Thursday, April 28, 2011

External Interference | Data Cabling

One hindrance to transmitting data at high speed is the possibility that the signals traveling through the cable will be acted upon by some outside force. Although the designer of any cable, whether it's twisted-pair or coaxial, attempts to compensate for this, external forces are beyond the cable designer's control. All electrical devices, including cables with data flowing through them, generate EMI. Low-power devices and cables supporting low-bandwidth applications do not generate enough of an electromagnetic field to make a difference. In addition, some equipment generates radio-frequency interference; you may notice this if you live near a TV or radio antenna and you use a cordless phone.
Devices and cables that use a lot of electricity can generate EMI that can interfere with data transmission. Consequently, cables should be placed in areas away from these devices.
Some common sources of EMI in a typical office environment include the following:
  • Motors
  • Heating and air-conditioning equipment
  • Fluorescent lights
  • Laser printers
  • Elevators
  • Electrical wiring
  • Televisions
  • Some medical equipment
Talk about electromagnetic interference! An MRI (magnetic resonance imaging) machine, which is used to look inside the body without surgery or x-rays, can erase the magnetic strip on a credit card from 10 away.
When running cabling in a building, do so a few feet away from these devices. Never install data cabling in the same conduit as electrical wiring.
In some cases, even certain types of businesses and environments have high levels of interference, including airports, hospitals, military installations, and power plants. If you install cabling in such an environment, consider using cables that are properly shielded, or use fiber-optic cable.

Monday, April 25, 2011

Alien Crosstalk (AXT)

Alien crosstalk (AXT) occurs when the signal being carried in one cable interferes with the signal being carried in another cable. This can occur in a cable that runs alongside one or more signal-carrying cables. The term alien arises from the fact that this form of crosstalk occurs between different cables in a bundle, rather than between individual wire pairs within a cable.

Alien crosstalk can be a problem because, unlike the simpler forms of crosstalk that take place within a single cable, it cannot be eliminated by traditional phase cancellation. Because AXT resembles noise rather than signals, alien crosstalk degrades the performance of the cabling system by reducing the signal-to-noise ratio of the link. As the signal rate increases in a cable, this form of crosstalk becomes more important. In fact, this is a major source of interference, and a limiting factor, for running 10GBase-T (10Gbps) over UTP cabling. A lot of work has been performed during the creation of the ANSI/TIA-568-C standard in understanding the causes of AXT and potential solutions.
Alien crosstalk can be minimized by avoiding configurations in which cables are tightly bundled together or run parallel to one another in close proximity for long distances. In a typical installation, however, this is difficult and impractical. Category 6A (augmented Category 6) cable tries to solve this problem by increasing the spacing between wire pairs in a cable using separators within a cable to space the conductors apart from one another. This has the added effect of separating the conductors in one cable from the conductors in another. As you can imagine, this increases the diameter of a Category 6A cable compared to a Category 6 cable.
Another recommendation for reducing AXT is to avoid using tie-wraps to bundle cable together and to try to separate the cables in a rack as much as possible. This in turn requires more space to run these cables.
Recently developed Category 6A cables use a special core wrap that is not electrically continuous, so it does not require grounding, but that isolates and protects the core from alien crosstalk and other forms of external interference (as you'll see in a moment). These cables can be routed and bundled like traditional UTP cables without the concern of AXT and their size is smaller as well.
The industry has created measurement methods to measure alien crosstalk in the field, but they are very time consuming. The best advice is to ensure all the components are verified to be Category 6A compliant and that they have been tested in a channel or permanent link configuration to work together.

Friday, April 22, 2011

Pair-to-Pair Crosstalk & Power-Sum Crosstalk

Pair-to-Pair Crosstalk

For both near-end crosstalk and far-end crosstalk, one way of measuring crosstalk is the pair-to-pair method. In pair-to-pair measurement, one pair, the disturber, is energized with a signal, and another pair, the disturbed, is measured to see how much signal transfer occurs. The following six combinations are tested in a four-pair cable:
  • Pair 1 to pair 2
  • Pair 1 to pair 3
  • Pair 1 to pair 4
  • Pair 2 to pair 3
  • Pair 2 to pair 4
  • Pair 3 to pair 4
The test is repeated from the opposite end of the cable, resulting in 12 pair-to-pair combinations tested. The worst combination is what is recorded as the cable's crosstalk value. See Figure 1.
Figure 1: Cutaway of a UTP cable, showing pair-to-pair crosstalk

Power-Sum Crosstalk

Power-sum crosstalk also applies to both NEXT and FEXT and must be taken into consideration for cables that will support technologies using more than one wire pair at the same time. When testing power-sum crosstalk, all pairs except one are energized as disturbing pairs, and the remaining pair, the disturbed pair, is measured for transferred signal energy. Figure 2 shows a cutaway of a four-pair cable. Notice that the energy from pairs 2, 3, and 4 can all affect pair 1. The sum of this crosstalk must be within specified limits. Because each pair affects all the other pairs, this measurement will have to be made four separate times, once for each wire pair against the others. Again, testing is done from both ends, raising the number of tested combinations to eight. The worst combination is recorded as the cable's power-sum crosstalk.

Figure 2: Power-sum crosstalk

Tuesday, April 19, 2011

NEXT, FEXT, ELFEXT | Types of Crosstalk

Near-End Crosstalk (NEXT)

When the crosstalk is detected on the same end of the cable that generated the signal, then near-end crosstalk has occurred. NEXT is most common within 20 to 30 meters (60 to 90 feet) of the transmitter. Figure 1 illustrates near-end crosstalk.
Figure 1: Near-end crosstalk (NEXT)
Crosstalk on poorly designed or poorly installed cables is a major problem with technologies such as 10Base-T and 100Base-TX. However, as long as the cable is installed correctly, NEXT is less of an issue when using 1000Base-T because the designers implemented technologies to facilitate NEXT cancellation. NEXT-cancellation techniques with 1000Base-T are necessary because all four pairs are employed for both transmitting and receiving data.
Cables that have had their twists undone (untwisted) can be problematic because the twists help cancel crosstalk. Twists are normally untwisted at the ends near the patch panels or connectors when the cable is connected. On the receiving pair of wires in a cable, the signal received at the end of the cable will be the weakest, so the signal there can be more easily interfered with. If the wires on adjacent transmit pairs are untwisted, this will cause a greater amount of crosstalk than normal. A cable should never have the wire pairs untwisted more than 0.5 for Category 5e, and 0.375 maximum for Category 6 cables.

Far-End Crosstalk (FEXT)

Far-end crosstalk (FEXT) is similar to NEXT except that it is detected at the opposite end of the cable from where the signal was sent. Due to attenuation, the signals at the far end of the transmitting wire pair are much weaker than the signals at the near end.
The measure of FEXT is used to calculate equal-level far-end crosstalk (ELFEXT). More FEXT will be seen on a shorter cable than a longer one because the signal at the receiving side will have less distance over which to attenuate.

Equal-Level Far-End Crosstalk (ELFEXT)

Equal-level far-end crosstalk (ELFEXT) is the crosstalk coupling between cabling pairs measured at the end of the cable opposite to the end of the signal source, taking into account signal loss. ELFEXT is calculated, not measured, by subtracting the attenuation of the disturber pair from the far-end crosstalk (FEXT) on the disturbed pair. The calculation describes the ratio of disturbance to the level of the desired signal; it is another indication of signal-to-noise ratio. Another way of looking at it is that the value represents the ratio between the strength of the noise due to crosstalk from end signals compared to the strength of the received data signal. You could also think of ELFEXT as far-end ACR (attenuation-to-crosstalk ratio, described later in this chapter).
Each pair-to-pair combination is measured, as the attenuation on each pair will be slightly different. If the ELFEXT value is very high, it may indicate that either excessive attenuation has occurred or that the far-end crosstalk is higher than expected.

Sunday, April 10, 2011

Noise (Signal Interference) | Speed Bumps

Everything electrical in the cable that isn't the signal itself is noise and constitutes a threat to the integrity of the signal. Many sources of noise exist, from within and outside the cable. Controlling noise is of major importance to cable and connector designers because uncontrolled noise will overwhelm the data signal and bring a network to its knees.

Twisted-pair cables utilize balanced signal transmission. The signal traveling on one conductor of a pair should have essentially the same path as the signal traveling the opposite direction on the other conductor. (That's in contrast to coaxial cable, in which the center conductor provides an easy path for the signal but the braid and foil shield that make up the other conductor are less efficient and therefore a more difficult pathway for the signal.)
As signals travel along a pair, an electrical field is created. When the two conductors are perfectly symmetrical, everything flows smoothly. However, minute changes in the diameter of the copper, the thickness of the insulating layer, or the centering of conductors within that insulation cause disturbances in the electrical field called unbalances. Electrical unbalance means noise.

Resistance unbalance occurs when the dimensions of the two conductors of the pair are not identical. Mismatched conductors, poorly manufactured conductors, or one conductor that got stretched during installation will result in resistance unbalance.

Capacitance unbalance is also related to dimensions, but to the insulation surrounding the conductor. If the insulation is thicker on one conductor than on the other, then capacitance unbalance occurs. Or, if the manufacturing process is not well controlled and the conductor is not perfectly centered (like a bull's-eye) in the insulation, then capacitance unbalance will exist.

Both these noise sources are usually kept well under control by the manufacturer and are relatively minor compared to crosstalk.

You've likely experienced crosstalk on a telephone. When you hear another's conversation through the telephone, that is crosstalk. Crosstalk occurs when some of the signal being transmitted on one pair leaks over to another pair.

When a pair is in use, an electrical field is created. This electrical field induces voltage in adjacent pairs, with an accompanying transfer of signal. The more parallel the conductors, the worse this phenomenon is, and the higher the frequency, the more likely crosstalk will happen. Twisting the two conductors of a pair around each other couples the energy out of phase (that's electrical-engineer talk) and cancels the electrical field. The result is reduced transfer of signal. But the twists must be symmetrical; i.e., both conductors must twist around each other, not one wrapping around another that's straight, and two adjacent pairs shouldn't have the same interval of twists. Why? Because those twist points become convenient signal-transfer points, sort of like stepping-stones in a stream. In general, the shorter the twist intervals, the better the cancellation and the less crosstalk. That's why Category 5e and higher cables are characterized by their very short twist intervals.

Crosstalk is measured in decibels; the higher the crosstalk value, the less crosstalk noise in the cabling. See Figure 1

Figure 1: Crosstalk

Wednesday, April 6, 2011

Attenuation (Loss of Signal) | Speed Bumps

As noted earlier, attenuation is loss of signal. That loss happens because as a signal travels through a cable, some of it doesn't make it all the way to the end of the cable. The longer the cable, the more signal loss there will be. In fact, past a certain point, the data will no longer be transmitted properly because the signal loss will be too great.

Attenuation is measured in decibels (dB), and the measurement is taken on the receiver end of the conductor. So if 10dB of signal were inserted on the transmitter end and 3dB of signal were measured at the receiver end, the attenuation would be calculated as 3  10 = 7dB. The negative sign is usually ignored, so the attenuation is stated as 7dB of signal loss. If 10dB were inserted at the transmitter and 6dB measured at the receiver, then the attenuation would be only 4dB of signal loss. So, the lower the attenuation value, the more of the original signal is received (in other words, the lower the better).

Figure 1 illustrates the problem that attenuation causes in LAN cabling.

Figure 1: The signal deteriorates as it travels between a node on a LAN and the hub

Attenuation on a cable will increase as the frequency used increases. A 100-meter cable may have a measured attenuation of less than 2dB at 1MHz but greater than 20dB at 100MHz!

Higher temperatures increase the effect of attenuation. For each higher degree Celsius, attenuation is typically increased 1.5 percent for Category 3 cables and 0.4 percent for Category 5e cables. Attenuation values can also increase by 2 to 3 percent if the cable is installed in metal conduit.

When the signal arrives at the receiver, it must still be recognizable to the receiver. Attenuation values for cables are very important.

Attenuation values are different for the categories of cables and the frequencies employed. As the bandwidth of the cable increases, the allowed attenuation values get lower (less loss), although the differences between Category 5e and 6 are negligible at the common frequency of 100MHz.

Characteristics that contribute to attenuation are detailed as follows:
  • Conductor resistance Conductor resistance acts as a hindrance to the signal because it restricts the flow of electricity through the cable conductors. This causes some of the signal energy to be dissipated as heat, but the amount of heat generated by LAN cabling is negligible due to the low current and voltage levels. The longer the cable or the smaller the conductor diameters (actually, the cross-sectional area), the more resistance. After allowing for dimensional factors, resistance is more or less a fixed property of the conductor material. Copper, gold, and silver offer low resistance and are used as conductors.

  • Mutual capacitance This characteristic is an electrical occurrence experienced when a cable has more than one wire and the wires are placed close together. The insulation material will steal and store some of the signal energy, acting as a capacitor between two conductors in the cable. A property of the insulating material called dielectric constant has a great influence over the mutual capacitance. Different materials have different dielectric constants. The lower the dielectric constant, the less signal loss. FEP and HDPE have low dielectric constants, along with other properties, that make them well suited for use in high-frequency cables.

  • Impedance Impedance is a combination of resistance, capacitance, and inductance and is expressed in ohms; a typical UTP cable is rated at between 85 and 115 ohms. All UTP Category 3, 5e, 6, and 6A cables used in the United States are rated at 100 + 15 ohms. Impedance values are useful when testing the cable for problems, shorts, and mismatches. A cable tester could show three possible impedance readings that indicate a problem:

    • An impedance value not between 85 and 115 ohms indicates a mismatch in the type of cables or components. This might mean that an incorrect connector type has been installed or an incorrect cable type has been cross-connected into the circuit.
    • An impedance value of infinity indicates that the cable is open or cut.
    • An impedance value of 0 indicates that the cable has been short-circuited.
Some electrons sent through a cable may hit an impedance mismatch or imperfection in the wire and be reflected back to the sender. Such an occurrence is known as return loss. If the electrons travel a great distance through the wire before being bounced back to the sender, the return loss may not be noticeable because the returning signal may have dissipated (due to attenuation) before reaching the sender. If the signal echo from the bounced signal is strong enough, it can interfere with ultra-high-speed technologies such as 1000Base-T.

Saturday, April 2, 2011

Hindrances to High-Speed Data Transfer | Speed Bumps

Electricity flowing through a cable is nothing more than electrons moving inside the cable and bumping into each other—sort of like dominoes falling. For a signal to be received properly by the receiver, enough electrons must make contact all the way through the cable from the sender to the receiver. As the frequency on a cable (and consequently the potential data rate) increases, a number of phenomena hinder the signal's travel through the cable (and consequently the transfer of data). These phenomena are important not only to the person who has to authorize cable purchase but also to the person who tests and certifies the cable.

The current specifications for Category 5e, 6, and 6A cabling outline a number of these phenomena and the maximum (or minimum) acceptable values that a cable can meet and still be certified as compliant.
Due to the complex modulation technology used by 1000Base-T Ethernet, and even more so with 10GBase-T, the TIA has specified cabling performance specifications beyond what was included in the original testing specification. These performance characteristics include power-sum and pair-to-pair crosstalk measurements, delay skew, return loss, and ELFEXT. Some of these newer performance characteristics are important as they relate to crosstalk—for example, AXT (alien crosstalk) to express the interaction between cables in a cable bundle. Although crosstalk is important in all technologies, faster technologies such as 1000Base-T and 10GBase-T are more sensitive to it because they use all four pairs in parallel for transmission.
All these requirements are built into the current version of the standard, ANSI/TIA-568-C.

Many transmission requirements are expressed as mathematical formulas. For the convenience of humans who can't do complex log functions in their heads (virtually everyone!), values are pre-computed and listed in the specification according to selected frequencies. But the actual requirement is that the characteristic must pass the "sweep test" across the full bandwidth specified for the cable category. So performance must be consistent and in accordance with the formula, at any given frequency level, from the lowest to the highest frequency specified.

The major test parameters for communication cables, and the general groupings they fall into, are as follows:
  • Attenuation (signal-loss) related
    • Conductor resistance
    • Mutual capacitance
    • Return loss
    • Impedance
  • Noise-related
    • Resistance unbalance
    • Capacitance unbalance
    • Near-end crosstalk (NEXT)
    • Far-end crosstalk (FEXT)
    • Power-sum NEXT
    • Power-sum FEXT
    • Alien crosstalk (AXT)
  • Other
    • Attenuation-to-crosstalk ratio (ACR)
    • Power-sum ACR
    • Propagation delay
    • Delay skew