Mariner 8

The Mariner Mars 1971 project was designed around two spacecraft doing two different jobs, because one orbit cannot do both. To map a planet you want a steep orbit that carries you over high latitudes as the planet turns beneath, so the ground track eventually covers everything; to watch a planet change you want a lower-inclination orbit whose period is tuned to the Martian day, so you come back to the same places under the same lighting again and again. Mariner-H, the spacecraft that became Mariner 8, was assigned the mapping job: 80 degrees inclination, a twelve-hour period and a closest approach of 750 miles, from which it would photograph roughly 70 percent of the surface over at least ninety days. Mariner-I, which became Mariner 9, drew the other job at 50 degrees, a 20.5 hour period and a 550 mile periapsis, revisiting six chosen areas to record seasonal and dust changes. The hardware was identical and either spacecraft could have flown either mission; only the flight plan separated them. Both were also new in a specific way. Every previous Mariner had flown past its target. These carried a restartable 1,340 newton engine and enough propellant for a 1,600 metre per second braking burn, an entirely redesigned propulsion system, because for the first time an American spacecraft was going to stop at Mars.

Mariner 8 lifted off from Launch Complex 36A at Cape Kennedy on 9 May 1971 at 01:11:02.294 GMT, which was ten past nine on the evening of 8 May in Florida. The Atlas performed. The trouble began at Centaur main engine start, and it began invisibly: the gain in the pitch rate channel, the loop that tells the engines how hard to push back against an unwanted rotation, came up at only 20 to 40 percent of what it should have been. A rate gyro measures how fast the vehicle is turning, and its signal is what damps a steering command so the rocket settles onto a new attitude instead of overshooting it. With that damping weak, the vehicle was stable only while it was being asked to do nothing. Four seconds after ignition, when the guidance system started steering, the pitch motion began to diverge. Twenty-four seconds after that the engines went to fixed positions and the stack started to tumble. Roll and yaw stayed normal the whole time, which is exactly the signature of a single failed channel. Eighty-six seconds after ignition the payload telemetry cut out as the spacecraft came off the stage, and two seconds later the tumbling had thrown the propellants away from the pump inlets and the engines starved. The wreckage hit the Atlantic about ten minutes after launch, 1,665 kilometres downrange and some 560 kilometres northwest of Puerto Rico.

The investigation was fast and unusually conclusive. A first-look review ran the night of the launch, with NASA Lewis, JPL, Kennedy, General Dynamics, Minneapolis-Honeywell and Pratt and Whitney in the room, and it put the fault in the pitch channel of the Centaur flight control system before morning. General Dynamics then built a computer simulation of that flight control system, worked through root-locus analysis and analogue simulation, and finally installed real hardware in the loop. A prototype gyro canister with a deliberately failed part reproduced the flight records. The single cause was a uA709 integrated circuit in the pitch rate gyro preamplifier. JPL's public statement on 26 May put it slightly differently, saying the integrated circuit failed apparently because a protective diode had malfunctioned, and both descriptions survive in later accounts, which is why some sources blame a diode and others a chip. The fix was a partial redesign of the rate gyro package plus much harsher acceptance testing: temperature cycling for dc offset, elevated temperature and vibration, deliberate attempts to provoke dc latch-up, and explicit checks of the amplifier input diodes. Engineers also looked hard at flying a replacement. The proof test model spacecraft could have been brought to flight standard, a Centaur and Atlas could have been found, and a nose fairing could have been modified in time. Everyone agreed it was possible and that it needed money. NASA said no.

What was left was one spacecraft and two missions, so JPL merged them. Mariner 9 flew on 30 May into a compromise orbit at 65 degrees inclination with an 11.98 hour period, a number chosen because it is 17/35 of a Martian solar day, so that after seventeen Martian days and thirty-five orbits the ground track repeats under nearly identical lighting. That single choice preserved the variable-features work, expanded from six target areas to seventeen, while the inclination kept enough polar reach for mapping. It worked better than the original plan deserved. Mariner 9 arrived on 14 November 1971 GMT to find Mars buried under a global dust storm, waited it out, and then returned a near-complete map of the planet plus the discovery of Olympus Mons and Valles Marineris, doing in one mission what two had been budgeted for. Mariner 8's own name is a small historical oddity: it flew as Mariner-H, never reached Earth orbit, and was still given the number 8, a point JPL's final report felt the need to footnote. NASA's history office describes it as the last American planetary spacecraft lost to a launch vehicle failure, a claim that holds if you count CONTOUR's 2002 destruction as a spacecraft failure rather than a launch vehicle one, since its Delta II had already delivered it to orbit.