A large number of the questions that we have been getting during our Random Access sessions has been related to networking computers - either in a local network or connecting to the internet. This article (which is the first of several installments) will address some of the details as to how it works. Note to any EEs out there: some details will be sacrificed for simplicity. Since ethernet is by far the most popular way for creating a wired network, I will start with the cabling that is used for creating a wired home network. The wiring must be Cat-5 (for Category 5) twisted pair cable. This cable had its origins in the telephone industry, but is most definitely not plain telephone wire. Lets start with a bit of basic physics: When electricity flows through a wire there is a discernable magnetic field (disturbance) around the wire. If the wire is spun into a coil (think of a spool of thread) the magnetic field is greatly enhanced - and you have an electromagnet. Now the interesting partif you have a magnetic field and a wire moves through it, the field can induce an electric current. This is the principle behind a generator. But physical movement isnt the only way to induce a current - it can be induced if the field strength varies rapidly. Have you ever been on a telephone call and faintly heard someone elses conversation? What is happening is that in some cable somewhere pulses from that conversations pair of wires are being picked up by your pair of wires - a phenomenon known as cross talk. Data being sent through a pair of wires is done via pulses of ON and OFF voltages, millions of times (megabits) per second. That most definitely qualifies for rapid variation of field strength. The way to reduce the cross talk to an acceptable level is to make use of the fact that the signal is traveling up one wire and back down the other in the pair. By wrapping (or twisting) the wires of the pair around each other, the magnetic fields cancel each other out - voila - no cross talk. Cat-5 cable has 4 pairs of wire. They are color coded: orange and orange/white stripe, green and green/white stripe, blue and blue/white stripe, and brown and brown/white stripe. If you get a magnifying glass and look at the plug end of a regular Cat-5 cable, you will see the individual wires in this sequence shown below as End A Only the orange and green pairs are used for ethernet. In a regular connection a patch cable is used to connect a PC to a hub or switch or modem device. If the PCs network card is set to send on the orange pair and receive on the green pair, then the other device must be set to receive from the orange pair and send on the green pair. Fortunately, you usually connect a PC (a so-called terminal device) to a hub, switch, router, or modem (a so-called communications device) so the manufactures set them up as to which pair in the jack are send and which are receive. At this point it might be well to explain
the difference between a hub, switch, router and modem. A switch has some intelligence within it. It keeps track of which device is attached to which of its ports (sockets) and directs the data transmission to the intended device without putting it on the cables for the other devices. This greatly reduces the amount of data traveling over the network cables. A router is a specialized switch. It knows which devices are local and which destinations are somewhere else - such as the internet. It determines whether to make a local connection or steer (route) the transmission to another router (say at your ISPs site) and hand it off. Typical broadband Cable Modem/DSL Routers (such as by LinkSys, D-Link, Belkin, SMC, Netgear, etc.) also perform other functions which I will describe in a future article. Since the price of a router is marginally (if at all) more than a hub, you wont find many hubs being made or sold today. A modem is a device which converts the local ethernet signal to something that is suitable for transmission over your broadband connection - such as DSL or cable. It does bidirectional conversion, i.e. ethernet to broadband, or broadband to ethernet. For an absolutely bare-bones network, you may connect two terminal devices, such as two PCs, back to back using a specialized Cat-5 cable, called a cross over cable. (Sometimes the cable will have the abbreviation X-over on it.) In this special-case cable, the wires at one end of the cable are in the plug in a sequence different than the other. In the diagrams below I only show two of the four pairs. You can now see that a signal put on (say) pair 1-2 at one end will come out on pair 3-6 at the other, and visa versa. This effectively corrects the transmitter of one device to the receiver of the other, and visa versa. Just as two similar terminal devices may be connected via a cross-over cable, two similar communications devices may be connected via a cross-over cable, effectively making a larger switch or hub. Another reason to have multiple switches or hubs would be to place them in the centers of work groups, thus reducing the amount of cable necessary to connect the clusters of work groups to each other. For example, in an office that occupies more than one floor of a building, there might be one switch per floor, with an uplink connection connecting the switches on each floor. More often, however, one of the ports will be designated an up-link port which has the cross-over function implemented via an internal mechanism. In the next installment I will discuss network cards and how to configure them and test your configuration using some very basic tools included in Windows - ping, winipcfg, ipconfig, and tracert. |
Bruce Preston is president of West Mountain Systems, a consultancy in Ridgefield, CT specializing in database applications. A DACS director, Bruce also leads the Access SIG. Members may send tech queries to Bruce at askdacs@dacs.org. |