Category 5: How Did We Get Here and Where Do We Go Next?
Copper Applications in Electronic & Communications
About the Authors
Jim Serenbetz is Vice President-Advanced Technology and Pete Lockhart is Senior Marketing Manager-Structured Cabling Systems for Anixter Inc., a global, value-added provider of integrated communication systems and services.
Is the reliable performance of your network infrastructure important to your organization's bottom line? According to a study commissioned by LeCroy, a high-end test and measurements equipment manufacturer, failures at the physical layer (structured cabling) account for an average loss of $250,000 per year per 100 users. Losses are measured in user productivity, network manager effort and business downtime. Couple this with the fact that the physical layer represents only about 10 percent of the overall network installation costs, when including the computers, software, structured cabling and support costs, and you can see a big reason to be concerned. Fortunately for the people responsible for cable infrastructure, a system of acceptable standards exists that defines the expectations and limitations of cable, and provides structure and direction for technological advances.
Looking Back: I Thought Wire Was Wire!
In 1989, Anixter Inc. began a dialog with customers about what seemed to be indiscernible differences in communications cabling construction. After all, isn't all wire created equal?
At about the same time, the computer electronics industry was all a buzz about a tiny startup company in Texas that promised to obsolete the coaxial cable and connectors we had all learned to love and hate. Sure, we got great performance from our Ethernet transceivers and cabling (affectionately called Thicknet and Thinnet)... when it worked, but when it didn't we had a hard time locating the problem. Today, words like "reflection", "terminator" and "vampire tap" have quickly faded from memory in the wake of robust and reliable 10 and 100BASE-T LANs and "categorized" unshielded twisted-pair cabling systems.
The Anixter engineers, concerned that vendor hype would distort the facts, developed and published a "Cable Performance Levels" purchasing specification for communications cables that emphatically stated that all cables are NOT created equal. The specification was their attempt to create some measurement of electrical uniformity and performance assurance in the cable manufacturing process. Three years later EIA/TIA published the cabling standard that set the baseline for interoperability in structured cabling and provided a consistent platform for networking devices to be built to.
The Original "Levels"
Someone entering the industry today wouldn't know (or care) about Level 1 telephone voice-grade copper cable - or "POTS" (Plain Old Telephone System) cable, as it was called. Level 2 handled IBM mainframe and minicomputer terminal transmission, as well as some early slow-speed (1-2 Mbps) LAN technologies like Arcnet. Level 3 was designated as the minimum quality twisted-pair cable that would handle 10 Mbps Ethernet and 4/16 Mbps active Token Ring without errors at the desktop.
The seven years since these original Levels were defined have seen America being rewired, information transmission technologies advancing and standards ratified. In 1992, a group of manufacturers marketed a copper version (CDDI) of an FDDI (Fiber Distributed Data Interface) transport system using thin coax and IBM type 1 cabling products. And in that same year, Anixter authored a Level 5 purchasing guide for 100 Mbps over UTP (unshielded twisted-pair) cable and delivered it to cabling manufacturers for specification compliance.
In 1993, ANSI ratified TP-PMD (twisted pair-physical media dependent) for FDDI over Category 5 UTP. Shortly after that, ELVRIA signed the "568" standard document, followed immediately by TSB-36, which adopted in total the "Levels" requirements set forth earlier by Anixter - although ELVFIA defined these levels as "categories." The Anixter engineers then worked with Underwriters Laboratories, Inc. to create the UL LEVEL testing and follow-up program, which assured end-users that the manufacture of cables would fully meet these levels/categories programs, and would give them an independent yardstick for cable performance.
It wasn't until the birth and availability of affordable 100BASE-T in 1996 that institutions and organizations saw a reason to enable 100 Mbps desktops, and then largely because it was an inexpensive and well-understood insurance policy. For a little extra money, whether turned on or left dormant, dual 10/100 Mbps Ethernet network interface cards became a "no-brainer" for network managers. Fiber optics and FDDI remained in the campus backbone, and became the server superhighways and intercloset infrastructures. In five short years (since Level 5 was introduced), the physical layer transport of the most future-thinking planners has become "maxed out" in terms of the high-speed networking options of the near future!
Isn't Cat 5 Enough?
"I might give my users 100 Mbps at their desktops since it's inexpensive and my support groups know Ethernet very well... and I've heard that intranets could very well deliver some high-bandwidth 'enabled' applications soon."
This is not an uncommon statement these days. Few of us can conceive of the need for anything beyond 100 Mbps. For example, 155 Mbps ATM (Asynchronous Transfer Mode) is seen by many to be a technology in search of an application, but we should remember how much things can change in a five-year period. In 1991, an article in PC Week magazine reported an analyst's opinion on the newly introduced 486 processor: "Outside the niche of graphics and CAD applications, there's no need to have a 486 sitting on your desk ... " Few mainstream applications existed to take advantage of the increased processing power. The article further observed: "Overall, the greatest impact of high-end 486s will probably be on applications that have not been computerized yet." At that time, Windows 3.0 was the greatest boon to the IBM-based PC since the Macintosh was introduced. It wasn't until later that WIN95 (code named "Chicago") made a ripple in the back pages of PC Magazine, and the Internet was just a scholarly and governmental vehicle to pass data back and forth.
Today in the age of Pentiums, the chicken-and-egg routine continues. Processing power ultimately drives innovation in user applications, specifically media-rich and collaborative functions, and these business and learning-enabled applications ultimately drive the need for more bandwidth when it's needed. Transport technologies like switching and ATM will likely catch on as the economic and sociological benefits of multimedia, distance learning and media conferencing are realized. When thinking about where applications will be in five years, think about the size of hard drives and modem speeds in 1991. Hundredfold increases in performance and plummeting costs make these technology innovations solid drivers in today's corporate America, as well as in education and healthcare.
If we consider that five years ago our high-end wiring choice was Category 5 cabling in the LAN and multimode fiber optic systems in the backbone, when LAN speeds were 10, 16 and 100 Mbps, the "headroom" or additional capacity we built into our systems seemed more than adequate for the future.
This past year the ATM Forum put its seal of approval on 155 Mbps ATM to run on existing Category 5 systems, and the first interface products have just recently started to appear on the market. We may ask what applications will require more than 100 or 155 Mbps at the desktop; but the more visionary question is: Will my Category "X" cabling system have enough additional "headroom" or TRUE electrical bandwidth to provide error-free transmission when I do need the extra throughput?
A few issues need to be explored to answer this question satisfactorily. All high-speed LAN standards require compliance with generic cabling specifications plus many additional parameters that are defined only in the specifications and standards for the network interface products. These extra requirements define the actual electrical and digital signaling, and usually assume a well-behaved and consistent cable and connectivity system. The figure below shows the relationship between cabling standards (center ellipse) and networking standards (outer ellipse), and demonstrates, unfortunately, that cabling requirements are just a subset of the overall requirements for a smooth-running network.
All high-speed standards need to conform to SNR (Signal-to-Noise Ratios) and maximum noise thresholds. But pair skew and propagation delay characteristics are important supplemental requirements for 100BASE-T, 100BASE-VG and for ATM above 100 MHz. Pair skew applies to technologies using multiple pairs for signaling. In essence, signals are divided between pairs and must be reassembled at the receiving end. If they arrive at different times, skewing of the signal occurs, resulting in transmission errors. Propagation delay, the time it takes for the signal to travel to the receiver, is a factor of the efficiency of the cable in moving the signal relative to the theoretical speed of electricity (light). Also known as the velocity of propagation, it is expressed as the percent of the speed of light represented by the cable's speed.
Network electronics manufacturers deal with electrical loss across cable distances by incorporating equalizers into their receivers. These equalizers attempt to amplify the received signal based on what they assume happened through attenuation or the electrical loss during transmission through the channel. This same received signal must also be identified with the noise picked up during its transmission and receipt, and in most cases a little bit of the noise is also reamplified. If this results in an incorrect representation of the original signal it is called a "bit error." Bit errors often lead to garbled information and/or retransmissions of the data.
As in the case of 155 Mbps ATM running on Cat 5 cable, anomalies can occur above the Cat 5 maximum signal frequency (in excess of 100 MHz and as far out as 200 MHz) that when seen by the equalizer are amplified as if they were part of the signal. This results in higher than acceptable bit errors and therefore corruption of the information. No additional headroom will help in this case. If the attenuation performance of the cable is not smooth, then the ATM signal will probably not be interpreted correctly even though the cable installation passes Cat 5 requirements below 100 MHz!
Aren't All Cat 5 Cables Created Equal? Isn't This A Standard?
Let us preface this section by saying that as active participants in the standards organizations, we at Anixter are firm believers in the need for standards and think the public needs to know about the standards process. Standards by definition are derived by consensus and often are open to interpretation. "Delay Skew" is an addendum to the ANSI/TIA/EIA-568-A specification that requires another test be performed on the cable before it leaves the manufacturer. The TIA task group has rejected suggested names for the addendum (Category 5.1 or Cat 5-1997), and has elected not to have the cables that would comply with the new standard marked differently from the other seven billion feet of four-pair cable already manufactured and currently installed in North America. The only way to know for sure if your cable meets this new requirement will be to get a copy of the actual product specification the manufacturer used to make the exact cable you purchased at that time. When was the last time you consulted the cable manufacturer's spec sheet? So, enhancements to cable can only be determined by looking at exactly what parameters the manufacturer has tested and guaranteed.
Performance is directly related to the chemical compounds used in the manufacture of cable. To date and largely because of a worldwide shortage of FEP (Fluorinated Ethylene-Propylene-Teflon, a registered trademark of DuPont, and Neoflon, a registered trademark of Daikin) from early 1994 on, there are more than 105 different electrical designs of plenum cables, including 15 high-end Cat 5 plenum designs and 33 standard and high-end nonplenum designs - all with varying electrical performance characteristics, yet still Cat 5-compliant.
In addition, a high-speed system must display Category 5 characteristics from input to output - in other words, across all connectors, cross-connects, patch panels and outlets. So, assuming our Cat 5 cable tests out at 155 MHz, we still must contend with the quality of the components and the installation. Some of the various plenum "flavors" that used different numbers of Polyolefin pairs mixed in with the FEP pairs to reduce the amount of FEP consumed, were very installer-friendly; others were not. This mixing of different materials can cause the propagation delay skew to exceed the 45 ns specified in the revised TIA-568 standard and has resulted in a recent addendum.
It's ironic that the original EIA/TIA-568 was signed in the summer of 1991 and only covered what essentially was 10BASE-T electricals, or the then-current Category 3. Immediately after the standard was issued, the committee came out with TSB-36 (Technical Systems Bulletin) for "Additional Cable Specifications for TwistedPair Cables," which defined the new Category 3, 4 and 5 electrical performance requirements based on the Levels Program developed by the Anixter engineering staff and the work done at NEMA (National Electrical Manufacturers Association) and ISO.
A TSB is not a standard but a preliminary look at what a standard might be as generated by the TIA working group. That is, if they publish such a standard, it might look like the TSB after the voting is done. So, a new standard can be approved then immediately made obsolete by the same working group. A rewrite of the 568 standard was signed into existence in October of 1995 as ANSI/TlA/EIA-568-A, and ANSI formed a working group to explore the issue of delay skew, resulting in another change or addendum. Many standards are obsolete the day they are signed because they cover existing, implemented and proven technologies that by design must be available from a number of different sources.
Frost & Sullivan, an information company specializing in high-technology market research, addresses the effect of standards on cable manufacturing in its 1997 report on the North American Premises Wiring Transmission Media Market, stating, "Standards have become so prevalent that brand awareness has become less of an issue. ... Because of this, many manufacturers tend to minimally meet specifications, which in turn fosters a market environment where the products become commodities. ... As a result, the advent of standards has impeded manufacturers from developing products that exceed the qualification of standards."
We may see the promise (or opportunity) of deploying even higher-speed technologies in the next five years as applications and processors address new creative and competitive business needs, and continue to consume more and more of the available bandwidth. The Gigabit Ethernet Alliance has concluded that this technology will have a significant impact on cabling. It will push the limits. Regardless of product and installation quality, there will be no slack if implemented on the current Category 5 cabling.
So while it seemed, a mere five years ago, that Cat 5 would be all that would ever be needed in the horizontal cable infrastructure, it appears that headroom and "structural return" concerns will open the books again with the help of the new Anixter Levels '97sm Program. This timely Performance Assurance Program is based on a stringent purchasing specification that requires Anixter suppliers to qualify their high performance unshielded twisted-pair products. (In order for any product to be acceptable to Anixter, it must be tested for compliance to the purchasing specification through a lot sample procedure by the ASCL, Anixter Structured Cabling Laboratory, an independent certified lab connected to the Anixter Technology Center in Mount Prospect, Illinois.)
This new purchasing specification sets guidelines for electrical bandwidth in excess of 100 MHz by reaching for a performance mark that has over twice the actual usable electrical bandwidth of the current Category 5. It also extends the data bandwidth to the 1.2 Gbps performance mark, making it useful in developing Gigabit Ethernet systems while incorporating less sophisticated encoding schemes than those required for conventional Cat 5 cabling.
The original Level 5 specification from 1992 was modified to cover the performance requirements for existing Cat 5 cables. The more stringent requirements for what has been called High-End Cat 5 or Cat 5+ cables are referred to as Level 6 in the updated levels program. And a new generation of recently launched products that meet the twice Cat 5 bandwidth requirement constitute Level 7. Note that Level 5 is different from the standard Cat 5 in that it now must meet the more stringent requirements included in the international standard document ISO 11801. This standard allows cables meeting these requirements to be used globally. This new definition for cable performance creates a "super-set" of the original Category 5 requirements.
At a recent meeting, the working group of a major standards organization elected to use the Anixter Levels '97 specifications as a guide for its new draft of the UTP Technical Requirement Standard.
Copper cabling technology has certainly come a long way in less than a decade!
As is usually the case, the implementation of a cabling infrastructure should fit the need. Corporations, financial institutions, healthcare providers, and colleges and universities are poised in many ways to take full advantage of the technology wave to enhance their competitive advantage. Is your organization ready to deploy tomorrow's technological advances? Is your physical layer infrastructure up to the task? Thankfully, there are cost-effective, future-proofing solutions in high-end copper still in the works. And as we move into the world of lightwave communications over optical fiber, the same guiding principles of price, performance and ease of installation and maintenance are at work in the engineering and standards committees of the industry.
© 1997 Anixter Inc.
Reprinted with permission of Anixter Inc.
Also in this Issue:
- Home Automation Comes of Age: The Future Has Finally Arrived!
- Category 5: How Did We Get Here and Where Do We Go Next?
- xDSL Glossary
- xDSL Technology and The Internet (Part I)
- Communications Wiring For Today's Homes
- Best of Both Worlds: Helping to Create a Good Environment to do Business