Ch5-4

Computers in Spaceflight: The NASA Experience

- Chapter Five -
- From Sequencers to Computers: Exploring the Moon and the Inner Planets -

Expanded memory and expanded functions


[146] The new sequencer had a 9-bit address field, providing a 512 address limit. Expanding the memory to 512 words did not require a change in the logic. So JPL added the extra memory for the Mariner [147] Mars 1971 orbiter missions. Still, the old fixed sequencer remained in charge of the Mars orbit insertion burn. After the spacecraft established orbits, however, the ground control center used the new sequencer to control the imaging of Mars and its moons. The expanded memory proved sufficient. Preflight estimates for Mariner VIII specified 150 words of memory and 225 words for Mariner IX, yet both grew to over 400 words in flight³⁰.

The mission that used the sequencer to its limits was Mariner Venus Mercury 1973, or Mariner X. Mission profile called for the spacecraft to turn its imaging equipment on the earth as it flew toward deep space, do some studies of the moon in flyby, and then research in the area of Venus during a gravity assist maneuver that would send it toward Mercury, where JPL planned three separate encounters with the innermost planet.


Figure 5-2. Mariner Venus/Mercury 1973 made the most use of the programmable sequencers. (JPL photo 251-135AC)

[148] Due to the more complex mission requirements, the design team wanted a bigger and better sequencer, but cost constraints killed any chance of building a new machine³¹. Adrian Hooke of JPL, one of the project's managers, decided to use planned memory updates at regular intervals. He also instituted a "suspenders and belt" approach to reliability. The sequencer would not only carry a detailed program for the next mission phase but also a constantly updated bare minimum program to complete the mission if the spacecraft lost contact with the ground. If a command was not received for a certain time, then the sequencer would follow whatever commands were in the backup program. Thus, software moved ahead in leap frog fashion. During the earth-moon phase the Venus backup was loaded, during the Venus encounter the backup Mercury encounter sequence was on board, and so on³². Software development was assigned to three programmers. Ronald Spriestersbach of JPL wrote the near-earth and post-Mercury sequences, George Elliot of the Boeing Company did the Venus encounter, and Larry Koga of JPL wrote all three Mercury encounters³³.

During the 1969 missions, most changes and subroutines were hand-coded and used once. By 1971, the COMGEN ground computer program that produced memory loads for the Sequencer could develop blocks of commands that functioned much like subroutines in a standard computer program or macros in an assembly language program³⁴ . In 1973, COMGEN resided in an IBM 360/75 computer that generated the commands and sent them via the NASA communications net to the appropriate Deep Space Network station for transmission. By this time, each station had a command computer, thus ending the voice/manual era³⁵. Another improvement to the Sequencer was that engineers could do memory checks by comparing a sumword stored in location 512 to the result of summing the first 511 locations. If a miscompare occurred, then a location-by-location check for error could be made³⁶.

The improvements both in the Sequencer and in programming and ground control techniques were not enough to ensure its use beyond the Mariner series of spacecraft. In spite of the success of the long and complicated mission of Mariner X, JPL's Hooke complained that memory limits were too costly due to excessive need for optimization and constant relocation of subroutines³⁷. Besides, the sequencers, regardless of their full name, were not computers. Spacecraft needed to do on-board computations, to have more room for software (and, thus, increased flexibility), and to use the central computer for other functions such as spacecraft health and safety monitoring done on other manned and unmanned spacecraft. Some missions intrinsically needed computers, as, for example, the Viking Mars orbiters and landers and the Voyager outer planet probes. The computer eventually designed, built, and used for the Viking Orbiter had its roots in the programmable sequencer, but it also owed some [149] concepts, at least in comparison, to a computer built in the research side of JPL and aimed at the long-duration, complex missions of the future. The story of that computer research project adds a necessary perspective for understanding the direction JPL's on-board computer development took in the 1970s.