View from Neil Armstrong’s position in the lunar module seconds before the touchdown of Apollo 11. This image was made by Industrial Light and Magic visual effects supervisor John Knoll. The “1202” error from the AGC — indicating an overflow condition — can be seen to the lower right. Courtesy of John Knoll

Companion to the Stars: The Apollo Guidance Computer

By Dag Spicer, Senior Curator

In August 1961, NASA contracted with the Charles Stark Draper Lab at MIT, which had already designed digital computers for space with its 26-pound, 0.4-cubic-foot on-board guidance computer for the United States Navy’s Polaris missile.

Left: View from Neil Armstrong’s position in the lunar module seconds before the touchdown of Apollo 11. This image was made by Industrial Light and Magic visual effects supervisor John Knoll. The “1202” error from the AGC — indicating an overflow condition — can be seen to the lower right. Courtesy of John Knoll

The Apollo Guidance Computer (AGC) project was radically ambitious. According to hardware design team lead Eldon Hall in a 1982 lecture at the Boston Computer Museum:

If the designers had known then what they learned later, or had a complete set of specifications been available as might be expected in today’s environment, they would probably have concluded that there was no solution with the technology of the early sixties.¹

The main contractor for the AGC was defense contractor Raytheon at their Sudbury, Massachusetts, division. When initial research and planning began in late 1959, the Draper Lab began experimenting with ICs — a new technology that combined multiple transistors into a single package. ICs provided enhanced reliability, much lower power consumption, and took up less space.

Another technology embedded in the AGC was rope memory, essentially core memory with the windings hard-wiring a zero or a one in each bit location. The information stored in rope memory could be changed, albeit at great cost and effort. To wire rope memory involved painstakingly threading wires the width of a human hair through the tiny magnetic cores of the memory.

Meetings of the guidance and navigation software team, led by 2017 CHM Fellow Margaret Hamilton, were often called “Black Fridays” because the bar for accepting any changes to the Apollo code was very high, resulting in many tense meetings and code reviews.

Margaret Hamilton , mathematician and programmer at the MIT Instrumentation Laboratory, sits in a mock-up of an Apollo command module, November 25, 1969. Hamilton led the team developing flight software for the AGC. Hamilton/Courtesy of MIT Museum; AGC/Collection of the Computer History Museum, X37.81A

From the user perspective, the AGC was designed to be simple. Nonetheless, in a typical Apollo mission, astronauts had to enter 10,600 keystrokes into the
AGC. This was made simpler through a special interface called a Display Keyboard, or DSKY, in which astronauts could communicate with the computer by using a “Verb + Noun” syntax.

While the AGC wasn’t particularly fast even compared to the minicomputers of its day, it was more reliable than commercial computers, much smaller given its performance, and it sipped a measly 70 watts of power. Not bad for putting a man on the Moon.

Notes

1. 1 The Apollo Guidance Computer,” part one, by Eldon Hall, June 10, 1982, Computer History Museum, 102624617.

About the Author

Dag Spicer leads the Museum’s collection strategy and supports multiple projects and initiatives across the institution, including those in research, education, fundraising, public programs, and marketing. He is the longest-serving employee at the Museum, having started in 1996, and holds degrees in history, electrical engineering, and the history of science and technology.

“Companion to the Stars: The Apollo Guidance Computer” is published in the Computer History Museum’s 2019 issue of Core magazine.