Local Area Network Technology: Part 2 - Hubs

by Jeff Carrell, Electronic Communications Chairman

[reprinted from IEEE - Fort Worth Section Signals May 1997]

Part 1 of this series discussed the first component of the OSI Physical Link Layer - cabling. In LAN technology, the electronics of the physical layer electronics have been referred to as: amplifiers concentrators, hubs, and MSAU's, depending on the topology used. Most of the time, "hub" is the operative word.

Hubs come in many variations: from a few ports, to hundreds of ports; from a small enclosure not much bigger than an apple, to larger than a mini-fridge; from under $50 to well in excess of $25K.

There are basically three types of hub designs:

Unless otherwise stated, this article assumes the use of the RJ45, an 8P8C style connector supported by most hubs.

In Ethernet, there are three different speeds: 10Mb/s, 100Mb/s and 1Gb/s (although gigabit Ethernet is not yet a standard). Additionally, there is the capability to support full-duplex speeds which doubles the effective throughput. Token Ring has three speeds: 4Mb/s, 16Mb/s and 100Mb/s (100Mb/s TR is referred to as FDDI or CDDI). Token Ring also supports full-duplex, but only on 16Mb/s and 100Mb/s technologies.

Many hubs have an LED indicator per port known as "link." This indicates that a device has successfully loaded or been powered up, and the connection between the device and the hub is active and OK. However, depending on many issues a link light does not always guarantee a signal is passing. Bad or improperly pinned cabling can allow some pins to provide a link light, but no signal passing or the physical port of the hub may be bad and still provide a link light.

When connecting devices to an Ethernet hub, a straight pinned cable is used (following either T568A or T568B wiring configurations - see April Signals). When connecting two Ethernet hubs together, a cross-over cable must be used (T568A on one end, and T568B on the other), or one of the hubs must have a port that can be configured to "cross-over," a common feature on higher priced stand-alone and modular hubs. If a cross-over cable is used, a link state LED may be active/on, but no signal will pass. If a straight pinned cable is used to connect two hubs together, usually no link state LED is on, and therefore no signal will pass (this assumes that neither hub has a "cross-over" port capability or it is not in use).

When connecting devices to a Token Ring hub, a straight pinned cable is used (following either T568A or T568B wiring configurations). When connecting two Token Rings hubs together, two straight pinned cables are used to connect a pair of ports designated as "Ring In/Ring Out" (not device ports). Because of the function of the signal is moving in a logical 'ring,' you must extended the ring by connecting the two hubs together so that the ring is maintained and lengthened. You cannot simply connect a device port (IBM "lobe" port) on one hub to another device port on the second hub - no signal will pass. If IBM MAUs or 100% compliant devices are being used, there are special cables available for connecting both devices to the MAU and MAU-to-MAU, since the connector (IBM Data Connector) is not an RJ45 type connector.


In the next article I'll discuss the active electronics that is installed in host devices…NICs.

Copyright © 1997 Jeffrey L. Carrell All Rights Reserved