Spaceborne computers are usually designed to satisfy rather severe physical constraints. These are discussed below.
Weight is related to size by the densities of materials used and the spatial efficiency of the structure. In the past, good packaging practice has led to densities in the neighborhood of the density of water (one gram per cubic centimeter). Early aerospace computers with 8K memories had masses on the order of 40 kg (88 lb), while equivalent present day machines have masses on the order of 20 kg (44 lb) or less.
Memory size, memory speed and processor operating speed are important factors in size since faster circuits, high-speed memories, and parallel operations all tend to lead to increased size and power dissipation. Low speed requirements may permit more efficient, serial or series-parallel processing and machine organization. Extremely small processors of the order of a few hundred cubic centimeters in volume, limited in throughput capability, can be made using large scale integrated-circuit packaging techniques.
Size reductions foreseen by the progress of the art of microminiaturization have been limited by problems in the areas of heat transfer, interconnections including the availability of test points), ability to replace subsections, and power conversion. Consequently, the greatest overall economy is not necessarily achieved by designing to the minimum possible volume but rather requires tradeoffs in volume vs operating features.
(3) Power Consumption
Except for certain special low power logic systems, the advent of integrated circuitry, MSI and LSI, has not produced dramatic reductions in total power consumption. However, the power efficiency has been substantially improved since more computation can be performed for a given amount of power. Low power consumption may enhance reliability since, for instance, some semiconductor failure mechanisms are strongly temperature dependent. Since the memory often consumes the bulk of the power, it is a major factor in power requirements.
Another important consideration is speed since faster circuits nearly always require larger currents, and, in particular, faster core or film memories normally require larger drive currents. Therefore, a faster computer with otherwise the same operating characteristics generally requires more power. Occasionally, the power distribution to spaceborne computers is controlled during different phases of the flight. For example, the Apollo lunar module computers are powered down during most of the flight to the moon. For these applications, fixed memories or nonvolatile NDRO memories are usually desirable since they retain their contents when the power is turned off. The computer may not be switched off entirely in some cases but "idled" at low standby power; periodically, it will awaken itself to perform tasks such as midcourse-guidance calculations. A few special circuits have been developed which consume very little or no power in the standby condition. The newer circuit technologies such as low power bipolar logic and metal-oxide-semiconductor-field-effect-transistors (MOSFETs) - particularly complementary MOSFETs - consume less power than the bipolar circuitry which is most commonly used in spaceborne computers today.
In the interest of reliability, power supplies are often duplicated, as in the Saturn LVDC, and provision is made for switching in spare power supplies when those in use fail during a mission. Also, protective circuitry is often included to protect against power surges caused by certain types of power supply failures. Computers with a restart capability incorporate an alarm to warn of power supply failure and a temporary power source to allow retention of system status. However, the advantages of these protective features are gained at the expense of increased weight and volume as well as the added complexity.
Home - NASA
Office of Logic Design
Last Revised: February 03, 2010
Digital Engineering Institute
Web Grunt: Richard Katz