, but system equipment innovations, beginning with the introduction of digital switching and stored program call control, forced changes in the cabling infrastructure during the late 1970s. The first generation of proprietary PBX telephones, first electronic and then digital, required multiple wiring pairs to support the more advanced features and functions available with the new technology. At the same time, the early data LANs required a wiring infrastructure of their own, based on coaxial cable. As customer premises voice networks and data networks evolved in the mid-1980s, issues such as a common infrastructure and increasing transmission bandwidth requirements needed to be addressed. The existing telephony wiring system, fine for voice but inadequate for data, needed a major overhaul.

In 1985, two standards committees began working on specifications for a generic telecommunications cabling system to support a mix of communications media (voice, data, video) in a multivendor environment. The TIA and the Electronic Industries Association (EIA) formed a joint committee known as the EIA/TIA 41.8 Committee. After 6 years of work, the TIA/EIA 568 standard was issued. TIA/EIA 568 is more formally known as the Commercial Building Cabling Standard and outlines specifications for a generic telecommunications cabling system. The American National Standards Institute (ANSI) also adapted this standard, so it is sometimes referred to as ANSI/TIA/EIA 568.

There is a corresponding series of specifications known as ANSI/TIA/EIA 569: Commercial Building Standard for Telecommunications Pathways and Spaces. The purpose of ANSI/TIA/EIA 569 is to standardize design and construction practices within and between buildings that support telecommunications equipment and transmission media. The standards are outlined for rooms or areas and pathways into and through areas where telecommunications media and equipment are installed. To simplify the implementation and administration of the cabling infrastructure, another series of specifications were developed, ANSI/TIA/EIA 606: The Administration Standard for the Telecommunications Infrastructure of Commercial Building.

In addition to the standards specified by the ANSI/TIA/EIA recommendations, the International Standards Organization (ISO) defined a generic cabling system recommendation known as ISO/IEC IS 11801. The ISO standard is intended for global usage and is broader in scope than the ANS/TIA/EIA standards for the North American market. The European version of ANSI/TIA/EIA standard is EN 50173 and is more similar to 568 than to the ISO standard.

Cabling Categories

The use of UTP copper wiring for multimegabit-per-second, in-building data networks was first proposed in the mid-1980s and became generally accepted after the Category 5 standards were defined in 1994. Category 5 (Cats) cables usually have four twisted pairs of 24-AWG copper wire and are terminated with eight-wire RJ45 miniature connectors. This cable delivers 100-Mbps Ethernet bitstreams up to a distance of l00m (325 ft) between the electronic equipment (e.g., a switch) and the desktop device. Typically, cabling system manufacturers will provide guarantees or "certification" of the performance of their cabling materials if the cable plant in a new installation has been designed and installed correctly.

We cannot be sure that full Cat5 performance will be delivered if the in-line components, as identified in Figure 1, between the desktop computer (or IP phone) and the server come from different cable and equipment manufacturers. For this reason cabling implementations should follow best practices for design and implementation, according to the manufacturer's specifications, and those outlined in the EIA/TIA Building Telecommunications Wiring Standards (to which the cabling performance specification must abide).


Figure 1: End-to-end components with a LAN.

Category 5 Enhanced (Cat5E) standards were defined by the telecom industry in early 2000. Cat5E performance has been so improved over the earlier standard that a string of Cat5E components will deliver guaranteed Cat5 performance, even if they come from a variety of manufacturers.

Category 6 (Cat6) cabling has been on the market, from several manufacturers, for a few years. It offers a 10 times improved performance bit rate over Cat5 (i.e., supporting 1,000 Mbps, or gigabit, Ethernet). Most applications do not justify gigabit Ethernet to the desktop, but it is used in LAN backbones and for connection to network servers. Standardization of the Cat6 specifications was due to take place by the end of 2002. Until that happens, there is some risk with using the Cat6 class of cabling and mixed-vendor implementations should certainly be avoided before there is industrywide acceptance of Cat6 standards.

Category 7 (Cat7) is a different class of cabling, because it includes a metallic mesh shield around the twisted copper pairs, underneath the outer layer of plastic insulation. Cat7 cabling was developed in Europe and is being used in several countries there, particularly in Germany, for highcapacity LAN installations.

Over the past 5 years many corporate LANs in North America have been built with Cat5E cabling, and a limited number of Cat6 infrastructures are now in place, in spite of its lack of standardization. Our experience has been that there is no more than a 10% cost differential between quotations for Cat5E and Cat6 implementations. One reason for this is that the actual cable materials make up well under one-half of the cost of a complete wiring installation.

Physical LAN costs have not changed much over the past decade, at around $200 per outlet for Cat6 (or about $180 with Cat5E), including needs analysis, system design, hardware, cable, cable-pulling, connecting, and testing, in small-to-medium installations (e.g., a school or a low-rise suburban office building). Expect cabling infrastructure costs to be above $200 per outlet in high-rise, downtown office towers, especially if installation is controlled by unionized workers.