COBBLED CABLING - Part 3

by Wayne M. Krakau - Chicago Computer Guide, December 1995

This is the third part of my series covering LAN cabling. Forget "Sick Building Syndrome". We’re talking about "Sick Cabling Syndrome".

I’ll start with a clarification of a potentially ambiguous paragraph in Part One of this series mentioning the use of plenum cable. The possibility of misinterpretation was pointed out to me by a reader via E-mail. (Yes, I do read and respond to E-mail!) Many localities require the use of plenum-rated cable if a suspended ceiling is used for air return. If the return air is ducted, you can often get away with non-plenum cable, though there may be rules detailing just how that cable is laid. Other localities such as the City of Chicago require the use of conduit (pipes) for any new cables going into a plenum ceiling. For them, even plenum-rated cable isn’t good enough.

The next incident was discovered during a conversion of an Ethernet system from coaxial cable to UTP (Unshielded Twisted Pair). We had to keep the existing system stable while transitioning to 10Base-T in stages. Luckily, the network was already split into two based on location by having two network cards in the file server. This made our task easier since it automatically isolated half the network.

The network administrator got a head start by switching all workstations to combination network cards that could use either coaxial or UTP cable. This meant we could move them from coax to UTP at will.

Since we needed a stable network prior to starting the switch, I inspected the existing coax. I found numerous open T-connectors, substandard BNC connectors (including the ever popular dangling wires), and a lack of grounding. I also found many kinked segments, including some with multiple kinks.

It took a little exploring to find out why there were so many kinks. I finally realized that whoever originally wired the place thought it was a good idea to stash the cable out of the way for convenience. Since the cable is naturally springy, they had to find a way to keep it from popping out from under desks and the like and getting underfoot. Their solution was simple. They just pinned it down by putting it under desk legs, cabinets, and printer stands! Over time, due to furniture moving, multiple kinks and crushed spots were created. I have no idea how any signal was getting through.

I went back a week later to help with the transition and found that all of the cubicles in the main portion of the office had been removed, leaving an open area. When I asked the administrator if that portion of the cable plant had been rerouted to accommodate the missing section that used to support the cubicles, she told me about her encounter with one of the workers. He had asked her if he could cut the cable in the ceiling. She had, of course, forbidden him, since the proper method would have been to install two BNC connectors at the cut and then connect them with a barrel (inline) connector - while the LAN was down. Later, he told her not to worry because he had "fixed" everything. He cut the cable on an active LAN and had taped the two ends together!

Since this part of the office was scheduled to be converted to UTP in less than two weeks, and, amazingly enough, the remaining workstations on the taped coax still seemed to work, I decided that retreat was the better part of valor and kept my paws off the system. I was afraid that if I touched the cable while it was in such a precarious state, the network would collapse. It had survived other abuse and I wasn’t about to fiddle with it in spite of the fact that I couldn’t figure out how the LAN stayed alive!

The final case I will present is the ultimate in cabling adventures. I was called in to optimize a LAN and to help prepare the administrators for an upcoming upgrade. As a part of my inspection of the existing LAN, I examined the cable plant. First, I checked the coaxial cable that was used to connect the 10Base-T concentrators (hubs). It was 92-Ohm RG62 ARCnet cable instead of RG-58 A/U Thin Ethernet cable. Also, the coax was partially unterminated and was not grounded.

Then I checked the front of the hubs and the patch panel. Almost all of the RJ-45 connectors were incorrectly installed. They had lots of unsheathed individual wire hanging out and didn’t look well crimped. The system also had Category 3 cable going into the wall and Category 3 patch panels, but used a random mix of Categories 4 and 5 jumpers. As I’ve stated in previous columns, you can’t mix cables with different characteristics (which is what "Category" refers to) without causing a point of distortion where the different types meet. Think of it as the difference between clear glass and translucent glass. The light gets through both, but only the clear glass provides a clear, undistorted "signal".

Just to make things really interesting, several jumpers were completely unsheathed and untwisted wires! The specification calls for Unshielded Twisted Pair (UTP) cable, not Unsheathed, Untwisted, and Inherently Unpaired (UUIU?) cable!

As I was about to continue my inspection in other areas, I noticed a metallic glint among the jumper cables. I pushed aside the intervening cables and discovered that one jumper was made up of NAKED COPPER WIRES! It was with great relief that I tracked down the ends of that jumper and found out that even though this cable (maybe "bunch of wires" would be more accurate than "cable") was tangled within the LAN jumpers, it was really a part of the phone system. Whew! It was only a false alarm. I strongly suggested to the LAN administrator that a company with a live LAN but a dead phone system would not necessarily stay in business. He got the hint and promised to notify the phone system administrator about the wacky phone cabling.

Now, onto the workstation side of the system. I checked several offices and found that the LAN sockets were RJ-11 four-wire jacks, not the required RJ-45 eight-wire. Although Ethernet only uses four wires (two pairs), the specs are based on the internal spacing within the RJ-45, not the smaller wire spacing found in RJ-11 or even six wire RJ-12 jacks.

To test some sample cable runs with my LANCat (Datacom Technologies, Inc. 800-486-5557), I had to disassemble the socket and experiment with alligator clips to find which wire was which. Every run that I tested completely failed even at the lowest quality (Category 3) settings. This meant that even though it was Category 3 cable, it was strung in a nonstandard manner and couldn’t carry an Ethernet signal reliably.

Because there were so many standards violations, I called my cabling authority Bruce Kahn of Telnet Communications Consultants, Inc. (Wheeling, Il, 708-215-0003) to discuss the results of my inspection. He confirmed my conclusion that the system was a disaster waiting to happen. It was so destabilized that the slightest change could cause it to collapse.

Because I had only tested a few cable runs, the administrator requested an estimate for Telnet technicians to come in and do a thorough test of the entire cable plant. He felt that my inspection was too limited for him to justify a major recabling job. He also felt that since I was, in essence, out of my major field of expertise (information that I had volunteered previously), that Telnet’s results would carry more authority.

I discussed this client with Bruce a few weeks later and found that upper management had vetoed the request for a thorough cable plant test. The system wasn’t broken (at least from their point of view), so why fix it? In retrospect, I actually agree with their decision to skip further testing, but not for the reason they would have expected. I am convinced that the system is so totally unsalvageable that any testing done in hope of saving portions of the existing cable plant would be a total waste of time and money! I think that they should just tear out the whole thing and start over from scratch - preferably before it collapses and heads start rolling.

The overall theme of this series harks back to one that I’ve repeatedly used in when covering other aspects of systems integration. Don’t put up with people dabbling outside their field - and for goodness sake, don’t dabble yourself! For LAN cabling, deal with specialists with heavy experience in high-speed data cabling for LANs. This excludes computer service companies, programmers, dealers, resellers, VARs, systems integrators (including my company), other assorted amateurs, and even telephone cabling companies. The job you save may be your own.

1995, Wayne M. Krakau