Specify cord testing to assure LAN performance
Specific contract wording, beyond referencing standards, can help ensure high-performance systems.
When the permanent portion (permanent link) of a LAN cabling system is installed and tested, it does not include any patch cords. The cords are usually added later, creating the channel. Unfortunately, many contractors and end users have discovered permanent links that formerly passed with comfortable margins have lost margin, and now fail the channel test when the cords are added.
How can a few small cords degrade a much larger network?
Patch cords, by virtue of their construction and placement near the ends of a typical LAN cabling channel, can have a big influence on a channel's performance. Consisting of a short length of cable with a modular plug (connection) at each end, non-compliant cords create relatively large near end crosstalk (NEXT) and return loss noise levels. To make matters worse, since cords are usually located near the active network gear, these noise levels are not subject to much attenuation.
A small impedance difference between the patch and horizontal cable, coupled with signal reflections due to bad connections, results in a huge noise source at the network interface card (NIC) or switch. Therefore, patch cords are often the weakest link in a high-performance Category 5e or 6 channel.
Distributors, engineering firms, and end users should expect the patch cord performance specified, and one way to get it is to demand actual test data.
In mid-2002, the Telecommunications Industry Association (TIA—www.tiaonline.org) issued the first true component-level patch cord test procedure, TIA/EIA-568B.2-1, Annex J. It is an effective test, but requires relatively expensive test equipment and is not typically suitable for field use or in a cord-manufacturing environment. Thus, most open-market patch cords (cords not sold as part of a specific manufacturer's LAN system) sold today are not fully tested to this specification. Most are assembled using low-cost labor and questionable components, and are tested only for continuity and wire map.
Fluke Networks (www.flukenetworks.com) recently introduced a patch-cord test adapter for use with its DSP-4000 series cable analyzer. The introduction of this test tool permitted cord testing to the full requirements of TIA/EIA-568B, including measurement of propagation delay, length, delay skew, NEXT, wire map, and return loss. Thus, the user could quickly certify Category 5e and 6 cord performance. Realizing this tool would have an impact on the cord marketplace, my company obtained several units, evaluated their performance, and began testing open-market patch cords.
We obtained these open-market cords via the same sales channels through which a contractor or end-user would obtain them—from distributors, retail outlets, cord-assembly houses, and catalog houses. A total of 34 sources widely distributed across North America supplied both Category 5e and 6 cords. We obtained 2- and 3-meter cords, with three or four samples of each category.
The samples totaled 96 Category 5e cords and 53 Category 6 cords from 20 distributors, four retail outlets, seven assembly houses, and three catalog houses. (Category 5e—63 from distributors, 13 from retail outlets, 10 from assembly houses, and 10 from catalog houses; Category 6—20 from distributors, nine from retail outlets, seven from assembly houses, and three from catalog houses.)
Each cord was identified as to vendor, bulk cable manufacturer, sales channel, and performance category. We also recorded the purchase price to correlate performance to cost. Statistical analysis, based on the assumption that there are an infinite number of cords available, shows that test results from this sampling should be accurate within 10%, with a 90% confidence factor.
Initially, we debated whether to test Category 5e cords. Some believed that because most of today's systems are Category 5e, and the requirements for 5e performance are relatively easy compared to those of Category 6, most or all Category 5e patch cords would pass. So, testing all the cords could be a waste of resources. Unfortunately, the test results showed otherwise.
Category 5e cord-testing results revealed a surprising 69.8% failure rate. Because Category 6 cord requirements are new and much more stringent, we anticipated higher failure rates. Those pessimistic expectations were met. The data showed 83% of the Category 6 cords tested did not meet the TIA requirements.
Category 6 failures were predominantly huge NEXT issues, indicating bad terminations; however, some failed both NEXT and return loss, adding the issue of questionable cable quality. No Category 6 cords failed return-loss alone. Failing Category 5e cords had smaller margins of failure, with NEXT and return loss problems more evenly distributed.
Many failing cords had damaged or deformed cable, inconsistent assembly techniques, and too-tightly coiled packaging. Some of the failed cords were assembled using cable and/or connectors from vendors that are well recognized and generally considered to be good quality (including our own). Overall, it was apparent that most cord assemblers did not have the proper manufacturing processes or testing capability to consistently produce compliant Category 5e or Category 6 cords.
There was no correlation between price and performance. The failure rates were roughly equivalent across all purchase channels; however, one Category 5e assembler had 100% passing samples.
We noted that this vendor uses high-quality bulk cable and plugs combined with good handling, assembly, and packaging techniques. Another assembler uses similar techniques to produce Category 6 cords. Although they did not test their cords at the time that they provided samples, they still only had a 25% failure rate.
You can conclude that it is possible to produce high-volume, fully compliant Category 5e and Category 6 patch cords if you combine the proper cable, plugs, assembly methods, and test gear. In fact, some Category 6 cords manufactured today in controlled environments achieve 95+% yields. Compliant cords can be produced.
This test program's dismal results provide a wake-up call to the industry. Remember that cord performance affects channel performance. Cords can be manufactured to meet published test requirements, and a simple, accurate cord-test capability is now available.
How to get what you expect
Distributors, engineering firms, and end users should expect patch cord performance specified, and one way to get it is to demand actual test data. Specifically, we strongly recommend you consider adding to your specification and bidding documents a requirement similar to the following:
100% of the patch cords furnished for this project shall be tested and shown to comply with the applicable Category cord requirements of TIA/EIA-568B, including wire map, NEXT, and return loss. Compliance must be proven by testing cords alone, not by inserting cords into a channel. Cord performance may be measured on-site using either the TIA's method detailed in Annex J or using a cord-test adapter to a handheld LAN cable tester. Alternatively, cord compliance may be proven by actual test reports supplied by the cord manufacturer.
If this statement becomes standard language in project documents, you will quickly make the patch cord market an honest one.
Tom Russell is vice president of marketing at Quabbin Wire (www.quabbin.com).