|
Post by Grand Lunar on May 7, 2008 19:02:09 GMT -4
While learning about electronic fuel injection (EFI) engines in my marine mechanics class, I noted interesting things about the comptuters used on these engines.
It seemed similar to Apollo, far as I know. The computer takes in data from various sensors, and has devices that can be used to make changes to the spacecraft, just as an EFI engine's computer (ECM) can make changes to the engine's conditions.
What more is how little memory is used. The first generation used a mere 500+ bytes of memory. The latest type uses 2K of memory. Last I read, Apollo's computer used 4K.
What more, is that by the use of scan tools and special software for PCs and laptops, the data by the computer can be displayed. It seemed similar to how the computers at mission control took in the data from Apollo's computer and displayed it on those large screens.
What I'm wondering is if these technologies are based on similar concepts. Are these computers from the same "family", but just of different generations?
|
|
|
Post by JayUtah on May 7, 2008 21:12:00 GMT -4
Well, all computers work by listening to inputs and generating outputs according to some appropriate algorithm.
But the branch of computation most closely associated with both Apollo and fuel-injection computers is control theory.
Fasten your seat belt.
Discrete controllers use combinatorial logic to map discrete sensor inputs to discrete output control signals. They express process actions such as, "When the fuel level falls below value X, turn on a warning light on the console," or "When the camshaft position sensor A is activated and the ignition bus voltage exceeds a minimum value, spray the injector." Timers and similar pseudo-inputs add sequentiality.
Discrete controllers can implement fairly complex finite state automata, a formal method of modeling computation. They tend to be rugged and simple, therefore highly reliable. As such they can implement fail-safes. One we use in the theater is a hydraulic position fail-safe: we hold certain hydraulically-powered components in place with brakes rather than hydraulic pressure when they have to be there for a long time. This lets us power down the pump and compensate for outlet valve leakage. Upon the command to move the component, the brake will not release until the hydraulic pressure in the actuator rises to a value sufficient to hold it in place without the brake.
Continuous controls take inputs from continuous process variables and compares them against a stored set point. The difference between the set point and the process value is the error. The error determines (by mathematical function) the value of a corresponding control variable. The most common implementation of a continuous control system combines the error function with functions that use the time-integrated error and the time-derivative error. You tune the system by controlling how much each function determines the control output.
The AGC digital autopilot was a kind of proportional-derivative continuous controller.
Both these types of controller can be implemented in terms of electronic components. Discrete controllers, for example, can be implemented by N-way relays: elevator controllers were built this way for many years. Continuous controllers can be built using analog electronic circuits.
However, these days both types are usually implemented with general-purpose microcontroller CPUs. The application-specific controller logic is built in software with special tools.
The Apollo computers and their various input and output controllers and related modules were designed and built very much along the lines of classic control-system methods.
|
|
|
Post by Grand Lunar on May 10, 2008 7:27:34 GMT -4
Thanks a bunch, Jay.
|
|