The AGC (for Apollo Guidance Computer) was the computer used in the command and lunar modules of the missions that, about 40 years ago, took men to the Moon. Even though integrated circuits already existed at the time, they were considered too modern for NASA to trust their reliability, so they were set aside. That is why the AGCs ended up using ROM memories “woven” with thin wires and magnetic cores, devices known as core rope memory. Hand-woven by anonymous textile workers from Massachusetts, they contributed to the success of the first space missions.
Core rope memory: the ROM of the Apollo
Despite the fact that some still debating whether man truly reached the Moon or not, part of the technology we use today was driven by the development of the Apollo missions. We all know the difficulties of sending an astronaut to the Moon, landing, then taking off and returning to our planet. In the 1960s this was even more complex, because one of the pillars on which any mission of this type must rely—computing—was still in its infancy. Guiding a spacecraft to our natural satellite and making it land softly on its surface is beyond human capabilities. Delaying the ignition or shutdown of an engine by tenths of a second can mean the difference between mission success and the most resounding and horrible failure. However, a computer, even without great computing power, can handle these tasks without problems. NASA knew this and decided to equip its spacecraft with an onboard computer. That is how the AGC (for Apollo Guidance Computer) was born.
Obviously, thinking of a computer 40 or 50 years ago meant thinking of something the size of an apartment, heavy as a dozen elephants, and powered by thousands of watts of energy. Something like that could not be included in the modules NASA was designing. The U.S. government had some smaller (and more limited) computers designed to guide the Polaris ballistic missiles, but their power was far below what NASA required. The Apollo computer would have to handle the navigation and autonomous guidance tasks of the craft, something like a “pilot” practically independent of the astronauts. The company North American had been designated by NASA to develop the command module (CSM) of the Apollo, and its managers turned to the Massachusetts Institute of Technology (MIT), where the leading computer experts of the time were, seeking proposals and solutions. After several meetings, they concluded that a computer like the one they needed could be built, with a volume of about one cubic foot—that is, roughly the size of a cube about 30 centimeters on each side.
Although the then-novel technology of integrated circuits was available, and NASA bought nearly a million chips at the time for various projects, the truth is that when it came to equipping the AGC with ROM memory, they did not trust silicon. Instead of using chips, the AGC programs were physically “wired”, a system similar to the ferrite core memories used in many computers of the era, which was well known. The core rope memory of the AGC had a capacity equivalent to about 36KB of 16-bit words (the RAM was only 2KB), where each bit was encoded (or “wired”) with copper wires “woven” around small toroids made of a steel-nickel alloy. In these memories, the ferromagnetic cores were used as small transformers, where the signal of the “word” passing through a given core and paired with a particular bit line was interpreted as a binary “one”. If the corresponding word line passed outside the core, it was read as a binary “zero”.
The system works, and it is extremely robust. In fact, during the launch of Apollo XII, several lightning strikes hit the Saturn V rocket, causing the onboard computer to “hang”. But once in orbit, the astronauts reset it and it started without issues. Very few modern computers could survive such an event. The company Raytheon had hired several women, some former textile company employees, to wire these memories. They would take the binary data and “sew” them onto the metal cores. NASA designers called these ROMs “LOL memory”, for “Little Old Ladies memory”, something like “memories of the little old ladies”, in reference to the women who assembled them. Each computer had six LOL memory modules, and each one required more than a month of work to be ready. Then they were verified to ensure that the programs and orbital parameters included were correct, and they were ready to be installed in the AGCs.
NASA of course used chips and transistors in other components of these computers, but the memories were based on this older technology, the only one—according to its engineers—that offered the level of safety the mission required. Some claim that NASA and the Apollo missions are responsible for driving the miniaturization of computers, which went from being monsters the size of a room to small boxes 30 centimeters on a side. Surely this is exaggerated, and it is likely that military applications of computers did more to drive the reduction in size. In any case, the effort of the MIT engineers made it possible to create computers small enough to take man to the Moon, and that is no small feat.
For more information, see Core rope memory on Wikipedia.
