Along with hubs came another form of Ethernet cable and connector: unshielded, twisted pair (UTP) wiring and RJ-45 “flat” connectors. These cables use multiple twisted pairs of wires instead of the coaxial cable specified in Metcalfe’s original Ethernet. The purpose of using twisted-wire pairs is to reduce magnetic signal coupling.

295

For 10 Mbps Ethernet over UTP cable (called 10BASE-T) and for 100 Mbps Ethernet (called 100BASE-TX), only two out of four available wire pairs are used:

296

It should be noted that 1000 Mbps (“Gigabit”) Ethernet over twisted-wire pairs does in fact use all four pairs in an eight-wire cable: a departure from traditional UTP Ethernet cable wiring.

298

Along with UTP cables and RJ-45 connectors came a significant alteration to the basic electrical scheme of Ethernet. Metcalfe’s original design used a simple coaxial cable as the “ether” connecting devices together. Such cables had only two conductors, meaning each device needed to transmit and receive data over the same two conductors. With UTP cable’s four pairs of conductors, transmission and reception of signals is handled over different wire pairs. This means connections made between Ethernet devices must employ a “swap” between TD and RD wire pairs in order for communication to take place, so that the “receiver” circuitry of one device connects to the “transmitter” circuitry of the other, and visa-versa. This is precisely the same problem experienced inherent to EIA/TIA-232 and four-wire EIA/TIA-485 networks, where separate wire pairs for “transmit” and “receive” are different.

In a typical Ethernet system, the interconnecting hubs perform this transmit/receive swap. Hubs are considered DCE devices, while computers and other end-of-the-line devices are considered DTE devices. This means the pin assignments of DTE and DCE devices must be different in order to ensure the transmit/receive pin swap necessary for straight-through cables to work. This also means  if someone ever wishes to directly connect two Ethernet DTE devices together without the benefit of a hub in between, a special crossover cable must be used for the connection, identical in function to the null modem cable used to connect two EIA/TIA-232 DTE devices together:

299

Furthermore, the same problem exists when multiple hubs are connected to form larger networks. Since each hub is a DCE device, a straight-through cable connecting two hubs together will pass transmitted signals from one hub directly to the “transmit” pins of the other hub, not the “receive” pins as it needs to. Consequently, a “crossover” cable should be used to connect two Ethernet hubs together in order to avoid this problem:

300

Some early Ethernet hubs provided a different solution to the “crossover” problem, and that was a crossover switch built into the hub, allowing a person to manually switch the transmit and receive wire pairs with the push of a button. In this next photograph of a four-port Ethernet hub, you can see the “Normal/Uplink” pushbutton on the right-hand side of the front panel, controlling the furthest-right port of the hub. This switch is supposed to be placed in the “Normal” position if the device plugged into that port is a DTE device, and placed in the “Uplink” position if the device is a DCE device (e.g. another hub):

301

Note the LED indicator lights by each port on the hub. One LED indicates whether or not the cable is active (when a powered Ethernet DTE device is plugged into that port of the hub), while the other LED indicates traffic on the cable (by blinking). These LEDs are very helpful for identifying a crossover problem. This hub even has an LED indicating the occurrence of collisions (the “Col” LED just below the main power LED), giving simple visual indication of collision frequency. Some modern hubs use auto-sensing technology to perform any necessary transmit/receive pin swaps, rendering crossover cables and crossover pushbuttons unnecessary for hub-to-hub connections. 1000BASE-T (“Gigabit” Ethernet) hubs have this as a standard feature.

Leave a Reply