Installing a Home Network: Part 2

In my previous article, I took you through the actual provisioning of cable Internet service into my house. With the service in place, my next task was to run CAT5 cable to all our computers. The cable entered the house on the east side of my basement. The family computer, used by my wife and kids, is located in the dining room. Since that is a finished room, I'd have to fish a cable up through the wall somehow, and install an outlet. My office contains the other three computers I was concerned with, and it is located in a large, finished room on the west side of my basement. My office presented some special challenges because I had no way to fish cable behind the outside wall where I needed to place the network outlet.

Wiring A Patch Panel

Before I could pull any cable, I had to think about where to place a patch panel. A patch panel is nothing more than a block of RJ-45 jacks. Each cable that I ran in my house would terminate at this patch panel. My UGate-3200 Internet gateway would be located very close to this patch panel, and I would use patch cables to connect each cable in my house to a port on the gateway. The following diagram gives you an idea of how this works:

A patch panel is important for a variety of reasons. The cable that you run in a building is fixed in place, so it shouldn't be connected to a moveable device such as a router, switch, or hub. You should run the cable between outlets that are also fixed in place. The cable I used was solid, not stranded. Any significant amount of flexing would break the wires. So that's another reason to pull it once, fix it in place, and not touch it again. Finally, there's the issue of flexibility in terms of configuration. With a patch panel, I can easily change the connections around using the patch cables.

Patch panels cost a bit more than I expected. I ordered most of my supplies from MicroWarehouse, and the least expensive patch panel that I found was a 16 port model by Hubbell that cost $116. I only needed four ports, and I certainly didn't want to spend that much money on ports that I didn't need. As a result, I decided to make my own patch panel using some Hubbell surface mount housings, face plates, and jacks. The following figure contrasts my solution with the standard type of patch panel that you would normally see in a wiring closet:

Patch panels that you buy off the shelf are designed to fit it in equipment racks. They are always 19 inches wide, and have holes on the ends for screws. I had no rack, so that didn't matter to me. The total cost for my solution, including the housings, face plates, and jacks, came to $37.00. The snap-in jacks at $6.00 each accounted for most of that cost.

I really liked working with the Hubbell housings and fittings that I bought. I found them to be sturdy, well designed, and easy to assemble. While working on the patch panel, I learned a couple of things:

• The Hubbell jacks are designed to snap in to the face plates only one way.

• You need to pass the wires through the housing before snapping the jacks in place.

• I should have ordered deeper housings.

Hubbell face plates are marked with a top and bottom, and you need to pay attention to that when you snap in the jacks. I didn't at first, and ended up forcing a jack wrong side up into the face plate. The jacks should snap in easily. If you have to put a lot of pressure on one, you're probably putting it in wrong.

It may seem obvious, but you need to run the cable through the housing box before you snap the jack into the face plate. I missed this at first too, but it was only a minor annoyance because the jacks are easy enough to pull out and snap in again.

Fortunately, the Hubbell jacks snap into the face plate from the back, so you don't need to worry about pulling the wire through that as well.

Finally, I should have bought deeper housings. I ordered the standard-depth surface mount housing boxes, and those were barely deep enough to accommodate the RJ-45 jacks when I screwed in the face plate. I had to push harder than I would have liked in order to attach the face plates. Next time, I'm ordering the "deep" housings.

Pulling Cable to My Office

Once I had the patch panel in place, I could think about pulling cable to my office. I first had to decide exactly where in my office I wanted to locate my ethernet outlet. I also had to consider how best to get the cables to that location. There was no way to get behind the finished walls in my office, so I had to run the cable on the surface. That meant using raceway, and the brand I ended up with was Wiremold, because that's the brand my local building supply store sold. I came to regret that choice, but more on that later.

I faced some interesting challenges running raceway around my office. Two of my walls are covered with flat paneling, while the other two are covered with recycled wooden house siding. The effect looks nice, but the siding is angled and the paneling is not. This made for a big problem in getting the raceway to go around the inside corners, which the following figure illustrates:

I thought about running the raceway across the ceiling, but the ceiling is acoustical tile, and there was no good way of fastening the raceway to it. There is also an air return duct blocking the ceiling path. There was no room to run the wire above the ceiling because the ceiling is flush with the bottom of my floor joists. An additional complication was that the outlet box had to be mounted on the wall with the angled siding. So how was I going to run the raceway down and have it mate with a box installed on a tilt? After giving the matter a great deal of thought, I purchased some 1x6 boards and fastened them over the siding as shown in this next photo:

Because the board covers the raised edges of two pieces of siding, it lays flat and vertically plumb. This solved the corner problem nicely, and by dropping a board down vertically I was able to create a flat surface on which to mount the outlet box. Eventually I'll either paint or stain the board to match the color of the siding.

You can see from the photos that I did not run the raceway at the extreme top of the walls. This was due to the room configuration. The raceway had to be low enough to pass under some casement windows, and yet high enough to clear a counter that is built into one wall. As a result, the raceway ended up being at the level of my chin.

I really hated working with the Wiremold raceway, and if I had the job to do over I would take the extra few days to mail order another brand, probably Hubbell. To start with, the Wiremold raceway was not adhesive backed. This meant that I had to pre-drill holes, and then screw the raceway into my walls. The raceway is very shallow, and I had an extraordinary amount of difficulty getting the cables to stay safely inside as I snapped on the covers. It's also very difficult to reopen this raceway once you have it closed. Finally, the depth of the raceway did not match the predefined openings on my surface mount housing boxes. Instead of snapping out a piece of plastic that had been pre-scored, I had to drill and then cut a hole to match the raceway that I was using. This is most likely a brand issue. I should not have tried to mate one company's surface mount housing boxes with another company's raceway.

The cable that I used was Category 5, 4 pair, solid plenum cable from Belkin. Plenum rated cable is somewhat more fire-resistant than standard PVC cable. It's also twice the cost. I ordered Belkin cable because it happened to be the only brand that Warehouse.com sold in 250-foot lengths. I had no need for a 1000-foot spool. I was satisfied with the cable, but if I ever do something like this again, I'm going to buy a cable stripper for use in stripping the outer jacket. For this job I used a wire cutter to slit the outer jacket. That was tedious, and when I slipped and hit my fingers it was painful.

   

Inside my office, the cables all run inside the raceway. Outside my office, the cables run along the floor joists back to my patch panel. As the pictures above show, I used two methods for fastening the cable. For the single cable to the dining room, I used nail-in cable clamps. For the three cables to my office, I bundled them together and used plastic clamps that I screwed into the floor joists.

The Dining Room Outlet

When I wired the dining room outlet, I bought the face plate and RJ-45 jacks at my local building supply superstore. I also ran the cable inside the wall rather than use raceway. For an outlet box, I bought a blue plastic one that's designed to be installed into an existing wall. You cut a rectangle into the drywall, slide the box in, and tighten up on some screws. The act of tightening the screws raises some plastic "ears", and those are drawn tight against the back side of the drywall to secure the box in place.

Finding the right place to drill up from the basement was my biggest challenge. I needed to come up inside the wall. Too far one way and I'd end up with an unsightly hole in my dinning room floor. Too far the other way would give me a hole in the kitchen. Fortunately, there was a heating duct in the same wall, so I was able to use that as a guide for where to drill. It still took me two tries. The hole I cut for the outlet box was very close to a wall stud. My first drill hole came up inside the wall, but on the wrong side of that stud. I had measured the distance from the heating duct to the outlet box very carefully, and I'm still not sure how I managed to screw that up. At least my mistake is hidden inside the wall where no one can see it.

Whenever you run cable, be sure to leave a bit of slack for future use. My dining room cable has an extra foot or two of slack that's hidden within the wall. This allows me to easily pull the outlet out should I ever need to work on it in the future. I can also afford to cut off the current cable end should I ever need to wire in a new jack.

With the cables going to my office, I left a bit of slack in the outlet box, and I also ran the cable in a loop just before it enters the raceway. That loop is clamped down, but should I ever need a few feet more cable, I can gain that by removing the loop and straightening the run.

The face plate and RJ-45 jack that I used in the dining room came from a company named Pass & Seymour/Legrand. It worked okay, but I wasn't really happy with it. It isn't as sturdy as the Hubbell equipment. The jack tends to snap out of the face plate a little too easily. The jacks also snaps into the face plate from the front, not from the back like the Hubbell jacks do. I didn't realize this until after I had punched down the wires on the jack and tested the connection. I was forced to rip them all loose, feed the cable through the face plate, trim the used ends off, and punch the wires down all over again.

I also had problems getting the face plate to mount flush with the wall. The outlet box stock protruded about 1/16" from the wall because of the way it clamped onto the drywall. The face plate wouldn't cover that, and I was left with an annoying gap between the back of the face plate and the surface of the wall. To fix this, I had to chisel away at the drywall in order to mortise the outlet box in far enough so that it was flush with the painted surface. Then I could get the face plate on flush with the wall.

Connecting to the Net

With the wiring in place, all I had to do was connect everything together using the patch cables. The following figure shows how the patch cables run from my UGate-3200 gateway to my patch panel. You can see the cable modem on the right, next to the UGate-3200. One patch cable runs from the cable modem to the UGate-3200, and four patch cables connect the Ugate-3200 to the patch panel jacks supporting my four computers. The patch cables are color coded to match the jacks. The green, black, and orange jacks in the left-hand outlet box correspond to the same color jacks in my office. The cable from the green jack in my office runs to the green jack in my patch panel, and so forth.

In its out-of-the-box configuration, the UGate-3200 uses DHCP to get an IP address from my cable company. It also acts as a DHCP server to my four computers. This was all fine. Two of my computers are workstations that do not need fixed IP addresses. I went to the TCP/IP properties window, which is shown in the following figure, and quickly verified that my two workstations were configured to obtain IP addresses automatically.

With my computers properly configured to obtain IP addresses automatically using DHCP, I was able to run Internet Explorer and bring up Web pages. I was online.

Cost

So what was the cost for all this? It wasn't cheap. The following table summarizes the cost of all the equipment and material that went into this project:

UGate-3200 Internet gateway     252.98 Raceway, fittings, clamps, patch cables, and jacks     257.64 CAT5 plenum-rated cable     99.99 Total Cost     610.61

My actual out-of-pocket cost was a bit more than the $610.61 listed. I made a few small purchases that never got recorded, and I had to buy a few tools in order to complete the work. I bought a level to keep the raceway straight and level as it wound it's way around my office. I bought a heavy-duty utility cutter with which to cut the raceway, and I had to buy a few drill bits for drilling through the floor, and for pre-drilling the raceway.

What Next?

In my final installment of this series tomorrow, I'll show how I configured my two servers. They both require fixed-IP addresses, so I had to manually reconfigure all their TCP/IP settings. I'll also show the UGate-3200 configuration screens, and talk a little about how that unit secures my network from outside attack. Finally, I'll show you how I opened up one of my machines to the outside so that I can run a low-volume web server.

Jonathan Gennick is a writer and an editor at O'Reilly & Associates. His writing career began in 1997, when he coauthored Teach Yourself PL/SQL in 21 Days (Sams). Since then, he has written Oracle SQL*Plus: The Definitive Guide (1998) and Oracle SQL*Plus Pocket Reference (2000), both for O'Reilly & Associates. In his previous life, Jonathan was a manager and a database administrator in KPMG Consulting's Public Services practice. He has more than a decade of experience with relational databases.