Published In April 2001
Last month we looked at the state of Gigabit Ethernet today and recognized that it will be confined to backbone networks and server connections for the reasonably foreseeable future, simply because most desktops can’t use such speed. We also noted that data cables carrying Gigabit transmissions must be handled much more carefully than those carrying lower speeds. This month, we will look at the tests required to certify cable for reliable Gigabit transmissions. This will help us understand the Gigabit testers we’ll survey later. The general measurement categories that are specified to ensure that a cable can function as required include the following: loss, pair coupling, delay, and signal-to-noise ratio. Each of these in turn can be measured in various ways, and several different variables can be identified in each category. Loss The simplest concept for many of us is that energy travelling through a data cable meets resistance and is reduced by that resistance. This loss, called attenuation, is measured on a logarithmic scale in decibels (dB) for voltage or in dB (W) for power. Because of the log function, the decibels can be added for each part of its journey. Thus, the attenuation of the horizontal wire cable, the cross connect, and the patch cord can be summed to find how much energy is lost on the entire trip. Increasing frequency and cable length increase the attenuation suffered by signals making the trip, which explains why it becomes more critical as a cable plant moves up from Ethernet to Fast Ethernet to Gigabit Ethernet. Pair coupling A number of measurements target the phenomenon of signal energy in one pair affecting the signal in another pair (or pairs). Instead of staying in the pair where it originates (and is desired to stay), some of it becomes electromagnetic energy (radio waves) and travels to any other close-by signal-carrying pair, thus corrupting that independent signal. Coupling of energy in one pair into another pair is called crosstalk. It is measured at both ends of the cable, and these measurements are called near-end crosstalk (NEXT) or far-end crosstalk, respectively, depending upon whether the measurement is done at the transmitting end or the receiving end. In 1000Base-T (Gigabit), four separate pairs are used and the new term powersum (PS) is used to describe the combined crosstalk effects of any three pairs on the fourth. So, we can now measure PSNEXT and PSFEXT. To capture the extra sensitivity of Gigabit signals to cable quality, the idea is introduced of measuring crosstalk with respect to the received signal, rather than to the transmitted signal, and this is referred to as equal level (EL). Thus, 1000Base-T re-quires measurement of ELFEXT and PSELFEXT, as well as the previous variables. Finally, any signal travelling through a real cable will encounter barriers such as connectors along its way,which will cause some of the signal energy to be reflected back toward its source. This echo is called return loss, and it is another new measurement required for Gigabit Ethernet. Delay Although light or electricity travels at about 286,000 miles per second in a vacuum, it slows down when it meets resistance in the form of a metal or glass cable, by an amount that depends upon the insulation, among other factors. A recent addition to the TIA/EIA spec to ensure proper Gigabit operation mandates that each pair should carry signals at least 61 percent of the speed of light itself. Thus, Teflon-insulated cable meets this requirement, and such insulation is typically used for 1000Base-T. Signal-to-noise ratio To get a simpler measurement that doesn’t require as much effort as the battery of tests listed above, the attenuation-to-crosstalk ratio (ACR) is often used. This single value for each cable allows an easy comparison between different available products. The fundamental idea remains the same, namely, if the signal is greater than the noise, it can be reliably detected and the data expressed by the signal will be adequately received. If the ACR tests negative, it means that the data and noise cannot be distinguished, and the cable fails the basic requirement for which it was created. Summary of cable measurements Each of the variables just discussed is specified with a threshold value in the TIA568-A standards. In fact, TIA568-A originally addressed only as high as Fast Ethernet (100 MHz) speeds, but the standard itself has grown with several added sections specifying faster operations. Thus, there are increasingly stringent thresholds for Cat 5 used for Fast Ethernet, Cat 5 for Gigabit Ethernet, Cat 5e for better Gigabit Ethernet, and proposed Cat 6 and Cat 7. For instance, the original NEXT at 100 MHz for Cat 5/GigE is 27.1 dB, but Cat 5E/GigE raised it to 30 dB. Next month, I’ll discuss testers for Gigabit Ethernet that implement these standard measurements. SHINN is principal of Compututor. He can be reached via e-mail at firstname.lastname@example.org.