How To: Getting Started With Ethernet Cabling

While many sites are installing wireless networks, wireless is not always the best choice. Security needs, building construction, and other reasons can lead a company to link its computers using old-fashioned wired Ethernet.

To tap into the Ethernet cabling market, a system builder needs to know three basic things: which special tools are needed, how to choose the right cable, and how to terminate the ends of a cable run for a reliable connection. Let's look at each of these three need-to-know factors.

Choosing the Right Tools

As with any job, having the proper tools for installing Ethernet cabling will make the work go faster and easier. Plus, you'll have a better chance of doing it right the first time.

To get started, you'll need to acquire some specialized network tools, which you can buy at either electrical-supply houses or retailers like Home Depot, or Lowe's:

• Fish tape: A fish tape (or "fish") is a long, semi-rigid piece of flat spring steel used to pull cable. Basically, you unwind it, poke it through a narrow space, and then fasten it to the cable you want to pull.
• A length of 3/4-inch diameter PVC pipe: This is useful for poking a cable through any area you can't get the fish into. Drill a 3/8-inch hole through the wall of the pipe near one end; that's where you'll secure the cable.
• Electrical tape: You will use this for securing the cable to the fish tape or pole while you're pulling it.
• Crimper for RJ45 connectors: Useful for making patch cords.
• Punch-down tool with correct blades: Needed for the RJ45 jacks.
• At least one sturdy ladder.
• Drywall saw: Handy for making holes for the network cables and jacks.
• Flashlights: Headlamps or small LED task lights that can clip to the brim of a cap let you see into dark corners while still keeping both hands free.
• Cable tester: Use this to verify that cable runs and patch cords are properly terminated.
• Label maker: While a Sharpie or other permanent marker will work in a pinch, the results aren't professional looking. Use this instead.

You'll need more than just bulk Ethernet cable to properly install a network infrastructure. Among the consumable items you'll need are:

• The proper type of Ethernet cabling (discussed in detail below).
• RJ45 connectors, if you'll be making patch cables.
• RJ45 jacks, wall mounting plates, and connection boxes.
• Surface-mount boxes for mounting jacks, if you're working in an area where the jacks cannot be mounted flush with the wall surface (for example, on a cinder block wall).
• Patch panel(s) to terminate the cable runs in the wiring closet.
• Cable ties and Velcro wraps for securing bundles of cables.
• D-rings for securing bunches of cables to a wall.
• Conduit for protecting cables in exposed areas.
• Labels for the label maker.

Choosing the Right Cable

Cabling is cabling, right? Wrong. Cables may have solid core or stranded core conductors, may or may not have shielding, and may or may not be rated for use in a plenum. You need to select the proper type of cabling depending upon the application. Among the factors you need to consider are whether the cables will go through a plenum (a space used in building HVAC systems that acts as an air return), the speed of the network, whether you'll be making patch cords, and the proximity of your cabling to sources of electromagnetic interference.

Many cable runs will need to be routed through a plenum or to penetrate floors or building firewalls. For this reason, I recommend that you buy only plenum-grade cable. Among other things, it emits fewer toxic fumes if burnt in a fire, and is therefore required by many fire codes for commercial buildings. By buying only plenum-grade cable, you'll have one less thing to keep track of—a big help on complex jobs.

Most cabling jobs you'll encounter will be for 100-Mbps Fast Ethernet, which requires the use of Category 5 or 5e ("CAT5" or "CAT5e") twisted-pair cable. The specific type of cable is generally printed on the cable's jacket. Virtually all of the bulk cable you'll run across will be CAT5e, which will serve the majority of customers well for a long time.

The differences between the types of cable impact how much data can be transmitted on them in a given time. CAT5 is suitable for up to 100 Mbps networks in runs of up to 100 meters long. CAT5e cabling can be used for Gigabit Ethernet patch cords, or for 100 Mbps runs that exceed 100 meters in length. And CAT6 is recommended when the entire network is being designed for Gigabit Ethernet. Also, if the cost isn't prohibitive, you can use CAT6 cabling to extend the long-term utility of the cable plant; for example, to future-proof a 100 Mbps installation that might be upgraded to Gig-E in the future.

You may encounter CAT3 cabling, but it's suitable only for either 10 Mbps Ethernet or, in its most common use, telephone wiring. Increasingly, however, even in-building telephone wiring is being run with CAT5e cabling, mainly to help future-proof the installation.

For permanent cable runs, such as those between the patch panel and a jack, you should use solid (as opposed to stranded) conductor cabling. While solid cabling is less flexible than its stranded counterpart, solid cabling conducts electrical signals better. On the other hand, for non-permanent cable—such as patch cords connecting an Ethernet card to a network port, or a patch panel port to a switch—use the more flexible stranded conductor cabling. Patch cords made with solid conductor cable can eventually fail, as the conductors break due to flexing.

Blocking Interference

The vast majority of Ethernets can be run using unshielded twisted pair, or UTP, cabling. However, some environments expose the network to significant electromagnetic interference. These situations call for the use of shielded twisted pair (STP) cable. But be sure to use shielded terminators on the cable ends (such as jacks or RJ45 connectors) of an STP run. Otherwise, you will cause the cable to act like a big antenna, making the electromagnetic interference worse, not better.

If you make patch cords, be sure that the RJ45 connectors or plugs are matched to the type of cabling you're using. There is a difference between RJ45 plugs meant for use with solid and stranded conductor cabling, and between UTP and STP. There are four kinds of plugs. Most patch cables will require plugs intended for unshielded, stranded conductor cable. But if for some reason you do a long run of UTP with solid conductor cable that isn't punched down into a jack, use plugs meant for use with solid conductor UTP.

In general, I recommend RJ45 connectors that include a separate little plastic guide through which you thread the individual conductors before inserting the end into the plug. This guide helps ensure that the conductors remain in the proper order. That way, when you crimp down the ends, the pinouts are correct. Before I discovered this type of RJ45 plug, I loathed making patch cords.

The first step in any cabling job is to organize and plan ahead. This helps you plan the work in an organized manner and purchase the correct quantity of materials. I like to assemble a collection of documents that note the following:

• Floor plan, with each cable drop marked.
• Details of interior wall construction. For example, will you need to drill through a cinder block wall?
• The number and type of ports needed at each workstation.
• Number and types of cable supports—such as conduit, trays, hooks, and D-rings—needed to keep the cable neat.
• Number and types of jacks—surface mount or wall mount.
• Location, type, and number of patch panels.
• Number, type, and length of each cable run.
• Number, type, and length of any patch cords.

Once you know where the cables go and roughly how much material you need, you're ready to start running cable.

In some cases, you can pull more than one cable simultaneously. For example, if you've got multiple drops running from the central wiring closet to a room or down a hall, you can pull them all at the same time. But you'll need multiple spools of bulk cable to do this.

You'll often use a fish tape to pull wire through confined spaces. First poke the fish through from the side you're pulling wire to. Then attach the wire to the end of the fish by threading it through the loop on the end of the fish, and secure it with several wraps of electrical tape so it doesn't snag. How to Terminate The Ends of a Cable Run

After you've run all the cabling, the next step is to punch it down at each end. In the wiring closet, you'll punch down into a patch panel. At the workstation end of the line, you'll punch it down into an Ethernet jack.

Both ends of the cable should be punched down according to the TIA-568B standard. (For a discussion of the merits of selecting TIA-568A or -568B, I recommend this Wikipedia entry.) This will result in a straight-though wiring scheme, which is the standard for connecting a PC or server to a switch, hub, or router. Most jacks and punch-down panels are color-coded; so to make sure they're punched down correctly, simply match the colors of the conductors to those on the jack.

Here's the correct order for a TIA-568B connector, moving from left to right: white and orange, orange, white and green, blue, white and blue, green, white and brown, and brown. The proper order for TIA-568B is shown in the following photograph. This picture also shows the small plastic insert I mentioned earlier. This would be inserted into an RJ-45 connector with the latter's tab facing away from you.

If you need to use the TIA-568A wiring scheme—for example, to make one end of a crossover cable—then the conductor order would be: white and green, green, white and orange, blue, white and blue, orange, white and brown, and brown. It should look like this:

It is very important to maintain the twists in the cable as far as possible up to where the cable is terminated, whether in a jack or in an RJ45 connector. The twists are there to reduce electrical crosstalk between conductors. If you untwist the conductors too far from the cable's end, you'll allow crosstalk, and that can cause data-transmission errors. Here's the end of a bad cable:

Instead, maintain the twists all the way to the end. You need to do this for both patch cords and terminating runs in Ethernet jacks and patch panels. Here is how you want it to look:

Once you've finished running and terminating each cable, next you need to test and label them. Label each run at the patch panel and on the jack. If you do a lot of cabling, then I recommend investing in a Fluke MicroScanner Pro Network Cable Tester, which retails for around $400. But if you're doing only an occasional cabling job, testers retailing for about $100 will work just fine.

Installing Ethernet cabling is not rocket science, but to do a good job, you will need to pay attention to the small details. Be sure to use the correct types of cabling and connectors. And always test all cable runs you make. By doing so, any system builder can gain new business from their SOHO clients.

DAVE MARKOWITZ provides PC, Mac, Linux, and networking support to small businesses and home users in the Philadelphia area.