Mitigating EMI in High-Speed Digital Transmission Networks, Part I

 

Classically, the network cabling that provides residential or commercial telecommunication access into a building has consisted of copper wires formed into unshielded and shielded twisted pairs (UTP and STP, respectively). This wiring runs from the curb (where the subscriber-loop carrier terminal is located) to a pedestal terminal and then into the building, where a standard jack-and-plug arrangement (such as RJ-11 or RJ-45) is used to connect it with a telephone or data terminal. When voice telephony was the major telecommunication service, the network cabling within a building was usually straightforward, although a variety of wiring systems evolved to accommodate the different interfaces needed by the host of service providers and a mix of wire sizes (ranging from two to 25 pair) was not uncommon. In such voice-alone systems, any impairments in the quality of transmission arising from the vagaries and variety of wiring could be tolerated.

Figure 1: Schematic showing (a) downstream and (b) upstream access network topology of ADSL systems. (ATU-C = the terminal unit at the central office; ATU-R = the terminal a the receiving end.)

With the advent of data communication, however, transmission quality became a significant concern, which led to the development of standards for a structured wiring system.¹ EIA/TIA-570, the wiring standard for re sidential buildings, and EIA/TIA-568, 569, 606, and 607, which cover commercial buildings, are aimed at the careful choice of materials, proper manufacturing, and high-quality cable installation so that both voice and data transmission in the 4-kHz bandwidth can be supported for the distances encountered in premise wiring.

Within these standards, the associated wiring basically refers to UTP and STP copper wire, but can include other media such as optical fibers and coaxial lines.

The standards cited above were conceived for the 4-kHz bandwidth used for telephone and voice-band modem communications. However, the new communication services that are or will be offered to subscribers at home and in business environments require the transmission of high-speed digital information from the curb into the building. Thus, questions arise regarding whether existing UTP cabling can support the broad bandwidths associated with these new services or whether wiring and related connective hardware will need to be replaced.

One viable and pragmatic way to support high-speed digital information access into UTP-based premise wiring is the technology known as asymmetric digital subscriber lines (ADSL).^2,3 Many operators and subscribers are currently pinning their hopes on ADSL, which analysts predict will be installed in the millions in coming years.² Shown schematically in Figure 1, this system has the capability to increase significantly the transport capacity of the embedded twisted-pair cable infrastructure. It is asymmetric in the sense that it supports high bit rates ( 6 Mb/sec and 64 Kb/sec) in the downstream transmission (from the service provider's central office [CO] to the building) and lower bit rates (384–576 Kb/sec and 64 Kb/sec) in the upstream transmission. These up- and downstream transmissions of digital information are done without any impairment to the conventional voice transmission supported by the twisted-pair cable. In addition to ADSL, the cohabited use of voice and high-speed, broadband transmissions may also be achieved with installations that support the B-ISDN protocol, namely, the asynchronous transfer mode.

In the context of achieving high-speed, broadband access by implementing ADSL with existing UTP premise wiring, the remainder of this article addresses the following queries:

In essence, it is necessary to identify the EMI problems associated with UTP-based premise wiring in order to investigate their mitigation during ADSL implementation as well as to decide on future-proofing strategies to support a growing number of high-speed, broadband services. The role of novel material-based schemes to achieve the required telecommunications performance can then be analyzed with reference to EMI shielding considerations.

 

Typically, the structured wiring in a residential or commercial building consists of some or all of the following:

Except for the power cabling, the above components constitute the telecommunications wiring. TIA/EIA 570A, the revised residential cabling standard that currently awaits final approval, categorizes this wiring into two grades. For Grade I wiring, which provides basic telephone and video services, the standard recommends the use of one four-pair Category 3 (or better) UTP cable and one RG-6 coaxial cable to each information outlet. For Grade 2 wiring, which provides enhanced voice, video, and data services, the standard recommends two four-wire Category 5 UTP cables and two RG-6 coaxial cables. One of the Category 5 cables is intended for voice transmission and the other is for data, while one of the RG-6 cables is for use with a satellite dish and the other is for local programming via an antenna or a cable TV connection.

Notwithstanding the new Grade 2 specifications, which are intended to future-proof wiring, the concept behind ADSL technology is to make use of the existing UTP copper premise wiring for high-speed, broadband access, despite the EMI-induced impairments to transmissions that may be expected.⁴

The sources of EMI in copper-based digital transmission systems include crosstalk, impulse noise, radio-frequency interference, and thermal or system noises.

 

A number of diverse sources can generate what is known as impulse noise. These include lightning strikes on switching equipment, power-line transients, and motors and other devices within a building. Such noise is a major concern in ADSLs inasmuch as the received signal on the high-speed data line may be weakened by heavy subscriber-loop losses. In addition, the origin of impulse noise is geographically variable, which makes it difficult to locate. Impulse noise is often characterized by a random-pulse waveform with an amplitude that is much larger than the system noise. Other typical characteristics include an occurrence rate of one to five pulses per minute, peak amplitudes in the range of 5–20 mV, spectral energy concentrated within 40 kHz, and an impulse duration of 30–150 µs.

The lines may also act as antennas and pick up external radiated interference. Radio signals usually do not impose a serious EMI threat to DSL lines. However, if there are strong radio transmitters in the vicinity, it may be necessary to place RFI filters on the lines.

Figure 2: Examples of NEXT and FEXT on (a) downstream and (b) upstream access lines.

NEXT and FEXT pose more serious EMI problems to DSL transmission systems. One concern is the possible corruption of signal waveforms that could seriously affect the logic elements in the digital transmission. These waveforms are known as critical signals. In high-speed DSLs, crosstalk can cause the false triggering of critical signals that exceed the voltage thresholds of prescribed logical states, which can impair synchronization significantly as well as increase the bit-error rate. In both NEXT and FEXT, the pair-to-pair coupling is related to the magnitude of the ratio (in decibels) of the voltage induced at the near or far end of the susceptible cable to the source voltage. This coupling also depends on the mutual- and self-inductance of the cable pairs and the capacitive bridge across them. A suggested technique for reducing this coupling is illustrated in Figure 3.

Figure 3: A suggested method for reducing crosstalk between cable pairs.

The origin and extent of NEXT and FEXT in the subchannel sections of an ADSL access network (shown in Figure 1) are illustrated in Figure 2. NEXT is independent of the length of the cable, whereas FEXT is attenuated almost to the same extent as the signal along the length of the cable.⁴ The coupling coefficient for NEXT can be approximated using the formula [–55 + 10 log₁₀(f/100 kHz)] dB. At the far end of the lines, this coefficient is lower and is given by [–50+10 log₁₀ (1/5 km)+20 log₁₀ (f/100 kHz)].

Self-NEXT can be a problem in the upstream ADSL lines, so simultaneous transmission and reception in overlapping frequency bands and overlapping time intervals should be avoided. FEXT is a concern in downstream transmissions, where cross-coupling of data can occur between two distinct subscriber lines. Although the level of such interference would be low (because of the attenuation along the line), its role as an added noise at the subscribers' reception equipment can lead to significant performance impairment.

The extent of FEXT is also frequency dependent and may increase significantly over a frequency range of 1 MHz. In ADSL, the capacities of the downstream subchannels that share a common cable environment will be weighted by the spread of FEXT versus frequency. Therefore, the spectral compatibility (i.e., the frequency plan) of the subchannels should be coordinated carefully, with consideration given to crosstalk effects.

The use of twisted pair, end-to-end grounding and the systematic separation of the directional channels (duplexing) into discrete frequency bands and time domains are the crosstalk reduction methods normally recommended for DSL networking. Both techniques are based on circuit theory and traditional communications technology. Other mitigation measures based on cable products and materials are described next.

 

The essential cabling products used in indoor telecommunication wiring can be grouped into six categories: UTP wires, STP wires, coaxial cables, connection devices, boxes and outlets, and distribution and routing equipment. The basic UTP wiring unit consists of four (or more) pairs of insulated conductors housed in a flame-retardant PVC jacket, which also contains a longitudinally applied rip cord and optimal center fillers. It may be rated for plenum or nonplenum use. Cables with the former rating are UL-approved, heat-resistant, low-toxic, low-smoke-emitting products suitable for use in heating ducts. In STP cables, each wire is entirely or partially wrapped with metallic braid or foil that provides shielding. The components, materials, and uses of these two types of cables and of the items in the other product categories are presented in Table I; accessories used to achieve EMI shielding are presented in Table II. The choice of materials for use in these products is based on such factors as electrical conductivity, electrical insulation properties, complex permittivity properties vis-à-vis the frequency range of operation, thermal durability (plenum rating), mechanical properties such as tensile and compression strength, and EMI shielding capabilities. Traditionally, steel, aluminum, and thermoplastics have been used to achieve the desired mechanical and structural properties. Annealed copper and beryllium copper are the classical conducting materials, while PVC, PE, PEP, Teflon, polyolefin, and fluoropolymer are used as insulating materials, and copper and aluminum are used as shielding materials. The summaries of material utilization in today's cabling products provided in the tables indicate that these materials are still used to realize the required functional characteristics.

Most new cable products and shielding accessories, however, are based on metals and insulator materials that have been modified so that they pose lossy dielectric relaxation to EMI energy.⁶ The lossy insulating materials are largely proprietary, but a variety of polymeric compounds can be either surface-treated or volume-loaded to realize such dielectric characteristics. The designing of composite materials to achieve EMI suppression over given frequency bands is beyond the scope of this discussion but has been described elsewhere by one of the authors.^7–9

As described earlier, the pair-to-pair coupling can be combated by using end-to-end single-point grounding (see Figure 3). This technique is effective for reducing differential-mode coupling. However, conduction coupling can also be induced between lines as a result of common-mode voltage levels fed across the ground-loop impedance. This ground-loop coupling is a function of frequency; it increases monotonically as frequencies increase and drops in a resonating manner when the cable length exceeds 1/4 to 1/2 wavelength. This common-mode menace can be reduced by increasing the ground path impedance or adopting a double enclosure, "shield-case-within-a-shield," technique. The conventional solution is to use RF chokes in the path from the case to ground. The RF chokes offer high impedance around the ground loop, but would prevent the low ground return required for ac power supply (hazardproof) grounding. Therefore, the double-enclosure method is preferable.

In this design, the signal reference plane is grounded to the inner shield, and the outer shield is grounded directly without developing a low-impedance ground loop. The ground-loop coupling reduction is determined by the shield-to-shield capacitance (C_S ), which depends on the surface area of the shields, the spacing between the enclosures, and the material characteristics of the inner surfaces of the enclosures. System designers can control all three of these factors and can maximize shielding effectiveness by using a composite structure of metal plus a dielectric material for at least one of the enclosures. In the example of double shielding enclosures illustrated in Figure 4, a composite material is used for the outer case, and the enclosures are installed at the source and terminal ends of the subscriber line.

Figure 4: The shield-case-within-a-shield method for reducing ground- loop coupling. In this example, the outer case (A) is a composite material, and the inner case (B) is a conventional shielding material. (Cs= the shield-to-shield capacitance.)

It is believed that composite materials can also be used for other components of DSL systems to provide protection against crosstalk and other types of EMI. The scope of such efforts is, however, still an open question.

Conclusion

As the demand for sophisticated telecommunications services has increased, techniques such as ADSL have evolved to support the necessary high-speed, broadband transmissions by using existing inside wiring. EMI, particularly pair-to-pair crosstalk on UTP wiring, is among the challenges to be overcome. Thus, the future-proofing of residential and commercial building cabling will depend on the development of insulating materials with lossy dielectric properties.

Part II of this article will discuss the EMI shielding of base-station equipment and the associated material considerations pertinent to cabling and equipment housings.