Since we are not studying electrical engineering, we shall gloss over some of the messier details on how devices connect to buses. However, a few words are in order.

First, many buses have slots into which connectors are plugged. This is electrically identical to wall receptacles in your house where you plug in electrical appliances so they can get the juice to operate. Actually, a better analogy might be the telephone plug in the wall. Any telephone receiver may be plugged into different houses and it will respond although it acquires a different address, or phone number, when it is moved to a different place. In a similar way inside a computer, the slots have addresses to which they respond.

There is usually an interface chip which sits between the actual bus wires and the device that logically plugs into it, such as memory or the CPU. This interface chip may do useful things like match voltage or amplify the signals so that they will be strong enough to assert logic 1 or 0 on the bus wires. Internal buses seldom have these problems since they are always connected and the chip designer knows how much current is passing through at any point and can match things up.

External buses, such as system buses, often connect heterogeneous equipment, often from different vendors, and it must all work together. A new trend, plug 'n' play as it is called, has emerged with Windows 95 (although a form of it was present in some earlier computers such as the Macintosh). In this system, no special addressing must be done in order to get the other devices on the bus to recognize the unit just plugged in. This is done automatically by the devices through sending and receiving signals to each other. In the (recent) past, you would have had to have changed configuration files when you added new hardware to your IBM PC or clone.

At the electronic level, tri-state buffers are often used to electrically disconnect the wires so that one device can be totally unaffected by what is going on in the bus, although it can be reconnected in nanoseconds by simply sending a signal to the tri-state buffers. Another type of connection is called a wired-or where a number of wires carrying logic values are electrically connected to a common wire. If one or more of those wires gets a high voltage, signaling logic 1, the common wire also gets a high voltage, thus getting logic 1. Since voltages are additive and since logic 1 is usually defined to be anything of +12V or higher, this makes a cheap form of OR gate.

Suffice it to say that it is a challenge to get different digital devices to communicate effectively on a single set of wires. In order to avoid reinventing the wheel every time a new computer or new device is introduced, hardware designers make up a list of characteristics regarding their bus and what can connect to it. We call such a standard a bus architecture. Up till recently, every vendor has its own bus architecture. A few of them are Massbus, Microchannel, Omnibus, Q-bus, Multibus and S-100 bus.

Recently there has been a trend towards more standardization, as exemplified by the SCSI-bus that allows many different devices to interconnect, such as printers, disk drives and CD-ROM players. SCSI stands for "Small Computer System Interface." In the past, vendors did not want users to be able to mix and match because they might buy components from competing vendors. Thus, locking customers into just one line of equipment was standard practice for decades and is only now vanishing as vendors see that having an open architecture allows other vendor's customers to come to their own market.

An example of using a bus standard to try to accomplish a business goal was IBM's Microchannel that was part of their PS/2 line of computers, introduced around 1987 and meant to grab back market share stolen by all the clone vendors of the immensely successful IBM PC. It did not help IBM at all and in fact many shunned the PS/2 because of its incompatibility.