There are a variety of important factors that today's installers must be aware of in order to maximize channel performance.
January 1, 2001
There are many factors that affect the electrical performance of a channel today. The evolution of UTP networks from 10BaseT to Gigabit has brought about a host of new electrical requirements such as return Loss (RL), powersum crosstalk (PSNEXT) and powersum equal level far-end crosstalk (PSELFEXT). Yet, a number of installation practices, commonplace to the industry, appear to have a negative impact upon these new electrical requirements.
By looking at the bigger picture, we can get a better understanding of the pitfalls of installation practices. This article will follow a typical installation path, from the initial specification to the termination of the cable, and will examine the impact of cable management and common installation practices upon new Category 5e and draft Category 6 channels.
MANAGING THE BULK CABLE
Installers want to minimize their efforts while maximizing their profits. One perceived method of doing this is to have horizontal cables pre-assembled into some type of bundle. This minimizes the number of pulls needed to complete a job. The bundled cables appear to be less susceptible to damage from the pulling process, due to the bundle’s physical size, and there are fewer reels to track and dispose.
Figure 1 illustrates a cable that was assembled for demonstrational purposes. Six 100m cables were twisted together and bound with two binders wrapping in opposite directions. The bundled cables look undamaged and have almost no jacket deformations.
The performance of the individual cables was evaluated prior to the assembly process. Notice that the crosstalk performance in Figure 2 is almost 9 dB below the Category 5 specification, 6 dB from Category 5e, giving very good initial performance. However, after the assembly process, the pair combination indicated by the red trace in Figure 3 is now very close to the Category 5 specification.
Note that the performance now fails Category 5e requirements. The money spent on a high-performance cable and the bundling process has been wasted. The cable geometry that is designed by the manufacturer requires many hours of research and development. Frequently, these designs are dependent upon a very stable manufacturing process to maintain consistent performance. Whenever a cable is bundled in the above fashion, a twist is imparted upon the pairs, disturbing the original design.
There are several things to be aware of when a bundled cable is specified. These cables are usually contracted out to facilities that merely pull the specified manufacturer’s cable into some type of bundling machine. Many of these facilities have been bundling heavier gage appliance, fixture or building type wires for a number of years and are unaware of the impact of bundling data cables. Frequently, the cables are pulled in excess of their rated pulling tensions, stretching the copper conductors. This may cause attenuation failures in the lower frequency range from 1-20 MHz. Sometimes these cables are numbered or reprinted in some fashion. The process of printing a cable, if done improperly, can squeeze the cable into an oval shape, disturbing the core’s geometry and causing problems with RL and crosstalk performance.
The installer must be aware of the rules regulating these bundled cables. When a bundled cable is ordered, often the only requirement on the purchase order is the number of cables in the bundle. When the installer receives bundled material with damaged cables, the problem is frequently not diagnosed until the material is already installed. New single cables may be pulled to substitute for damaged cables within the bundle. If there is no further room for additional cables, the entire bundle may have to be cut and removed, with a substantial increase to the installation cost and time. At this point, the process of establishing responsibility may be very tricky. Most cable manufacturers will not warrant their cable if bundling was performed by an unqualified third party. If the electrical specifications of the finished product are not specified on the purchase order, the assembly house may claim that the product was delivered as ordered (merely bundled cables), leaving the installer with little recourse and an angry end user.
The TIA/EIA has issued an Addendum 3 to the 568A document to address these issues. The document states that the assembled cables must meet the original specifications placed upon individual cables. Additionally, within the bundle there is a cable to cable PSNEXT requirement. This cable-to-cable PSNEXT is sometimes referred to as “alien crosstalk”. In order to ensure that the installer receives material that will conform to the standards requirements, the installer must specify the TIA/EIA 568-A-3 requirements when ordering, and confirm that the company has the ability to verify the assembled products.
ALIEN CROSSTALK VS. VARIOUS BUNDLED CABLES
Alien crosstalk is the noise power that is coupled from adjacent cables. There are many methods to bundle cables, including binders, cinch ties and conduit. The alien crosstalk performance will vary upon the method chosen to rout the cables from the wiring closet to the work area. Figure 4 indicates the alien crosstalk performance of the bindered cable bundle discussed above.
The straight line in the above figure represents the TIA/EIA-568A Addendum 3 requirement for a Category 5 cable bundle. There are 24 traces representing the powersum performance of each pair within the six-cable bundle. The assembly fails the requirement by 4.5 dB at 3 and 25 MHz. This is primarily due to the extended distance (100m) of closely bound cable, allowing a significant amount of alien crosstalk to couple onto the pairs.
Another method of bundling cables is to use some type of tie. Cable wraps come in many forms — from hook and loop straps to plastic cinch ties. In an attempt to simulate this type of cable installation practice, another six cables were bundled with plastic cinch ties every 48 inches. The cable jackets were slightly deformed as indicated by Figure 5.
The alien crosstalk performance with this bundling method improved by over 9 dB. There was now a 4.6 dB of margin to the Addendum 3 specification.
Conduit is also frequently used to route cables. This process of cable management essentially bundles cables into a small area. Another conducted trial consisted of 328-ft. cables pulled into 280 ft. of one-inch EMT conduit. This left 24 feet of exposed cable on each end of the sample under test. This was an attempt to simulate an installation with cable on each end, representing patch cable at the desktop and cabling inside the wiring closet. The first trial measured six cables within the conduit, equating to a 22 per cent conduit fill. In order to represent a more realistic environment, after the first six cables were evaluated, four more cables were pulled into the conduit, resulting in a 40 per cent fill. The results were only slightly worse than with the cinched cables. The alien crosstalk margin to the Addendum 3 specification was 3.5 dB for the 22 per cent fill and 3.0 dB for the 40 per cent fill, respectively.
Non-contiguous cable supports are another method of distributing cables. A prominent cable manufacturer provided a study to the TIA/EIA standards meetings in which the company varied parameters such as the distance between supports and the number of cables bound within J-hooks. The support span was varied from five to 50 feet and the number of cables was varied from one to 32. (Note: The standards violating span is in no way a recommendation or endorsement of these lengths between J-hooks. The extended length was used only for an extreme worst case demonstration and should in no way be interpreted as an endorsement to violate recognized industry standards). The manufacturer’s conclusion was that there was no significant degradation to the RL, internal crosstalk or far end crosstalk (FEXT) from these conditions. Additionally, the alien crosstalk for the suspended cables was within acceptable levels. However, it should be noted that heavily filled support rings co
uld cause long term jacket and electrical problems if the surface area of the support is inadequate.
There are several ways of bundling cable, including binders, cinch ties, J-hooks and conduits. Many of these methods have been examined for their effects upon cable performance. Of the methods mentioned above, the one most likely to cause problems is bundling. Cables may be abused during the assembly process, causing failures within the bundle. Since alien crosstalk cannot be verified in the field at this time, it is important that bundled cable be obtained from a source that can verify the performance of the finished bundle.
Once the cable is pulled near the work area in an open office, a common recommendation is that a service loop should be employed in anticipation of any future furniture moves. A prominent manufacturer of hand-held cable testers that was investigating return loss failures in marginal channels, discovered that the service loops caused a variation in RL performance. Figure 6 comprises the RL variation of a single pair when a service loop 10 feet from the wall outlet is varied from 0 to 30 feet. In this instance, the RL performance may vary from 3-4 dB. Removing or fanning the service loop may help to reduce the severity of the problem.
In the wiring closet, jumper wire is sometimes used to make cross-connections and may be jacketed or unjacketed material. A 100m channel was wired with three different types of jumper wire cables in a cross-connect fashion. The channel was measured from the wiring closet with 3m jacketed cable, 3m of various types of jumper wires, 90 metres of horizontal cable and 3 metres of patch cable at the work area. Figure 7 represents a channel with a remaining sample of the horizontal material used as the jumper wire. Note that the RL performance passes both Category 5e and 6 channel RL requirements.
Figure 8 demonstrates the results of using a “field-made” horizontal cross-connect cable. This field-made jumper wire consisted of removing 10m of jacket from a remaining sample of horizontal material. The unjacketed core was then punched down and the channel re-measured. While barely passing Category 5e specifications, the channel has been compromised to the point of failing Category 6 between 10 and 20 MHz. Additionally, the effects of the jacket removal can be observed up to 160 MHz. This degradation has occurred because the average impedance of the jumper wire has been increased with the removal of the jacket. When the impedance of this 10m segment increases, it causes signal reflections due to the impedance mismatches at each of the 110 type connections.
Finally, in Figure 9, a special unjacketed jumper wire with impedance compensation was used as the cross-connect. This material performs on the same level as jacketed jumper wire.
In drawing conclusions from these experiments, the optimal material to be used in a punchdown cross-connect is the same material that was used in the horizontal pathways. Unfortunately, this is not always possible. Making unjacketed jumper wire in the field will compromise the channel’s RL performance to the point of failure. However, it is possible to maintain the channel’s RL integrity with unjacketed jumper wires designed to operate in free air. However, it should be noted that the crosstalk of Category 6 channels is considerably tighter than the crosstalk of Category 5e channels. This may preclude the use of unjacketed jumper wire due to the exposure of the wires to environmental effects such as manipulation or alien crosstalk in a fully packed cross-connect.
JACKET REMOVAL AND UNTWISTED PAIRS
As the cable is terminated into various types of connectors, the jacket must be removed and the pairs must be untwisted. Excessive removal of the jacket (over 1.5 inches) can cause a 2-3 dB decline in RL margins in a similar fashion to the jumper wire graphs. Screened material will be more at risk due to the nature of the effect the foil tape has on the impedance of the core. Shield integrity must be maintained as far as possible along the core.
It has been documented that the pair untwist will degrade the crosstalk performance. However, as the pair is separated, the impedance of that point is increased and the RL performance degrades. For Category 5e installations, adhering to the recommended 0.5 inch maximum untwist will help assure that the channels crosstalk and that RL performance stays within the specifications. Future Category 6 installations’ crosstalk performance will be more susceptible to termination techniques. The channel RL performance will be assured through 0.5 inches. However, due to the tighter crosstalk requirements, a 3/8″ maximum untwist is recommended.
After the horizontal cables are terminated, patch cables are selected and installed into the channels. In the old days of 10BaseT, patch cables were an afterthought — something that could be made onsite or purchased down the street in an electronics store. The requirements of Category 5e and Category 6 have essentially brought those days to an end. Patch cable return loss has been a very troubling issue within the industry. Return loss problems with these cables have been well-documented in the industry, with many cables experiencing wild swings in performance by merely changing the position of the cord. While the situation has improved to some degree, it not uncommon to see changes in RL performance in the order of 5-9 dB as illustrated below.
Figure 10 illustrates the variation in RL performance of a single pair in a Category 6 patch cable. Patch cables continue to have a high attrition rate and are one of the major reasons behind the inability of hand-held testers to verify a Category 6 channel. As of the fourth quarter of 2000, there was no recognized test set-up to verify the performance of a Category 6 patch cable. Installers and end users should only purchase these products from trusted vendors.
Much has been discussed about the incompatibility issues with Category 6 connectors. The purpose of a standard is to allow for the option of using a “Brand X” jack with a “Brand Y” plug. However, in many early designs of Category 6 connecting hardware, Category 6 performance was only achievable if the plug and jack were from the same manufacturer. The absolute worst case of these proprietary designs involved a particular combination of connecting hardware manufacturers that could not provide continuity from one side to the other.
The industry has made sizable steps in improvement, however at the time this article was written, some connectors were not truly compatible. Others may provide marginal Category 6 channel performance across different vendors, but have better performance when the jack and plug are from the same manufacturer. The recommendation at this time is to use connecting hardware of the same manufacturer when connecting your Category 6 channels.
LOOKING AT DETAILS
To maximize the channel performance, the installer must be aware of several new details. Bundled cables must be properly specified from an assembly house with the capability of verifying the performance. Service loops may be a culprit if RL failures are located toward the work area of the channel. Jumper wire must be properly designed to ensure the RL performance. Additionally, the removal of jacket and pair untwist will become more critical for Category 6 channels.
Finally, the patch cables can make or break the performance of the final channel. Currently, hand-held testers cannot verify a full Category 6 channel with the patch cables attached. With the high failure rate of patch cable assemblies and the possibility of connector incompatibility, it remains critical to obtain Category 6 materials from reputable sources.CS
Paul Vanderlaan is a Senior Product Development Engineer at Belden Electronics Division in Richmond, IN. He has more than eight years of experience with Belden developing high performance premise cables and holds two patents. Paul actively participates in working group
s responsible for the publications of industry standards such as the ANSI/TIA/EIA-568-A. He currently co-chairs the TR42.7.2 copper cabling work group.