Estimated reading time: 2 minutes

Beyond Design: Losing a Bit of Memory

In the early 1980s, when I was working in the Microprocessor Research Lab at the University of Western Australia, we had the opportunity to dismantle an old DEC mainframe that had served its useful purpose. The main reason was to salvage any useable components. Whilst rummaging through the scrap, we found a number of blocks of magnetic core memory. I extracted one of the cards from a block and found it was made up of thousands of small toroidal magnets that seemed to have three enameled copper wires running through each toroid. This was just one kilobyte of memory. I’m not sure if this is fact, but at the time, I was told that the women of the Fijian Islands sewed the threads of wire, which were used to weave fine fishing nets. Even with my excellent eyesight (at the time), it was difficult to see.

This magnetic core memory was the first random access memory technology. The three wires were X, Y, and a sense/inhibit that ran diagonally through all cores. These wires were used to create the magnetic charge, remove the charge, and to read the state of the charge. And since the sense/inhibit wire ran through the entire matrix, only a single bit could be read at one time. Also, since the process of reading erased the data it held, it was automatically restored after read (write-after-read cycle). It is hard to believe, but core memory was the memory of choice for about 20 years, until semiconductors took over.

Core memory was extremely durable compared to today’s technology, as it had no moving parts, was non-volatile and resistant to heat and electromagnetic interference. It was used extensively in the early space shuttles and survived not only the explosion but also the plunge to Earth when the Space Shuttle Challenger exploded shortly after launch in 1986. NASA was able to recover and read the contents of the memory. The shuttle’s IBM AP-101 computers originally had a massive 424 kilobytes of magnetic core memory and used a high-level language called HAL/S (what else?).Read the full column here.Editor's Note: This column originally appeared in the May 2013 issue of The PCB Design Magazine.