8 inches FLOPPY

In 1967 IBM gave their San Jose, California storage development center a new task: develop a simple and inexpensive system for loading microcode into their System/370 mainframes. The 370s were the first IBM machines to use semiconductor memory, and whenever the power was turned off the microcode had to be reloaded ('magnetic core' memory, used in the 370s' predecessors, the System/360 line, did not lose its contents when powered down). Normally this task would be left to various tape drives which almost all 370 systems included, but tapes were large and slow. IBM wanted something faster and more purpose-built that could also be used to send out updates to customers for $5. David Noble, working under the direction of Alan Shugart, tried a number of existing solutions to see if he could develop a new-style tape for the purpose, but eventually gave up and started over. The result was a read-only, 8-inch (20 cm) floppy they called the "memory disk", holding 80 kilobytes. The original versions were simply the disk itself, but dirt became a serious problem and they enclosed it in a plastic envelope lined with fabric that would pick up the dirt. The new device, developed under the code name Minnow, became a standard part of the 370 in 1969. A Japanese inventor, Yoshiro Nakamatsu (aka Dr. NakaMats), claims he independently came up with the floppy disk principle back in 1950, and so a sales license had to be acquired by IBM when they started manufacturing their floppy disk systems. Alan Shugart left IBM, moved to Memorex where his team in 1972 shipped the Memorex 650, the first read-write floppy disk drive. In 1973 IBM released a new version of the floppy, this time on the 3740 Data Entry System. The new system used a different recording format that stored up to 250¼ kB on the same disks, and was read-write. These drives became common, and soon were being used to move smaller amounts of data around, almost completely replacing magnetic tapes. The IBM standard soft-sectored disk format was designed to hold just as much data as one box of punch cards. The disk was divided into 77 tracks of 26 sectors, each holding 128 bytes. Note that 77×26 = 2002 sectors, whereas a box of punch cards held 2000 cards. When the first microcomputers were being developed in the 1970s, the 8-inch floppy found a place on them as one of the few "high speed, mass storage" devices that were even remotely affordable to the target market (individuals and small businesses). The first microcomputer operating system, CP/M, originally shipped on 8-inch disks. However, the drives were still expensive, typically costing more than the computer they were attached to in early days, so most machines of the era used cassette tape instead. This began to change with the acceptance of the first standard for the floppy disk, ECMA-59, authored by Jim O'Reilly of Burroughs, Helmuth Hack of BASF and others. O'Reilly set a record for maneuvering this document through ECMA's approval process, with the standard sub-committee being formed in one meeting of ECMA and approval of a draft standard in the next meeting three months later. This standard later formed the basis for the ANSI standard too. Standardization brought together a variety of competitors to make media to a single interchangeable standard, and allowed rapid quality and cost improvement. Shugart moved on in 1973 to found Shugart Associates. They started working on improvements to the existing 8-inch format, eventually creating a new 800 kB system. However, profits were hard to find, and in 1974 he was forced out of his own company. Burroughs Corporation, meanwhile, was developing a high-performance dual-sided 8-inch drive at their Glenrothes, Scotland factory. With a capacity of 1 MB (MiB), this unit exceeded IBM's drive capacity by 4 times, and was able to provide enough space to run all the software and store data on the new Burrough's B80 data entry system, which incidentally had the first VLSI disk controller in the industry. The dual-sided 1MB floppy entered production in 1975, but was plagued by an industry problem, poor media quality. There were few tools available to test media for 'bit-shift' on the inner tracks, which made for high error rates, and the result was a substantial investment by Burroughs in a media tester design that they then gave to media makers as a quality control tool, leading to a vast improvement in yields.