This document last updated 24 May 2023 (c) Copyright 2023 Herb Johnson. It's going to be a rough-draft and changable, due to circumstances explained in the Introduction.
In May 2023, I came across some discussion of microcomputer history of bussed systems. They were identified in the discussion as "S-100, Apple II, IBM PC, microcomputer slots" and "in the micro[computer] era". The goal was to produce some kind of general "microcomputer bus" exhibit. The initial discussion was on these examples of bussed systems and referenced the IBM PC's "ISA" bus and its origins, developed by Lou Eggebrecht. Others responded with more history about subsequent PC-compatible busses. Or about minicomputers like DEC's which started as "random-wired logic", to busses like Unibus, QBus and the companies which produced bus compatible products.
It seemed to me, chunks of history were missing in the discussion. Also there was no distinction made about what "a slot" meant technically; and not just a thing that accepts boards, and not just a bunch of boards. I recall from years ago, a 1990's description of S-100 computers as "ugly boxes full of boards".
I started to write a reply, but it became a lecture on how microcomputers contributed to the history of computer bus architectures, and why that mattered. Also, some mention of why computers were not *always* bussed, and the micro and mini computers composed of quote "random-wired" boards. It got longer, as I referenced the S-100 and IBM ISA bus history, and other matters of that era.
Rather than post a lecture, and hijack the discussion thread; I constructed this Web page and referenced it, as well as my S-100 bus description Web page I published decades ago. I'll update this page, if the discussion generates some responses and as that discussion resolves. - Herb Johnson
Calling common-bus architectures "slots", puts attention on the boards and not on how the boards connected together. What we technical people (used to) think of as a "bus", is some particular standard of interconnecting signals, among standard-sized backplane connectors, with signals common to each connector. So (for the most part) any bus-compatible card could be plugged into any bus-compatible slot.
This seems normal and reasonable, for many people able to look technically at the systems mentioned in the subject-line. But the history of small computers was not always that "simple".
Connections before bussed boards, was referenced as "random-wired logic". That's a NON-standard, NON-common set of signals per cards and between cards in some kind of computer or control system. Why do that? Because there was a time when a computer was a bunch of boards, with individual logic-chip IC's, or even earlier transistors and passive components. What's called the "CPU" was on several boards; not one chip, not one board. Multiple boards were also needed to support mass storage, or printers, or terminals, or paper-inputs, or ran motors or equipment, etc. etc.
Companies in the 60's, like DEC and IBM and others now forgotten, started producing products like DEC's "Flip Chips", with a set of logic *functions* per board. Those boards were wired together - intentionally but "randomly" - to create industrial controllers and minicomputers. Like the "straight 8" in the VCFed Museum at Infoage. No bus, because no common signals, because those boards were too simple. Because, there were no highly integrated (complex) IC's for those boards.
Bus architectures (as common signals, etc.) began to emerge around the 1970's. Some designers created large boards with many, simple logic chips to perform major functions (CPU, memory, I/O). Minicomputer manufactures created busses for their product lines (brand and models). Mainframe computers had "plug compatible" standards for *cabling* their systems together.
An example of minicomputer buses is DEC's "Omnibus" for their PDP-8's just before the microprocessor era. Here's a PDP-8/f model. Because they were made from small-scale logic chips, these were large boards of lots of TTL chips.
As micro-processors began to emerge in the 1970's, each "family" (brand and model number) had their set(s) of CPU-chip signals for data, addressing, and status/control. Semiconductor companies then made micro-controller IC's, supported by those *same signals* - like UARTs, floppy controllers, and of course memory. And, they introduced boards with those signals at their edges, for interconnection. Those interconnections became the "bus architecture" for those microprocessors. I'm saying: the idea of a bus *emerged* over time.
Example semiconductor companies and busses, started in the industrial world. Intel, with their Multibus (preceeded by intellec development products); Motorola with their VersaBus (preceeded by EXORbus for development but a product-bus as well).
Even so, those companies and others, also produced "computers on a board", with CPU and memory and peripheral chips on ONE board. Some of those boards had "busses", some had "I/O connectors", some had both. The KIM from MOS Technology (6502) is a popular example. So "random" or non-bus signaling on a connector was also in use on some microcomputers.
Meanwhile, companies (often very small ones) started producing "compatible" boards that were based on those same bussed signals and connectors. Those signals and connectors then were named a "bus standard". I'm saying, the standards came *after* the products were designed and produced. Later, companies got wise and established a bus-standard *before* they announced a new architecture.
An example of "standards after"? The so-called "S-100". It began when MITS sold the "Altair 8800", with a 100-pin connector backplane with common signals between the MITS 8080 CPU board and various memory and I/O boards (some not yet designed!). They called it "the Altair bus". For many reasons, the MITS 8800 became popular, thousands were ordered in months, Many claim the Altair 8800 was the first popular and widespread personal computer.
When the Altair 8800 became popular, lots of little companies produced "Altair compatible" boards. But they got tired of naming the Altair in all their advertizing and documentation - MITS was a competitor. So they chose to call the bus "S-100", and changed some of the signaling. Some years after that, S-100 producers redesigned the bus signaling, and called that version "IEEE-696". I cover this S-100 bus history on my Web site at this link.
I'd certainly claim, that by MITS establishing their Altair Bus as the Intel 8080 became relatively inexpensive and accompanied by a host of Intel supporting chips, the MITS 8800 started a bus-compatible industry; one of many based on a bus architecture.
A similar history occurred, for the SS-50 bus of SouthWest Technical Products Corporation (SWTP or SWTPC). And they never got around to "standardizing" their bus! I have a Web page discussing the SS-50 as I worked on a SS-50 Z80 board.
An example of "standards before"? The IBM PC of 1981. IBM was well familiar with standards - and keeping them proprietary, secret - when it decided in the late 1970's to compete in the microprocessor-computer market for business (and personal-business) use. Management saw how the microcomputer market worked at that time - open bus standards, common operating systems, schematics, documentation: like the Altair/S-100 experience. That was the *opposite* of IBM's practices of closed-source, proprietary documentation, and lawyers to keep competitors away from IBM's customers.
So in a bold move, IBM created a design group that *ignored IBM's prior practices* and adopted the market's strategies. The IBM PC documentation included schematics! *Source code* for the BIOS! And, IBM defined their PC bus architecture - calling it "Industry Standard Architecture" or ISA - and did not restrict other's use!
Almost immediately, many small and large companies produced "PC compatible" or "ISA compatible" cards. Thus IBM's success with the PC, was because of the prior market created by the MITS Altair, the IMSAI, and other S-100 products; as well as similar architectures for the 6800/6809 (Motorola's busses, SWTPC's SS-50, Gimix), and hosts of other microcomputer brands and microprocessor producers now considered obscure. IBM created their own market for their IBM PC, and enlarged that market thanks to allowing their competitor's compatible products. Eventually that led to a PC market bigger than IBM could compete in, but that's another story.
An example of a kind of "multi standard", was a less-than-successful microcomputer with multiple busses and different microprocessors. That's the namesake computer from from "the Digital Group", of Denver Colorado (USA). Their computer and company were not in production for very many years, TDG had business issues, and it's now an obscure reference. But it's plausible to say that because it did not have a singular bus or singular microprocessor, few other companies produced products that worked with that computer.
Meanwhile, non-bussed computers became popular in the 1980s, such as the video-gaming home-computer systems produced at the time by Atari, Commodore, and others. Their means of expansion were cartridges for game/program ROMs, and various board-modules or specialized connectors, to add memory or features or peripherial. Some computer models used a *serial bus*, like the Commodore models' serial version of the HP-IB or IEEE-488 bus. So these models, teetered between some kind of means of expansion, versus the reduced cost and size of avoiding a bussed backplane.
By the 1980's, microprocessors and peripheral chips and memory were "concentrated" enough (had the capacity) such that a single board could provide a complete-enough computer for use. Of course, different eras defined what was considered "complete". Also, lack of a bus, did not stop people from producing board that could be wired into some of these products. All means to add new chips with new features to old computers. But a bus, makes that much easier.
By the 1990's, there was a "bust" in home video gaming computers. It's another subject, to discuss why that occurred. (And it's out of my areas of experience.) But in this context of bussed computers, one consideration in 1990 became the capability of (IBM)PC-compatibles and Apple Macintosh computers, to perform gaming as well or better than those now-aging, non-expandable systems. And thanks to VGA graphic cards, denser memory chips, and faster processors - and a buss to accomodate some of them - these computers *could* be upgraded.
The dominant personal computers of the 21st century, all became bussed systems and expandable; or of course replacable. But like the cheaper computers decades ago, there's also room for software-compatible but non-bussed personal computers in the Windows/Linux/OSX world as well.
The point of this Web page is to show that there was an entire generation of bus-compatible microprocessor-based microcomputers, each unique, before the recent era of only a few bus-standard computers. And, there was a time before bus architectures, because there was no means to put complex logic on a small-enough board to support a common "bussed" connection at all.
Common busses, not only support means to easily adapt
computers to new or different hardware and devices; they create a
standard that permits other companies to provide competing products. A
bigger market, plus ever-more-complicated and compact integrated
circuits, led to lower prices for more features and more capability.
"Moore's Law" applied to whole boards and computers, not just chips,
thanks in part to standards that avoided multiple products to cover each
computer, each brand and model. That all had a beginning, prior to the
world of S-100 and Apple II's and IBM PC's which established microcomputers
with bussed architectures. - Herb Johnson
Copyright © 2023 Herb Johnson
New Jersey, USA
To email @ me, see see my ordering Web page.
Copyright © 2023 Herb Johnson