| Table of contents |
|
2 Description 3 Software |
After leaving the US Air Force, Ed Roberts decided to use his electronics training to produce small kits for hobbyist uses. Along with a small group of other ex-Air Force technicians, Roberts created MITS (Micro Instrumentation Telemetry Systems) in his garage in Albuquerque, New Mexico, and started selling radio transmitters for model airplanes. In 1969 Roberts bought out the others and moved to a larger office, where he manufactured calculator kits for hobbyists. In 1972 Texas Instruments developed their own calculator chip and started selling complete calculators at less than half the going rate. MITS was devastated by this (as were many companies) and Roberts returned to being a dentist while trying to reduce his quarter-million dollar debt load.
With the release of the Intel 8080 in 1974, a number of hobbyists started designing microcomputer kits. In July 1974 one such design was advertised in Radio Electronics magazine, Jonathan Titus' Mark 8. The design was purely on paper, requiring the builder to track down the parts one at a time, a task that was basically impossible outside of California. Although the Mark 8 was not a success, the editors at Popular Electronics realized that someone was going to be the first to deliver a "real" kit, and decided they wanted to do it. At this point the story becomes somewhat less clear.
In one version, told by the series editor Art Salsberg, they had another simpler design in the works by Jerry Odgen, a long time contributor. This design was a pure "hack" however, and Salsberg realized they needed something much better to counter the Mark 8. At about this time Roberts approached them with his own design, based on a pre-printed circuit board rather than hand-soldered wiring. Les Solomon's version of the story is similar, but places the key meeting between himself and Forrest Mims, another contributor to the magazine, who told him about Robert's project. Forrest Mims' own version places him outside the loop, with Solomon and Salsberg deciding to call Roberts.
Whatever the truth, a deal was soon struck. Roberts looked for a deal on CPUs, and eventually talked Intel into supplying him with cosmetically blemished 8080's for $75, when they normally sold for $360. In fact the deal wasn't quite as shrewd as Roberts thought at the time; Intel chose the $360 price simply as a play on the famous IBM System/360 mainframe. The name finally decided upon for the computer came from Soloman's 12-year-old daughter, Lauren. She suggested Altair, which was the destination for the Enterprise during an episode of Star Trek that she was watching.
The first working sample was immediately shipped, by train, to New York. However, it never arrived due to a strike by the shipping company. The first example of this groundbreaking machine is thus lost to history. Solomon had already taken a number of pictures of the machine and wrote the article based on them, while Roberts got to work on building a replacement. Everything came together, and the kit was officially available on December 19, 1974.
The kit was first announced in the January 1975 edition of Popular Electronics.
The timing seemed to be just right. The electronics hobbyists were moving on to computers as more and more electronics turned digital, and yet they were frustrated by the low power and flexibility of the few kits that were already on the market. The Altair had enough power to be actually useful, and was designed around an expandable system that opened it up to all sorts of experiments. Roberts needed to sell 200 over the next year to break even, but instead received thousands of orders in the first month, including 200 in one day.
Within only six months competition arrived in the form of the IMSAI 8080, which included a keyboard, monitor and a floppy disk controller. Roberts was furious, and spent an increasing amount of his time trying to "knock off" these competitors instead of improving the Altair. By 1976 there were a number of much better built machines on the market, and when Roberts started demanding the newly-appearing computers stores sell only Altair machines, they instead turned to the competition and MITS was quickly squeezed out of the market they had created.
In the first design of the Altair, the parts needed to make a complete machine would not fit on a single motherboard, and the machine consisted of four boards stacked on top of each other with stand-offs. Another problem facing Roberts was that the parts needed to make a truly useful computer weren't available, or wouldn't be designed in time for the January launch date.
So during the construction of the second model, he decided to build most of the machine on removable cards, reducing the motherboard to nothing more than an interconnect between the cards, a backplane. The basic machine consisted of five cards, including the CPU on one and memory on another. He then looked for a cheap source of connectors, and came across a supply of 100-pin edge connectors. The rest, as they say, is history.
For all intents, the Altair bus consists of the pins of the Intel 8080 run out onto the backplane. No particular level of thought went into the design, which led to such disasters as various power lines of differing voltages being located next to each other, leading to easy shorting. Another oddity was that the system included two unidirectional 8-bit data buses, but only a single bidirectional 16-bit address bus. A deal on power supplies led to the use of +8v and +18v, which had to be "pulled down" on the cards to TTL (+5v) or RS-232 (+12v) levels.
The Altair shipped in a two-piece case. The backplane and power supply were mounted on a base plate, along with the front and rear of the box. The "lid" was shaped like a C, forming the top, left and right sides of the box. The face plate, reportedly inspired by the Data General Nova, included a number of large switches to feed data directly into the memory of the machine, and a number of LEDs to read those values back out.
When the machine first shipped the switches and lights were the only interface, and all one could do with the machine was make programs to make the lights blink. Nevertheless, many were sold in this form. Roberts was already hard at work on additional cards, including a paper tape reader for storage, additional RAM cards, and a RS-232 interface to connect to a terminal.
Around this time Roberts received a letter from a Seattle company asking if he would be interested in selling their BASIC programming language for the machine. He called the company and reached a private home, where no one had heard of anything like BASIC. In fact the letter had been sent by Bill Gates and Paul Allen from the Boston area, and they had no BASIC to offer. When they called Roberts to follow up on the letter he expressed his interest, and the two started work on their BASIC interpreter using a self-made simulator for the 8080 on a PDP-10 minicomputer. They figured they had 30 days before someone else beat them to the punch, and once they had a version working on the simulator, Allen flew to Albuquerque to deliver the program, Altair BASIC (aka MITS 4K BASIC), on a paper tape. Miraculously it worked the first time, and Gates soon joined him and formed Microsoft, then spelled "Micro-Soft".
Another piece of groundbreaking software was the CP/M operating system, which could handle floppy disk drives attached to the machine, to make it considerably more useful. CP/M had been written by Gary Kildall starting in 1973, and was the only such system on the market when the Altair appeared. By 1976 it was the most-used operating system in the world.History
Description
Software