News

The birth of the personal computer: everything you know is wrong

Unless you lived in Texas in the 70s

Jon Peddie

This could be an entertaining or totally aggravating, lawyer invoking party game—ask when the personal computer was invented. Someone did—ask that is. There’s a charming site located in Seattle called Blinkenlights Archeological Institute. It was established in 1997 to excavate, preserve, research, and present interesting and historically significant computing devices. The website, http://www. blinkenlights.com/pc.shtml, lists seven qualifications of what constitutes a PC and then goes on to list the relay-based Minivac and the Simon that used electric rotary switches.

So, in reading the site’s entries, it occurred to me that additional classification is needed: the type of switch—i.e., relay, vacuum tube, delay lines, transistor, integrated circuit, bios…etc. Some people may say that to qualify as a PC it has to be solid-state.

Blankenbaker’s Kenbak-1, circa 1971. (Photo courtesy of John Blankenbaker) IMSAI circa 1975. (Photo courtesy of the Obsolete Technology Website.)

 

And asking that question exposes a qualifying question in return, doesn’t it? How much of it had to be built using a specific technology? (i.e., solid-state). For example, the first computers to use transistors also had vacuum tubes, and the first to use ICs also had transistors, etc. So I’d say it’s the first occurrence of a specific technology in the machine and, therefore, it does not require that the entire machine be built with the newest technology.

So, refining the question, what was the first PC to use a(ny) ICs? Some might say the Apple 1, built by Steve Wozniak in 1976 based on the 6502 microprocessor. 

In reality, other PCs like the Datapoint 2200 and the Xerox Alto were selling up to six years earlier, and they didn’t have wooden cases or were built with second-hand parts.

Probably most of you would say the first computer was the MITS Altair or the IMSAI 8800 in 1975, based on the Intel 8080 processor. 

And, in France in 1974, the Micral-N microcomputer was sold as a fully assembled system rather than a kit. Based on Intel’s 8008 CPU, the Micral-N was programmed by Philippe Kahn, who later immigrated to Silicon Valley and founded Borland.

They were newcomers to the market

In 1971, John Blankenbaker built the Kenbak-1 before there were microprocessors, and he used logic chips. Blankenbaker’s machine was program- mable. It was introduced in the spring of 1971, according to Blankenbaker’s website http://www.kenbak-1.net/.

In between the late-comers like MITS and Mircal, HP brought out the HP9830, which had a 16-bit microprocessor, in 1972.

The processors

The first commercial microprocessor was built by Intel, the 4004, in 1971, and it is generally credited with being the first complete CPU on a chip, albeit only 4-bits.

But before the 4004 was built, a small Texas firm, Datapoint Corp., was trying to convince Bob Noyce in 1969 to produce a more powerful processor, which Intel ultimately did and it became the 8008; it was introduced in 1974 and used in the famous MITS Altair that Bill Gates fell in love with. The 8008, which was developed primarily for Datapoint (with NRE from Datapoint) spawned the 8080.

Intel and Datapoint (then named Computer Terminal Corporation, CTC) were both founded at about the same time in 1968. In the summer of 1969, Intel was approached by a Japanese company that wanted to build a scientific calculator. Intel convinced the customer that would be better to go with programmed logic rather than hard-wired logic since it could be done with four chips rather than a dozen or more. The project then stalled because Intel did not have any chip designers, according to Lamont Wood, a former Data- point employee.

During that time, CTC raised money for their second product. Their first one, the Datapoint 3300, was a successful teletype replacement that used shift registers to store the screen contents. Semi-conductor RAM memory was not available. They decided their second product would be a replacement for the IBM 029 card punch machine using programmable logic rather than hard-wired logic. They called it the Datapoint 2200. They hired a number of people for the new project, including Vic Poor, and they drew up specifications for a simple 8-bit computer. It, too, used shift registers for memory.

Around Christmas of 1969, Poor visited Intel. Poor says he was interested in Intel building a single-chip version of the CPU for the eight-bit 2200. He believed a single chip would have ad- vantages for space and heat. There is some dispute about this version. Intel engineer Stan Mazor insists that Poor was only inquiring about a single-chip 512-bit pushed-down stack. Mazor says that he came back to Poor with several proposals, one of which was to put the entire CPU on a chip. He thought that Poor would assume the single-chip idea was crazy, but he knew it would work because of the work he had been doing designing the four-bit 4004.

Either way, CTC wanted the CPU chip. Jack Frassanito, one of the founders of CTC, remembers being part of a business meeting with Gus Roche in early 1970 in which Bob Noyce (head of Intel) turned down the idea of making the eight-bit processor chip, saying it could be done but they would be stupid to do it. Apparently, he feared that the computer vendors who were buying In- tel’s memory chips would see Intel as a competitor and stop buying their chips. But he agreed to a development contract with CTC. The terms of the agreement remain obscure and Intel does not have a copy of the contract in its archives. Some say it was a bet between Intel and CTC. Some say there was a six-month deadline. (Intel does have the 4004 contracts in its archives. Curiously, it specifies no deadline, according to Wood.)

Work began on the CTC chip in March 1970 after Intel hired a chip de- signer, and work also began on the 4004 at nearly the same time. However, Intel put the CTC chip on the back burner during the summer of 1970, when it seemed that CTC had lost interest in the single-chip design. Time was growing short, CTC had to go into production, and they had made changes in the architecture that substantially increased performance. Apparently, CTC felt they could not wait for the new chip from Intel. Instead, CTC went on to build the Datapoint 2200 with a board-level CPU based on TTL logic. Shipment of the Datapoint 2200 began in early 1971. Despite myths to the contrary, it was designed, sold, and used as a desktop business computer rather than a “programmable terminal.” It was immediately popular with those who understood what it could do. (Few, if anyone, ever used it for IBM 029 emulation, however.)

Intel revived the CTC chip project in early 1971 after they got samples of the 4004 working when another Japanese calculator company expressed interest in a chip using the same kind of design as the CTC chip.

Meanwhile, Texas Instruments also built a version of the chip for CTC. It was delivered in March of 1971 and only worked for a few seconds a time. CTC rejected it. TI, however, went on to patent it. The TI design was based on the Intel specification done for CTC, and there is some dispute as to how TI got hold of the specification.

Intel delivered its version of the chip to CTC at the end of 1971, CTC rejected it as grossly overdue. Also, CTC had since gone on to develop Version II of the 2200, which embodied a number of improvements and was much faster (thanks in part to its use of RAM memory chips.)

At the same time, Intel put the 4004 in its catalog. It is now considered the first microprocessor, but Frassanito and Wood contend the CTC chip would have come out first if development had been continuous.

Intel added the CTC chip to its catalog in April 1972, calling it the 8008, or double the 4004. It had little in common with the 4004, because it used standard memory chips rather than the custom support chips used by the 4004. Since it used an 18-pin package (the same as their latest RAM chip), the 8008 required as many as 40 support chips to de-multiplex the pins, undercutting the rationale for a single-chip processor. It was adopted as a controller and so it did not anger the established computer vendors. Intel later enhanced it as the 8080, with a rational number of pins, and founded the x86 PC dynasty.

TI later sued the PC vendors based on its patent, and, as a  result, talking to the participants, says Wood, “is like entering a room where an argument has ended, since they were all deposed, some multiple times, with contentious results.” Since the case never went to trial, their depositions remain confidential. How- ever, adds Wood, “I understand that TI had to settle for pennies per PC.”

The first person computer 

CTC couldn’t wait for Intel, and so it introduced its first machine, the Datapoint 2200 in 1970, using medium-scale integration (MSIs), like Blanken- baker’s machine. CTC’s founders claim the 2200 was the first programmable personal computer and the Datapoint line was so popular that CTC changed its name to Datapoint. 

The world’s first PC, the Datapoint 2200—1970—notice the 70’s Texas hairstyle. (Photo courtesy of Jack Frassanito)

Accordingly, subject to the definition of what qualifies as a PC, Datapoint claims to be the developer of the first programmable PC that used ICs—and that was in early 1970. 

It was also the first to have a built-in keyboard and use a built-in CRT (with an 80 x 12, 7-bit-character display) which was chosen to mimic an IBM card layout (according to Vic Poor, the CTO of Datapoint.) The original target market was defined as a plug-compatible replacement for an IBM 029 card punch machine. “As far as I know we never sold one as a 029 replacement,” said Jack Frassanito (one of the founders of Datapoint).

In the original design, there was a row of register switches just above the keyboard. “We were going through a design scrub and every element was scrutinized to understand whether it was actually necessary, and I decided we should remove them,” said Frassanito. “As it was a business computer, so we did everything we could to simplify the design and make it less intimidating.”

There’s also the distinction of being a production system vs. a technology demonstration or a kit. And, one could argue about the keyboard—a PC should have one. Prior to the Datapoint unit, the console or keyboard was an ASR33 teletype. Mainframe and time-share computers had CRT terminals with built-in keyboards.

In addition, as Jack points out to qualify as a Personal Computer, it should support word processing, email, spreadsheets, and other PC applications. Datapoint, says Frassanito, wrote spreadsheet, word processing, and email programs and licensed them to Bill Gates. “We have the attorney who did the license agreement on the record,” he adds.

Epilog

Early PCs that came after the Datapoint 2200 also used CRT terminals, and two of the first to incorporate CRTs and keyboards into a consumer PC were the 6502-based Commodore and Radio Shack’s machine. These PCs were the first to offer bit-mapped graphics.

Commodore PET, circa 1977, with built-in 9-inch CRT.
(Photo courtesy of the Obsolete Technology Website.)
Tandy Radio Shack TRS-80, circa 1977. (Photo courtesy of the Obsolete Technology Website.)

 

So, this marks the 29th anniversary of the personal computer and the beginning of an era that was one of the most revolutionary developments in history, and it all started in San Antonio Texas—betcha didn’t know that.
Frassanito, Wood, and others are finishing up a book on the Datapoint story titled: “Stolen Thunder, the Lost  Story of the Texans Who Invented the Personal Computer Industry.” ..&