The PC has had its fair share of bus slots. What started with the ISA bus has culminated, so far, in PCI Express slots, M.2 slots, and a few other mechanisms to connect devices to your computer internally. But if the 8-bit ISA card is the first bus you can remember, you are missing out. There were practically as many bus slots in computers as there were computers. Perhaps the most famous bus in early home computers was the Altair 8800’s bus, retroactively termed the S-100 bus, but that wasn’t the oldest standard.
There are more buses than we can cover in a single post, but to narrow it down, we’ll assume a bus is a standard that allows uniform cards to plug into the system in some meaningful way. A typical bus will provide power and access to the computer’s data bus, or at least to its I/O system. Some bus connectors also allow access to the computer’s memory. In a way, the term is overloaded. Not all buses are created equal. Since we are talking about old bus connectors, we’ll exclude new-fangled high speed serial buses, for the most part.
Tradeoffs
There are several trade-offs to consider when designing a bus. For example, it is tempting to provide regulated power via the bus connector. However, that also may limit the amount of power-hungry electronics you can put on a card and — even worse — on all the cards at one time. That’s why the S-100 bus, for example, provided unregulated power and expected each card to regulate it.
On the other hand, later buses, such as VME, will typically have regulated power supplies available. Switching power supplies were a big driver of this. Providing, for example, 100 W of 5 V power using a linear power supply was a headache and wasteful. With a switching power supply, you can easily and efficiently deliver regulated power on demand.
Some bus standards provide access to just the CPU’s I/O space. Others allow adding memory, and, of course, some processors only allow memory-mapped I/O. Depending on the CPU and the complexity of the bus, cards may be able to interrupt the processor or engage in direct memory access independent of the CPU.
In addition to power, there are several things that tend to differentiate traditional parallel buses. Of course, power is one of them, as well as the number of bits available for data or addresses. Many bus structures are synchronous. They operate at a fixed speed, and in general, devices need to keep up. This is simple, but it can impose tight requirements on devices.
