Apollo Guidance Computer revisited
The Apollo Guidance Computer was more capable than the myth would suggest. E&T pays tribute to the hard-wired hardware that flew Apollo 11 to the moon 40 years ago.
We often hear that the on-board Apollo Guidance Computer (AGC) had less power than a modern digital watch; but it all depends on what we mean by 'power'. In its way, it was one of the most capable computers ever invented. It absorbed data from a complex gyroscopic inertial navigation system, allied to an optical star telescope and two radar range finders. It also mediated between the astronauts and the thrusters and rocket engines that drove the ship through space. Apollo's Command Module (Columbia) and Lunar Module (Eagle) had an AGC each, but it's the latter's that has attracted the most attention over the 40 years since it took Neil Armstrong and Buzz Aldrin from lunar orbit to their landing on the Moon's surface.
Translating the AGC's capacity into modern computing parlance can be misleading, but its magnetic core ROM stored the entire guidance programme in the equivalent of about 36 kilobytes. Each tiny ring-shaped core acted as a miniature transformer. Signals from wires running through a core were interpreted as a '1' while those running past it registered a '0'. Up to 64 wires could be threaded in or around a core.
Nothing was 'stored' in ROM when the computer was switched off, but once activated, it booted up within less than a second. It was a totally 'hard-wired' system, because the software was encoded as patterns of wiring, snaking in and out of the little ring-shaped cores, that could not be overwritten or erased.
Raytheon, the company assigned to build the AGC, was located in a region of eastern Massachusetts that had been renowned since the 19th century for its textile industries. Raytheon made clever use of these local skills. The ROM's many thousands of wires were threaded through the cores by middle-aged women with nimble fingers and a painstaking ability to work from coded knitting patterns.
The computer processor itself depended on a relatively untried device: the integrated circuit (IC). The first examples were invented in 1958 by Jack Kilby of Texas Instruments. A year later, Robert Noyce of Fairchild Semiconductors (and later, founder of Intel) refined the process by putting all the components on a chip of silicon and connecting with copper lines.
It's a cliché that Apollo kick-started the microchip revolution, but there's some truth here. NASA bought up 60 per cent of America's entire output of ICs in the early 1960s, temporarily shoring up an industry for which few other markets yet existed.
The AGC's processor ran on just 5,600 ICs, all in the form of NOR logic gates. NASA insisted that only one gate design be used throughout, so that quality inspectors could check them more easily. This sounds limiting, but by stringing a sufficient number of NORs together in the right way, all the required logic functions could be carried out. The AGC's designers delivered a very capable computer, despite its modest 2K of RAM.
Home PCs of 2009 may be more 'powerful' than an AGC, but they tend to be plugged into just a few interfaces with the outside world: a printer, a backup disc drive, a screen, and a router. By contrast, an AGC was connected to a wide range of dynamic devices, from radio telemetry links with Mission Control, radar rendezvous systems and landing altimeters, gyro compasses and optical star trackers, and ultimately, to its propulsion systems.
The Apollo astronauts insisted that an AGC had to fit into their world, rather than the other way around. It needed to deliver readouts that they could immediately understand and trust, while accepting simple commands from them that they could learn by rote, or crib from prompt sheets.
Astronauts communicated with the AGC via a small display and keyboard, called DSKY. They inputted requests or commands in the form of short numeric codes, signifying programmes that they wished to initiate, or actions that they wanted the computer to perform. The DSKY then 'talked' back at them with five lines of numeric displays, and a small panel of 16 labelled lights.
DSKY revolutionised the relationship between people and computers. With our plasma screens and colour graphic interfaces, we take for granted our ability to communicate with computers without having to know about their inner workings. In the early 1960s, the idea of non-specialists having anything to do with them was quite radical.
After Apollo, the distinction blurred between the people who operate complex flight systems, and the systems themselves. Apollo treated the human as an embedded component within the feedback and control logic of the spacecraft as a whole; but this meant finding a practical 'man-machine interface.'
The AGC was the mediator of a strange and successful new union. David Mindell, Professor of the History of Engineering and Manufacturing at MIT, has studied the system in detail for his authoritative book, 'Digital Apollo'. "The Eagle's landing is one of the great technological mythologies of the 20th century," he says, "but there are elements that usually hide in the background: the interaction between human and machine, and the role of the computer in mediating the astronauts' responses."
A famous Apollo 11 incident proves Mindell's point. Just as the Lunar Module Eagle was on final approach for landing, its DSKY suddenly started flashing. "Program alarm," said Armstrong. "1202," confirmed Aldrin. Mission controllers decided that as the computer was continuing to function, the landing should proceed; but they sent out urgent messages to Raytheon and MIT, asking for an explanation. What was a 1202?
Suddenly, just as the touchdown should have occurred, the LM pitched forward, and shot violently across the lunar terrain at 60km/h. At last it settled down onto the surface, but by then, Mission Control was in a state of nervous exhaustion.
Armstrong was apologetic when he radioed to explain what had happened. "That may have seemed like a very long final phase, but the auto-targeting was taking us into a crater with a large number of big boulders and rocks." With his fuel running critically low, he had needed precious extra seconds to nudge the ship forward until he could find a safe place to land.
The media got excited about the 1202 alarms, and talked excitedly of Armstrong "switching off the computer" and "seizing manual control" to steer the lander away from hazards. In fact, there was never any conflict with the AGC, because the astronauts were always supposed to be able to choose the exact patch of ground they wanted to touch down on. That's why the LM had windows, after all. After the mission Armstrong defended his "God-given right to be wishy-washy about where I was going to land".
The landing site on the Sea of Tranquility had been chosen from detailed orbital scans, but it was only when the LM was hovering just a few tens of metres above the surface that hazards too small to be observed from orbit could make themselves apparent. Armstrong took evasive action, working in tandem with the computer - not against it - to bring the craft down safely.
Balancing the pencil
It would have been impossible to land without the computer. The LM was balanced on a plume of engine thrust from a single nozzle, like an upright pencil poised precariously on a fingertip.
A pistol-grip 'translation controller' enabled Armstrong to steer the descent engine's nozzle, which was pivoted on gimbals that were nudged by electromechanical actuators, but those were under the AGC's control. Armstrong could not have maintained the balance of an unstable vehicle without having his steering commands refined, at one-tenth second intervals, by the computer.
As for the DSKY alarms: prior to the mission, a last-minute decision had been made to keep one of the LM's radars locked onto the orbiting mothership throughout the landing phase, while the other dish pointed down to the lunar surface. Given the urgency of the landing, the computer decided to send non-essential jobs, such as the mothership range data calculations, to the back of the line. It flashed a 1202 code to warn that it was running at near capacity. But it did not malfunction.
No wonder the AGC's designers at MIT and Raytheon felt that their achievements were too little understood by the media; and in their old age, how weary they must have become of hearing that you could today navigate to the Moon with a child's toy.
David Mindell suggests that Apollo "entangled humans and machines, defining new forms of heroic action in a technological world". That heroism was demonstrated by the designers, as well as by the men and equipment of Apollo itself.