Saturday, July 11, 2026

San Antonio's Datapoint Corporation

This is a special post discussing the work of two San Antonio engineers and the birth of the type of terminals associated with Decwar. Lamont Wood's Datapoint: The Lost Story of the Texans Who Invented the Personal Computer Revolution is highly recommended. Phil Ray in particular is a classic example of a certain type of loveably eccentric Texan and UT Austin graduate from back in the day. Phil certainly would have been someone to hang out with.

San Antonio’s Datapoint, originally the Computer Terminal Corporation, is a fascinating story of computing and the birth of video terminals in the late sixties. It was founded in 1968 by two engineers, Phil Ray and Gus Roche. Ray graduated from UT Austin in 1957. He began his career at Texas Instruments, where he contributed to missile telemetry and electronic speech analysis projects, before advancing to roles in aerospace telemetry at International Data Systems and as a senior staff engineer at General Dynamics' Dynatronics division, supporting NASA's Saturn V rocket and Lunar Orbiter programs. In 1967, while working on NASA projects in Florida, Ray partnered with colleague Gus Roche to explore entrepreneurial opportunities amid anticipated space program cutbacks. Ray recalled discussions with one of his instructors at UT Austin, and events soon took a surprising turn.

Roche and Ray also turned to Bob McClure, then working as a computer consultant in Dallas while teaching as well. Although younger than Ray, he had been a teaching assistant at the University of Texas when Ray was there. Ray had not only been in one of his classes, but the two had worked on an electromechanical demonstration computer, built from relays scavenged from pinball machines, that played a counting game called Nim. “I was sitting in my office in 1968 when I got a call from Charlie Skelton about consulting,” he recalled. Skelton was another Texan and a mutual friend of McClure and Ray, who was at that time involved with Ray and Roche in their effort to start a company. “He and Phil came over and we had a discussion about computer terminals. One of them said they had raised money to start a new firm to build computer terminals. They had read an article in Businessweek saying that computer terminals would be the next big thing but that neither of them had ever seen a computer terminal. By that they were joking, of course. I said I had been thinking about the topic, as I had been doing work in computers and knew that the availability of computer terminals was thin. Nearly all input was done with Model 33 Teletypes at 110 baud. So I said that they should build a glass Teletype. But I told them to make sure that they emulated the protocol of the Teletype exactly, so that no one at the computer end has to change any software, since that would be a hang-up to getting it accepted. There was a CRT terminal already on the market that was not precisely compatible and therefore had a lot of problems. There were some expensive machines available, but I assumed they could make one for less since there was really not that much in one,” McClure recalled. [1]

Following through on that initial advice, Ray and Roche went on to create the Datapoint 3300, released in 1969. The Datapoint 3300 was explicitly designed, named, and marketed as a silent, electronic glass teletype replacement for the loud, mechanical Teletype Model 33. Introduced in 1963, the Teletype Model 33, often referred to as the ASR-33 for its Automatic Send and Receive variant, was a massively popular electromechanical teleprinter. Because it was relatively inexpensive at around $1,000, and was one of the first machines to utilize the newly standardized 7-bit ASCII code, it became the ubiquitous terminal for the early computing era. It printed on continuous spools of paper at an agonizingly slow 110 baud, about 10 characters per second, and its heavy, motor-driven hammer mechanisms generated intense, continuous noise. To solve the noise and speed bottlenecks of machines like the ASR-33, Ray and Roche developed one of the earliest standalone cathode-ray tube displays. This new device was named the Datapoint 3300 and the number 3300 was chosen to explicitly position the device as a direct, futuristic advance over the Teletype Model 33. Datapoint even declared in its marketing that their new terminal was "100 times better than the 33".

To make the transition from paper to screen as seamless as possible for businesses, the Datapoint 3300 was designed to perfectly emulate the Teletype Model 33. It offered complete interchangeability with standard teletypewriter equipment, meaning operators could simply unplug a loud, mechanical Teletype and plug in a Datapoint 3300 to the exact same mainframe connection without requiring any software changes. Because memory was still highly expensive, the Datapoint 3300 used a digital shift-register design to store screen data and featured a display format of 72 columns of text, slightly less than the 80 columns that would later become standard.

The Datapoint 3300 was a massive, immediate hit when it debuted. In fact, demand so vastly outpaced Datapoint's initial manufacturing capabilities that the company had to temporarily outsource the production of the terminal's sleek, streamlined casings to a local San Antonio motorcycle helmet manufacturer. The terminal proved so successful that major computer vendors purchased them to rebadge and sell alongside their own systems. DEC sold it as the DEC VT06, and Hewlett-Packard sold it as the HP 2600A. The explosive success of the Datapoint 3300 provided the capital and momentum to begin developing a smarter successor terminal with its own internal processing power, a project that would shortly become the legendary Datapoint 2200.

The 1970 Datapoint 2200 is considered by some to be the true birth of the personal computer. Though marketed as a terminal, it was actually a programmable desktop system with its own internal memory, keyboard, and screen. Its digital architecture and instruction set was reproduced in Intel’s 8008 microprocessor. While the Intel 4004 hit the market first, the conceptual development for what became the 8008 actually began slightly earlier. Intel initially designed the 8008 on a separate track for Datapoint, who later abandoned the chip. Intel then renamed it the 8008 and released it to the public. It is the ancestor of the x86 standard, and every x86 chip instruction set contains 1970 Datapoint instructions and is in a sense backward compatible with the 2200. The internal architecture of the Datapoint 2200 was a marvel of its time, but it relied on a dense, multi-board processor built from approximately 100 discrete transistor-transistor logic chips. Seeking to reduce the machine's size, heat, and manufacturing costs, Datapoint contracted Intel and Texas Instruments in 1969 to consolidate this entire processor board onto a single silicon chip. Neither company could initially meet Datapoint's needs. Texas Instruments produced the TMX 1795, but it was notoriously buggy and ultimately abandoned. Intel’s single-chip version was severely delayed, forcing Datapoint to proceed with their original TTL design for the Datapoint 2200 to meet production schedules. Because Datapoint declined the delayed Intel chip, Intel retained the intellectual property. Intel released this chip commercially in April 1972 as the Intel 8008, and its instruction set, originally created for the Datapoint 2200, became the direct foundation for the x86 microprocessor architecture that dominates global computing today.

Because the video terminal market was so new and experimental, there were no standards for what the video screen should look like and how it should function. Datapoint decided that it made sense for the shape and visual feel of the screen to match with standard IBM punchcards, since computer users were familiar with punchcards at the time. This is why the Datapoint 2200 screen looks “short” to the modern eye. It matches the shape of a punchcard. In fact, the terminal was designed specifically to replace punchcards, just as the Datapoint 3300 had replaced the Teletype Model 33. The concept was that, in the corporate world data entry was often performed by creating decks of punchcards at remote offices and then sending the decks to a central data processing site. With the 2200, data entry would be done as if it were on punchcards, but on the video screen and keyboard of the 2200 rather than on physical punchcards using a keypunch. The 2200 would store the data internally, and write it to a standard consumer audio cassette tape. This system of storing computer data on audio cassette tapes would become familiar to early home computer users. The 2200 actually had two cassette tape drives integrated into its top surface. Remote offices sent cassette tapes to the central data processing site, rather than punchcard decks. This system turned out to be highly effective in practice, and made Datapoint rich by the mid and late seventies.

Phil Ray in the center and Gus Roche to the right
.
1957 UT Austin engineering grad Phil Ray.

1969 Datapoint 3300 "100 times better than a Teletype Model 33 ASR".

1970 Datapoint 2200 on the right. Father of the x86 instruction set and architecture.

Phil Ray and Gus Roche.

[1] Wood, Lamont (2010). Datapoint: The Lost Story of the Texans Who Invented the Personal Computer Revolution. Hugo House Publishers, Ltd. ISBN 978-1-936-449-36-1.

[2] Ken Shirriff has done many excellent blog posts on Datapoint, and particularly on the 2200 and the x86 instruction set.

No comments:

Post a Comment

San Antonio's Datapoint Corporation

This is a special post discussing the work of two San Antonio engineers and the birth of the type of terminals associated with Decwar. Lamon...