Mercury-Atlas 3

By the time MA-3 reached the pad, Project Mercury had solved half of its problem and still owed the other half. The capsule had come through the Little Joe abort flights, the beach abort test and Mercury-Atlas 2, which in February 1961 had demonstrated the integrity of the structure, the ablation shield and the afterbody shingles during a reentry from a critical abort. What remained unproven was the launch vehicle. The Atlas was an intercontinental ballistic missile being turned into a crew carrier, and that conversion reached into nearly every one of its systems: a missile built to throw a warhead or an uncrewed satellite carried nothing designed to keep a passenger alive.

Mercury-Atlas 1 had been lost during ascent in July 1960, and the post-flight analysis had forced a redesign of the liquid-oxygen tank skin. Meanwhile the programme's commitment stood unchanged: put an astronaut into orbit. MA-3 was the rehearsal before that flight, with the spacecraft fitted out as though it carried an occupant and a mechanical simulator taking his place. NASA framed the objective plainly as a quick test of the spacecraft-Atlas combination in orbit.

The capsule assigned to MA-3 was Mercury spacecraft No. 8, delivered to the launch site on 18 November 1960. It spent twenty-three weeks in preparation at the Cape before flying, a long stay even by the standards of those early missions, when each spacecraft arrived with systems still evolving and left the pad trailing a list of requested alterations. It carried the mechanical crewman simulator — the "mechanical astronaut" — along with the normal equipment of an orbital mission. There was no animal aboard, unlike MR-2 with Ham or the later MA-5 with Enos.

The launcher originally planned for MA-3 was Atlas 77D, which passed its factory rollout inspection in September 1960. Shortly afterwards the post-flight findings from MA-1 came out, and the thin-skinned 77D was recalled and replaced by 100D. That substitution made 100-D the first thick-skinned booster delivered to the programme: on Big Joe and MA-1 only the fuel tank had used heavier gauge skin while the liquid-oxygen tank kept the standard missile thickness, and after the loss of MA-1 NASA concluded that the standard skin was not enough. MA-2's booster, 67D, was still a thin-skinned vehicle and had to be fitted with a steel reinforcement band at the interface between capsule and booster; 100-D no longer needed it.

The vehicle also introduced a transistorised telemetry box in place of the vacuum-tube equipment used on earlier Atlas flights, which was bulky, drew a great deal of power and whose signal strength tended to degrade during launch. The replacement had been designed for the Atlas-Centaur and was adapted quickly for Mercury.

MA-3 was the first Mercury-Atlas to fly with a live escape tower. Big Joe and MA-1 had flown without a tower — and in MA-1's case the absence was suspected of having spoiled the aerodynamic profile of the stack — while MA-2 carried a dummy tower. The launch escape system itself was identical on Redstone and Atlas flights; what differed was the abort sensing system, far more complex on the Atlas because it was a much larger vehicle with five engines, two of them jettisoned in flight, a more sophisticated guidance system and pressurised balloon tanks that collapsed if they lost pressure.

That system, the Abort Sensing and Implementation System (ASIS), watched a deliberately short list of parameters drawn from the most likely failure modes of the D-series vehicles: excessive deviation from the planned trajectory, engine thrust or hydraulic pressure falling below limits, loss of propellant tank pressure, signs that the intermediate bulkhead was losing structural integrity, failure of the booster's electrical system, or the failure of ASIS itself. Redundancy was built in by design: if ASIS died, the loss of power alone triggered an abort. It had flown on missile research flights and then, on MA-1, open loop, able to raise an abort signal but not to command engine cutoff. MA-3 was the first flight on which it operated closed-loop.

Liftoff came at 11:15 a.m. Eastern time from Launch Complex 14 at Cape Canaveral Air Force Station. The climb was normal until about twenty seconds into flight, the point at which the pitch-and-roll sequence should have begun. That manoeuvre turned the initial vertical trajectory into a flatter one carrying the Atlas out over the Atlantic. It never happened: the rocket simply kept going straight up.

In the control room the flight dynamics officer, Tec Roberts, reported that there was no roll-and-pitch programme, and the meaning was instantly clear to everyone present, as flight director Gene Kranz later recalled: an Atlas climbing vertically threatened the Cape and the surrounding communities, and the higher it flew before breaking apart, the wider the debris footprint would be. The range safety officer confirmed the missing manoeuvre, lifted the cover from the destruct button and gave the vehicle a few seconds to recover and start its track across the ocean. It did not. The destruct command went out as the Atlas converged on the limits of the safety plot; radar then reported tracking multiple targets, the unmistakable sign that the booster had come apart.

The escape tower fired at the very moment the destruct command was sent and lifted spacecraft No. 8 away from the vehicle. The capsule climbed to an apogee of 7.2 kilometres and landed only 1.8 kilometres downrange; the whole flight lasted 7 minutes and 19 seconds, nearly all of it descending under the parachute. A United States Marine Corps CH-37 helicopter picked it out of the Atlantic about twenty minutes after liftoff.

For the Mercury team that was the single consolation of the day: the launcher's failure had turned into an unplanned test of the escape system, and the system worked. The mission itself was recorded as a failure, and the launch site summary put it bluntly — launched and aborted, mission objectives not accomplished.

On 17 May 1961 an investigation board was convened to establish the cause of the MA-3 launch vehicle failure. It considered several areas and, from a review of the test data, isolated three as probable causes. Telemetry analysis had quickly placed the fault somewhere in the guidance system without pinning down its exact nature: the evidence suggested that the missile programmer had either shut down completely shortly after liftoff or suffered a power interruption, restarted, and then failed to execute the pitch-and-roll sequence.

Two months after the flight the Atlas programmer was found buried in mud on a beach not far from the pad and was examined. The examination did not settle the cause with certainty — contamination of pins inside the programmer and a transient voltage that reset it were both proposed — but it exposed something more uncomfortable: the unit as a whole had several serious design flaws that needed correcting. The practical consequence was that the uncrewed orbital qualification flight slipped by several months, to September.

Spacecraft No. 8, recovered from the sea, was refitted and returned to the launch site on 11 May 1961 as spacecraft 8A, by then assigned to Mercury-Atlas 4. After eighteen weeks of preparation it flew on 13 September 1961 in the same configuration, this time atop Atlas 88-D. MA-4 carried special vibration and noise instrumentation and a mechanical crewman simulator in addition to the normal spacecraft equipment, and it became the first Mercury spacecraft to reach Earth orbit, with an apogee of 123 nautical miles and a perigee of 86 according to the programme chronology. After one orbit the spacecraft's orbital timing device fired the retrorockets and the capsule splashed down. The launch site summary records it as a spacecraft refitted and flown in the same configuration, mission successful.

That the very capsule torn away from a disintegrating Atlas should reach orbit five months later says a great deal about how the programme worked: hardware was recovered, repaired and flown again, and every failure was converted into a list of corrections. MA-3 met none of its objectives, but it left the programme three things: proof that the escape tower worked under real conditions, a catalogue of defects in the Atlas programmer, and an intact capsule with which to fly the mission MA-3 could not.