Token Ring Physical Layer
I mentioned in the introduction that, for all intents and purposes, the IEEE 802.5 specification for Token Ring LANs is the same as the IBM Token Ring, on which the IEEE specification was based. Our description of the physical layer will be in accordance with the IBM Token Ring, since that is how the LAN was marketed, and the IBM Token Ring had certain characteristics (such as the hub-based architecture) that were not specified in the IEEE specification. However, it should be noted that devices conforming with the IEEE 802.5 specification could attach to and communicate over an IBM Token Ring.
Token Ring was originally implemented over shielded twisted pair (STP) cable using an IBM Data Connector. This method allows a maximum of 260 stations at either 4 Mbps or 16 Mbps (which doesn't at all seem like much, now, but remember that this was at a time when the main competition was 10 Mbps Ethernet). Though later specifications defined higher speeds, such products were never marketed.
Because STP is more expensive (both material and labor) than unshielded twisted pair (UTP), the option of Token Ring over conventional UTP (using RJ-45 type connectors, which are also less expensive than the IBM connectors) was also defined, with the tradeoff being fewer stations (maximum of 72).
The Token Ring’s wiring hub is called a concentrator, or Multi-Staion Access Unit (MSAU), and the original MSAUs had eight ports, as depicted in Figure 31.
Figure 31: Token Ring Multi-Station Access Unit
In Figure 31, we see two MSAUs "daisy-chained" together (with patch cables, in green) to form one larger ring. The Ring In port on one hub must be connected to the Ring Out port on the other, and vice versa, thus creating a ring. Each connected station is also part of the ring, and if a station is removed (or a port is not used), then a bypass switch would close the ring at that port. (If there were not such a feature, then a failing station or link would "break" the ring.) The red dotted path shows the path of the ring. As you can see, the cable that connects a station to the MSAU (called a lobe cable, indicated by a double black line) is actually a double cable, with two sets of twisted pair, one for send and one for receive, or one for "ring in" and one for "ring out". Thus, each station actually acts as a repeater on the ring.
Another feature of the IBM Token Ring is "dual ring redundancy". With this feature, if a several MSAUs are "daisy-chained" together (as shown in Figure 32) and the cable that connects two of them breaks, the signal will "wrap" and then be routed in the opposite direction, in order to complete the ring, as depicted in Figure 32. Thus, not only is a ring resilient against station failure; it is also resilient against cable failure. However, with the original hubs, this ring recovery required that both ends of the failing cable be removed from the hubs manually - not something that would be very practical in a large ring.
Figure 32: Cable Break in an IBM Token Ring
In Figure 32, the green and red path is the normal transmission path. The dotted green line that passes through a hub is my "short-cut" way of indicating that the ring actually passes through all of the stations, as I attempted to depict in Figure 31, above. Since the cable that connects the middle hub with the bottom hub failed, the signal wraps (little blue line) in the middle hub, and is rerouted into the purple redundant path, the ring being completed because the signal also wraps in the bottom hub.
