Cool Retrotech

Here’s an interesting piece about the Apollo guidance computer, which played an important role in the moon-landing mission. The computer’s read-only memory, which stored the program and various constant data, was a “rope memory,” woven by women working at a factory near Boston. The pattern of the weave determined the “ones” and “zeros” of the permanantly-stored data. (via Isegoria)

Among the strange people who assert that the moon landing was a fake, one of the arguments used is that computers in 1969 lacked the computational capacity to guide such a mission. This ignores the fact that the guidance problem for intercontinental ballistic missiles is similar to that for space flight–do they also believe that the American and Soviet missile fleets were make-believe?

It is interesting, though, to compare the AGC with present-day computers. The AGC clock speed was about 2MHZ…around 500 to 1000 times slower than that of the computer on which you are probably reading this. The computer’s RAM was 2000 words, or 4000 bytes (that’s bytes, not kilobytes or megabytes) and the rope-memory ROM was 36KW, or 72KB.

And here’s a guy who built his own working replica of the AGC.

4 thoughts on “Cool Retrotech”

  1. Reminds me of the Minuteman missile computer. (showed up surplus around MIT sometime in the late 70s)
    It was built on two printed circuit boards, one was an entirely discreet CPU (made from actual transistors, resistors and capacitors, no integrated circuits), the other was 500 bytes of core memory, which stored the entire program and data. And this thing could fly to Moscow. It’s amazing how much could be done with so little. And the programmers who wrote software for these things must have been brilliant to fit it into so little space. No wasting thousands of lines of code just for the convenience of using Microsoft Basic or some other piece of bloatware.

  2. I don’t believe it’s quite true that the guidance for a ballistic missile is “similar” to that used in orbiting and landing on the moon.

    The very idea of “ballistic” is that the rocket takes a more-or-less fixed trajectory, just like a bullet, once aimed in the right direction. Guidance thereafter is minimal, especially considering the lack of atmosphere for most of the trip and the lack of ground communications when there is any atmosphere.

    The V-2, primitive as it was, was a pretty accurate ballistic missile that had a simple but clever mechanical guidance system.

  3. Jimbino…depends what you mean by “pretty accurate”…my understanding is that the circular error probable of the V-2 was about 17km. One of the early-1950s U.S. cruise missiles had a CEP (which had a range of thousands of miles vs a couple hundred for the V-2) had a CEP of around 20 *miles*. (There was originally a plan to use a vacuum-tube computer to guide this system but it wasn’t practical at the time and they had to revert to analog guidance)

  4. I actually had the chance to program on a simulation of some these early flight control computers. Very interesting nit-picking work.

    I think that there is complexity parity between the ballistic controllers and the apollo controllers. The ballistic controllers had to work much faster than the apollo controllers so they did more calculations per unit of time. The apollo controllers also had a much larger window to hit in space because for much of the mission, they could correct for deviations of hundreds of klicks with just a touch of thrust.

Comments are closed.