The History of the IBM 5100
IBM's First Personal Computer
Today, most people consider the IBM PC 5150 to be IBM’s entrance to the personal computing market. The 5150 is arguably also the machine that created the market for personal computers as prior to the 5150, computers were largely incompatible with one another, they were not common household items, and until the world wide web was made accessible with Windows 95 on IBM PC compatible hardware, personal computers were not something one would expect to find in any given household. Yet, the numbering of the 5150 belies the idea that that selfsame machine was IBM’s first attempt at personal computing; the 5100 preceded it.
This story actually starts with a programming language called APL. The language was designed and developed by Kenneth E. Iverson in the 1960s and is named after Iverson’s book A Programming Language. APL was first made commercially available in 1966.
Sometime in late 1972, the General Systems Division of IBM located in Atlanta, Georgia, asked the IBM Scientific Center in Palo Alto, California, to develop a product that would raise the visibility of the programming language APL. Paul Friedl was working at IBM’s Palo Alto location and had been working on a project of his own. He made some sketches and general outlines of a machine that would be portable, personal, and fully integrated into a single unit. He presented this proposal to IBM Management, and Jack Rogers of IBM’s GSD in Atlanta commissioned the project to take this concept to a working prototype. This was the SCAMP (Special Computer APL Machine Portable) which fulfilled Friedl’s goals as well as those of the GSD. Rogers gave Friedl six months to get this done. Friedl then gathered a team of ten at IBM’s Los Gatos research center and got to work. Joe George was the lead hardware engineer, Patrick Smith was the lead software engineer, and Tom Hardy was responsible for the industrial design work. The machine made use of many off-the-shelf components due to the time constraints. For example, the CRT display was a Ball Brothers model, and the tape drive was a Norelco. The processor, keyboard, memory, and system software were all IBM.
The processor for the SCAMP was the PALM (Program All Logic in Microcode). The PALM was implemented on a circuit board and was made of thirteen bipolar gate arrays and three TTL DIP ICs. The processor had a sixteen bit data bus with two additional parity bits, and was capable of addressing 64K RAM. The microcode part of this made things particularly weird. The SCAMP team wrote an emulator of the IBM 1130 in microcode and ran APL on it. This saved them quite a bit of time as they’d otherwise have needed to write both an OS and an APL compiler.
The portability mentioned in Friedl’s proposal is on full display in the SCAMP with the screen being able to fold down, with a built-in cover to slide over it, and the whole machine existing within a suitcase style enclosure. As for the personal nature of the device, it was personal in the sense that it was designed to be used by a single person at any one time. The intended use case for this machine was shown through its menu system, and IBM had rather specific ideas. Specifically, it was thought to be used as a calculator, a tool for financial, statistical, or engineering analysis, for project planning, or for education. The machine was, however, programmable in APL and it had an on-line (meaning on the machine and displayed on the screen) HELP system. These last two aspects were what Friedl found most important. He had a vision of a machine that would be used primarily to run APL programs by users who weren’t the most technically inclined. To sell this vision, Friedl took the machine to various IBM managers and had them use the machine themselves. The Norelco was the first portable cassette recorder ever made commercially available, and it wasn’t very reliable. As a result every time Friedl wanted to do a demonstration of the machine, he and his team would get the machine setup and all of the software loaded the night before, and they’d leave the machine running until the demo the next day. Despite all of this, the managers were all able to use the machine and its software thanks to Friedl’s efforts at ensuring the machine was easy to use.
With the prototype working, with happy managers, and with a company that had nearly zero recent retail operations experience, ideas began to fly. The usual IBM playbook was to have professional installation, end-user training, rental agreements that included maintenance directly from IBM, and of course on-going support agreements for any problems that may arise. None of this would apply to a personal computer. Early thoughts were that the customer himself/herself would handle both maintenance and installation. It was also soon realized that the customer would also be doing the troubleshooting. IBM even considered opening retail stores of their own. The most obvious issue was that rental wouldn’t be an option. People would purchase the machine outright. This machine, the SCAMP, eventually became the IBM Model 5100 which was announced on the 9th of September in 1975.
The 5100 was about seventeen and a half inches wide, eight inches tall and about two feet deep. The portable description is rather generous as this thing weighed roughly fifty pounds. This is especially true were one to include the IBM 5103 printer which frequently accompanied this machine which also weighed in at around fifty pounds. The printer was about two feet wide, eight inches tall, and fourteen inches deep.
The production machine used the PALM processor just as the SCAMP did.
For the 5100, this processor was clocked at 1.9 MHz. While the SCAMP had APL in microcode, the 5100 included both APL and BASIC. In addition to the microcode, the IBM 5100 had over one hundred kilobytes of ROM, which then required the machine to do banking to address all of it. The 5100 maintained use of magnetic tape for data and program storage, but these were DC300 cartridges with a capacity of 200K rather than cassette tapes. This was a far more reliable setup than the Norelco while also being cheaper. Some discussion had been had about switching to floppies, but eight inch disks would have made the 5100 a non-portable machine. The 5100 added a numeric keypad, and the keyboard had special functions encoded for APL and BASIC syntax elements. The 5100 maintained the five inch CRT of its predecessor which was capable of sixty four characters wide by sixteen vertical rows. Breaking with the SCAMP, the 5100 didn’t have its carrying case integrated. The machine could fit into an optional carrying case, and that would add another five pounds to the overall weight. Another change was that while SCAMP emulated an 1130, the 5100 partially emulated a System/360. This provided APL/360 instead of APL/1130, but it’s also what provided compatibility with BASIC. For memory, the 5100 could be bought with 16K, 32K, 48K, or 64K RAM.
Pricing for this machine was both complicated and very high. A version with only APL/360 started at $8975 (about $51326 in 2023) with 16K RAM. Getting 32K raised this to $11975, 48K would take the price to $14975, and for 64K the price was $17975. BASIC would have increased those prices by an additional thousand, and getting both languages was another one thousand again. So, if a customer wished to have 64K RAM, BASIC, and APL, he/she would need to spend $19975 (around $114232 in 2023). The 5100 did have some capabilities for expansion. A second tape drive was available for $2300 (around $13153 in 2023) and connected to a proprietary port on the back of the unit, and the 5103 printer was available for $3675 (around $21016 in 2023). This used the same port as the tape drives, but these could be daisy chained (required a terminator) allowing multiple tape drives and a printer. The machine sported a daisy chainable BNC port for connection to video monitors (compatible with BNC to VGA adapters too), a dedicated communications port for use as a terminal, and it also had two serial ports.
With the high price, the IBM Model 5100 was primarily used in the Defense Department (as well as other government agencies), research laboratories, universities, medical offices, real estate offices, insurance companies, and similar places. A few were used at large cattle ranches and lumber yards. The portable nature saw a 5100 make its way onto an oceanographic research ship, a use case that no other computer at that time could so easily have filled.
The IBM 5100 wasn’t a huge commercial success. It was pulled from the market three years after its introduction when it was followed by the 5110. The 5110 was on the market for two years, and then replaced by the 5120 which wasn’t portable. Sales of the 5120 were ceased shortly after the introduction of the IBM PC Model 5150.
Your story of the 5100 and its pictures reminded me of the Datapoint 3300 (1970, San Antionio Tex). It’s interesting how IBM chose to put the screen on the left, as Datapoint did – why not on the right?
What a great series you’ve done (are doing), thank you.
Jon
Dr. Jon Peddie PE, President