High Performance UTP Cabling
18 months ago the TIA looked at expanding the performance standards for cabling beyond category 5. The chairman of the TR41.8.1 Systems Subcommittee went on record as being against expanding the performance, stating that there were no applications that require it. What he meant is that within data communications there were no applications that at that time could use it. He was right for data but not for other areas. For example, the Society Of Motion Picture and Television Engineers had a standard for Serial Digital Video (SEMPTE 259M) that sent up to 360 Mb/s scrambled NRZ data. There are proven baluns (259M was designed for 75ohm coax cable) and UTP wire that can transport that signal 500 ft. and meet FCC requirements (see Belden web site).
The chairman has changed his mind. Why? Progress and marketing. The IEEE is finishing up 1000BaseTX (Gigabit Ethernet over UTP) and the market for Cat5 was languishing. Cable companies have been successfully pushing extended range cable and a major distributor, Anixter, created the levels program. They defined levels 5, 6 & 7 with level five being the equivalent of Cat5 and level seven being a 200MHz performer.
The TIA responded with the enhanced Cat5 and Cat6. Note that within the last 6 months Cat6 was defined by the IEC standards group as a 200MHZ UTP while they also defined a 600MHz STP to satisfy the Germans. So for the last few months the TIA has been grappling with Cat5E (enhanced) and is just now looking at Cat6.
However, in order to place gigabit Ethernet on Cat5 (the IEEE had gone on record that they would shoe horn gigabit Ethernet on Cat5 because of the huge installed base), the IEEE needed some measurements that were not part of Cat5. Because the IEEE used all four pairs they needed far end crosstalk (FEXT) numbers And because, for performance reasons, they send and receive simultaneously, they needed channel based return loss numbers. So the TIA accommodated them by developing additional specifications for Cat5. These are ELFEXT and return loss. The equal level (EL in ELFEXT) was to recognize that the signal is weaker at the receiver and the measurement must be related to this weaker signal.
The TIA is currently exploring specifications for Cat6. There are serious problems in repeatable measurements across labs that are likely to slow down the process of standards development. We are unlikely see cable specifications before next year and connecting hardware no sooner than the year 2000. Today any high performance systems (and many are claiming to have them) are proprietary and not interconnectable.
When categories of cable first appeared there were lots of claims and poor solutions arrived at in haste. Then product design improved rapidly with a few instances of product recall. Cat4 disappeared quickly and today many organizations are installing Cat5 even for telephones. Many early users installed Cat5 with unclassified connecting hardware because specifications were developed a year later than the cable. And all the early users paid the pioneer's penalty, high prices.
The point of the above history is that:
Gigabit Ethernet will work with Cat5 cabling but with small margins for sloppy installations or marginal equipment.
Enhanced Cat5 will supply the desired margins at minor additional cost.
No one can honestly claim Cat6 until the standard is developed and agreed to.
All existing claims-to-be Cat6 solutions are proprietary and unlikely to inter-operate.
What should a user do? The user should install the best cable that he can buy without a large premium. No one knows if the "better" cables have the properties necessary for future applications. The user is trying to buy insurance, not knowing whether it will be useful.
However a useful guide to cable selection is to look at the return loss specifications.The cable should have a low return loss value as this is likely to be a critical specification for complex coded bidirectional signaling.. A low return loss also says that the cable is well constructed (concentricity and uniformity) so the likelihood of meeting a specification not know specified, measured or controlled is enhanced. On the other hand the user should not buy the latest proprietary connecting hardware. Since the cost of retrofit is modest for connecting hardware and today's cost for hardware that exceeds the Cat5E requirements is relatively low, the user can defer getting so called Cat6 until the inter-connectability and price are stabilized.
This last point can be illustrated with patch cords. Users have been buying patch cords made with Cat5 wire but without a TIA patch cord standard (there is none, it is still in committee) ever since the Cat5 standard for wire came out. The price has fallen because the only standards are continuity, pinout and proper color pairing. They can be easily made in a garage. Now users are finding that when they install 100 Base-TX they have to replace these patch cords with higher performance ones (the replaced patch cords worked fine with 10 Base-T). But replacement when needed is easy and cheaper than buying unnecessary high performance patch cords five years ago.
Manufacturers hype is dangerous. For at least two years many cable companies have been bragging about their 350MHz wire. Only some of that wire will meet enhanced Cat5 specifications even though those specifications are only to 100MHz. Even the best wire out there today that passes Anixter's level 7 (only a few brands now qualify) is only 200MHz wire. Yes, these companies measured several parameters to 350MHz, but what did it mean?
The wire needs to meet a family of specifications dictated by the applications. This family of specifications needs to be hammered out by all the companies in an open forum and even then there is no assurance that what is hammered out in advance of the application will be sufficient. Just look to the many revisions to Cat5 over the last five years.
The siren song of some of the biggest companies out there is beguiling. But, you don't need the yet-to-be-defined but somehow-they-offer Cat6 performance. You do need to search for the the well-made, low-return-loss, economical wire that is out there. And find jack and patch panels with good FEXT (at least 3 dB above the Cat5E specification supplying double the margin). Good FEXT indicates that the jack compensation design has an inherently low inductive element, a critical factor at higher frequencies.
© Energy Transformation Systems, Inc.