Thomas Russell,
Quabbin Wire & Cable Co.
When the tia/eia-568 standard of the Telecommunications Industry Association (TIA--Arlington, VA) was revised and reissued in October 1995 as tia/eia-568A, the new document incorporated many changes to the original but left intact the basic "category" classifications for rating the performance of unshielded twisted-pair (UTP) copper cabling. With tsb-67 adding the requirements and methods for field-testing installed cable systems to the information included in TIA/EIA-568A, the two documents together provide the "bible" on how to design, install, and test a Category 5 local area network (LAN).
As recently as a year ago, the TIA`s authoring committees were adamant that they would not seriously consider any changes or additions to the wiring document that went beyond the system requirements for Category 5.
Today, this attitude has changed. The technologies governing lans have evolved rapidly, and faster lan protocols are being introduced as issues around grounding and shielding are being discussed. In addition, higher-bandwidth European specifications have emerged. The tia committee now suggests that network managers and other end-users consider the TIA/EIA-568A specification as a minimum requirement and set their own system performance specifications higher if necessary.
In the meantime, work has already begun on addenda to the TIA/EIA-568A standard that will have an impact on this important topic, and they will soon be issued. So, just as the cabling industry has begun to understand the requirements for Category 5 performance, get ready for some brand-new terminology and specifications. Whether you call it Category 5-plus or Category 6, a new performance criterion is on the way.
New and faster lan protocols are the primary reason behind these new requirements. Four-pair utp cabling will still be the medium of choice, but the data transmitted over the medium will go much faster and in different ways than is currently the case. Asynchronous Transfer Mode, for instance, crams approximately 1.5 bits of data into every 1-hertz cycle, so copper wiring operating at 100 megahertz can transmit at about 150 megabits per second. A new version of Ethernet, called 100Base-T4, will use all four pairs of the cable, operating at 30 MHz each, to achieve a data rate of approximately 100 Mbits/sec. In one to two years, it is predicted that Gigabit Ethernet, or 1000Base-T, will be both sending and receiving date on the same wire pair--in so-called full-duplex mode--resulting in a transmission speed of 1000 Mbits/sec over copper.
By contrast, in the earlier Token Ring and Ethernet encoding systems, only two of the four pairs were used, making network testing requirements relatively simple. You could afford to ignore far-end crosstalk; near-end crosstalk could be measured quite simply; and, many other electrical characteristics of the cable could be assumed to be insignificant. At higher transmission frequencies and with the newer lan protocols, however, these assumptions about performance and test methods are no longer valid, calling for tia/eia-568A and tsb-67 to be updated.
When existing data-communications transmission standards were being developed, 16-Mbit/sec Token Ring was the cutting-edge technology, and faster protocols were little more than blue-sky dreams. Over the next few months, though, you will be confronted with new terminology and procedures for network components and their performance testing. The current simple test for near-end crosstalk will be replaced by a power-sum near-end crosstalk measurement, and power-sum far-end crosstalk criteria will be developed. In addition, 100-MHz lan cable tests will be replaced by 350-MHz tests. There will be new test requirements, as well, for return loss, balance, resonance, conversion loss, delay, and delay skew.
Are these test requirements really needed, or are they just clever marketing? The answer to this question is that they are a little of both. The newest lan protocols will require the very highest system performance at higher operating frequencies, and this performance will result, in turn, from superior system design and cutting-edge manufacturing controls in producing system components.
For instance, the electrical characteristic of return loss becomes very important for transmission above 100 MHz. The cable`s impedance varies slightly along its length, creating slight reflections and signal losses which nevertheless add up. Return loss is a measurement of the relative power of all this reflected energy. System designers prefer to use cable with lower return loss and more stable impedance for high-speed networks because there will be less jitter or signal distortion, the attenuation-to-crosstalk ratio will be superior, and analog video signals can be transmitted more clearly.
If, for example, impedance is plotted to 350 MHz for a simple channel consisting of 90 meters of horizontal cable plus about 2 meters of patch cable, the potential problem with cable and connectors meeting the minimum requirements for Category 5 becomes apparent. Even though performance is relatively stable to 100 MHz, impedance becomes very unstable at higher frequencies. Such a cabling setup would probably not support the newest high-frequency lans.
Hardware and cable manufacturers are rushing to develop and introduce a new generation of products that will support these new data-communications protocols. Many Category 5-compliant products that are currently on the market may not provide sufficient performance to support these protocols. In fact, existing products cannot be tested or have their performance levels evaluated above 100 MHz, because at this time there is no test method or pass-fail criterion established for networks operating at such frequencies.
Until specifications and test methods are available, what should a system specifier or lan owner do? One recommendation would be to compare actual test data on existing products. Do not simply rely on a manufacturer`s claims or a company`s glossy brochures. If no testing procedure has been defined, the manufacturer can claim anything. You should not buy cable or other supporting products that are ordinary Category 5 items tested to a higher frequency--the method many manufacturers currently use to support their advertising claims. Rather, look for manufacturers who have isolated the critical production variables needed to make superior next-generation cabling products.
And finally, get ready for 1-Gbit/sec networks over copper. Just a few years ago, such networks were considered to be a technical impossibility, but they will be here very soon.
The impedance of a typical horizontal copper-cable run varies somewhat below a transmission frequency of 100 megahertz, but swings much more widely between the Category 5 maximum of 100 MHz and 350 MHz.
Thomas Russell is vice president of marketing and sales at Quabbin Wire & Cable Co. (Ware, MA), tel: (800) 368-3311, fax: (413) 967-7564, Web site: [email protected].