Abort sensor

Central to these efforts was the development of the Abort Sensing and Implementation System (ASIS), which would detect malfunctions in the Atlas's various components and trigger a launch abort if necessary. Added redundancy was built in; if ASIS itself failed, the loss of power would also trigger an abort. The ASIS system was first carried on a few Atlas missile R&D flights, then flown open loop on Mercury-Atlas 1, meaning the ASIS could generate an abort signal but not send a cutoff command to the propulsion system. It was operated closed-loop on MA-3 for the first time.

The Mercury launch escape system (LES) used on Redstone and Atlas launches was identical, but the ASIS system varied considerably between the two boosters as Atlas was a much larger, more complex vehicle with five engines, two of which were jettisoned during flight, a more sophisticated guidance system, and inflated balloon tanks that required constant pressure to not collapse.

Big Joe and MA-1 had no escape tower, the latter's in-flight failure was possibly due to the lack of the LES negatively affecting its aerodynamic profile and so MA-2 carried a dummy tower. A live LES was carried for the first time on MA-3 (and ended up proving its functionality in an unplanned test).

Atlas flight test data was used to draw up a list of the most likely failure modes for the D-series vehicles, however simplicity reasons dictated that only a limited number of booster parameters could be monitored. An abort could be triggered by the following conditions, all of which could be indicative of a catastrophic failure:

• The booster flight path deviated too far from the planned trajectory
• Engine thrust or hydraulic pressure dropped below a certain level
• Propellant tank pressure dropped below a certain level
• The intermediate tank bulkhead showed signs of losing structural integrity
• The booster electrical system ceased operating
• The ASIS system ceased operating

The ASIS system was deemed necessary because some flight failures of Atlas vehicles (for instance, Atlas 6B) occurred so fast that it would be nearly impossible for the astronaut to react in time to manually activate the LES. Other failure modes such as a deviation from the correct flight trajectory did not necessarily pose an immediately danger to the astronaut's safety, and the flight could be aborted manually.

Not all of the modifications listed below were carried on every Mercury flight and numerous changes were made along the way in the interest of improvement or as a result of flight data obtained from failed Atlas launches. Quality control and checkout procedures also improved and became more detailed over the course of the program.

Rate gyros

The rate gyro package was placed much closer to the forward section of the LOX tank due to the Mercury/LES combination being considerably longer than a warhead and thus producing different aerodynamic characteristics (the standard Atlas D gyro package was still retained on the vehicle for the use of the ASIS). Mercury-Atlas 5 also added a new reliability feature—motion sensors to ensure proper operation of the gyroscopes prior to launch. This idea had originally been conceived when the first Atlas B launch in 1958 went out of control and destroyed itself after being launched with a non-functioning yaw gyro, but it was phased into Atlas vehicles only gradually. One other Atlas missile test in 1961 also destroyed itself during launch, in that case because the gyroscope motor speed was too low. The motion sensors would thus eliminate this failure mode.

Range safety

The range safety system was also modified for the Mercury program. There would be a three-second delay between engine cutoff and activation of the destruct charges so as to give the LES time to pull the capsule to safety. More specifically, if the Range Safety destruct command was sent, the ASIS system would allow the engine cutoff signal to go through, while blocking the destruct signal for three seconds. The decrease in engine performance would then be sensed by the ASIS, which would activate the LES, after which the destruct signal would be unblocked and destroy the launch vehicle. Engine cutoff and destruct commands were also blocked for the first 30 seconds of launch to prevent a malfunctioning vehicle from coming down on or around the pad.

Autopilot

D-series Atlas missiles as well as early SLV variants carried the old-fashioned electromechanical autopilot (known as the "round" autopilot due to the shape of the containers its major components were housed in), but on Mercury vehicles, it was decided to use the newer transistorized "square" autopilot developed for the E and F-series missiles, and for the upcoming Atlas-Centaur vehicle. The first three Mercury-Atlas vehicles still had the round autopilot and it was flown for the first time on Mercury-Atlas 3, but failed disastrously when the booster did not perform the programmed pitchover maneuver and had to be destroyed by Range Safety action. Afterwards, the missile programmer was recovered and examined. While the exact cause of the failure was not identified, several causes were proposed and a number of modifications made to the programmer. On Mercury-Atlas 4, high vibration levels in flight resulted in more modifications and it finally worked perfectly on Mercury-Atlas 5.

Telemetry

Beginning on MA-3, a newer transistorized telemetry system replaced the old vacuum tube-based unit, which was heavy, had high power consumption, and tended to suffer from signal fade as vehicle altitude increased. As with most SLV configurations of Atlas, Mercury vehicles carried only one telemetry package while R&D missile tests had three.

Antenna

The guidance antenna was modified to reduce signal interference.

LOX boil-off valve

Mercury-Atlas vehicles utilized the boil-off valve from the C-series Atlas rather than the standard D-series valve for reliability and weight-saving reasons.

Combustion sensors

Combustion instability was a repeated problem in static firing tests of the MA-2 engines and had also caused the on-pad explosion of two Atlas vehicles in early 1960. Thus, it was decided to install extra sensors in the engines to monitor combustion levels and the booster would also be held down on the pad for a few moments after ignition to ensure smooth thrust. The engines would also use a "wet start", meaning that the engine tubes would contain an inert fluid to act as a shock damper (the two failed Atlas D flight tests used dry starts, with no fluid in the engine tubes). If the booster failed the check, it would be automatically shut down. By late 1961, after a third missile (27E) had exploded on the pad from combustion instability, Convair developed a significantly upgraded propulsion system that featured baffled fuel injectors and a hypergolic igniter in place of the pyrotechnic method, but NASA were unwilling to jeopardize John Glenn's upcoming flight with these untested modifications and so declined to have them installed in Mercury-Atlas 6's booster. As such, that and Scott Carpenter's flight on MA-7 used the old-style Atlas propulsion system and the new variant was not employed until Wally Schirra's flight late in 1962.

Static testing of Rocketdyne engines had produced high-frequency combustion instability, in what was known as the "racetrack" effect where burning propellant would swirl around the injector head, eventually destroying it from shock waves. On the launches of Atlas 51D and 48D, the failures were caused by low-order rough combustion that ruptured the injector head and LOX dome, causing a thrust section fire that led to eventual complete loss of the missile. The exact reason for the back-to-back combustion instability failures on 51D and 48D was not determined with certainty, although several causes were proposed. This problem was resolved by installing baffles in the injector head to break up swirling propellant, at the expense of some performance as the baffles added additional weight and reduced the number of injector holes that the propellants were sprayed through. The lessons learned with the Atlas program later proved vital to the development of the much larger Saturn F-1 engine.

Electrical system

Added redundancy was made to the propulsion system electrical circuitry to ensure that SECO would occur on time and when commanded. The LOX fuel feed system received added wiring redundancy to ensure that the propellant valves would open in the proper sequence during engine start.

Tank bulkhead

Mercury vehicles up to MA-7 had foam insulation in the intermediate bulkhead to prevent the super-chilled LOX from causing the RP-1 to freeze. During repairs to MA-6 prior to John Glenn's flight, it was decided to remove the insulation for being unnecessary and an impediment during servicing of the boosters in the field. NASA sent out a memo to GD/A requesting that subsequent Mercury-Atlas vehicles not include bulkhead insulation.

LOX turbopump

In early 1962, two static engine tests and one launch (Missile 11F) fell victim to LOX turbopump explosions caused by the impeller blades rubbing against the metal casing of the pump and creating a friction spark. This happened after over three years of Atlas flights without any turbopump issues and it was not clear why the rubbing occurred, but all episodes of this happened when the sustainer inlet valve was moving to the flight-ready "open" position and while running untested hardware modifications. In addition Atlas 113D, the booster used for Wally Schirra's flight, was given a PFRT (Pre-Flight Readiness Test) to verify proper functionality of the propulsion system. On MA-9, a plastic liner was added to the inside of the pumps to prevent this failure mode from recurring.

Pneumatic system

Mercury vehicles used a standard D-series Atlas pneumatic system, although studies were conducted over the cause of tank pressure fluctuation which was known to occur under certain payload conditions. These studies found that the helium regulator used on early D-series vehicles had a tendency to induce resonant vibration during launch, but several modifications to the pneumatic system had been made since then, including the use of a newer model regulator that did not produce this effect.

The flow of helium to the LOX tank on Mercury vehicles was limited to 1 lb per second. This change was made after Atlas 81D, an IOC test from VAFB, was destroyed in-flight due to a malfunction that caused the pressurization regulator to overpressurize the tank until it ruptured.

Hydraulic system

The hydraulic system on Mercury vehicles was a standard D-series Atlas setup. The vernier solo accumulator was deleted as Mercury vehicles did not perform vernier solo mode. A hydraulic pressure switch on MA-7 was tripped and flagged an erroneous abort signal, so on subsequent vehicles additional insulation was added as cold temperatures from LOX lines were thought to have triggered it.

Propellant utilization system

In the event that the guidance system failed to issue the discreet cutoff command to the sustainer engine and it burned to propellant depletion, there was the possibility of a LOX-rich shutdown which could result in damage to engine components from high temperatures. For safety reasons, the PU system was modified to increase the LOX flow to the sustainer engine ten seconds before SECO. This was to ensure that the LOX supply would be completely exhausted at SECO and prevent a LOX-rich shutdown. The PU system was set up in the Atlas C configuration through MA-6 in the interest of reliability, the standard D-series PU setup not being used until MA-7.

Skin

Big Joe and MA-1's boosters sported thicker gauge skin on the fuel tank but the LOX tank used the standard D-series missile skin. After the loss of the latter vehicle in flight, NASA determined that the standard LOX tank skin was insufficient and requested it be made thicker. Atlas 100D would be the first thick-skinned booster delivered while in the meantime, MA-2's booster (67D) which was still a thin-skinned model, had to be equipped with a steel reinforcement band at the interface between the capsule and the booster. Under original plans, Atlas 77D was to have been the booster used for MA-3. It received its factory rollout inspection in September 1960, but shortly afterwards, the postflight findings for MA-1 came out which led to the thin-skinned 77D being recalled and replaced by 100D.

The LOX tank skin was thickened still further on MA-7 as the operational Mercury flights carried more equipment and consumables than the R&D ones and capsule weight was growing.

Guidance

The vernier solo phase, which would be used on ICBMs to fine-tune the missile velocity after sustainer cutoff, was eliminated from the guidance program in the interest of simplicity as well as improved performance and lift capacity. Since orbital flights required an extremely different flight path from missiles, the guidance antennas had to be completely redesigned to ensure maximum signal strength. The posigrade rocket motors on the top of the Atlas, designed to push the spent missile away from the warhead, were moved to the Mercury capsule itself. This also necessitated adding a fiberglass insulation shield to the LOX tank dome so it wouldn't be ruptured by the rocket motors.

Engine alignment

A common and normally harmless phenomenon on Atlas vehicles was the tendency of the booster to develop a slight roll in the first few seconds following liftoff due to the autopilot not kicking in yet. On a few flights however, the booster developed enough rolling motion to potentially trigger an abort condition if it had been a crewed launch. Although some roll was naturally imparted by the Atlas's turbine exhaust, this could not account for the entire problem which instead had more to do with engine alignment. Acceptance data from the engine supplier (Rocketdyne) showed that a group of 81 engines had an average roll movement in the same direction of approximately the same magnitude as that experienced in flight. Although the acceptance test-stand and flight-experience data on individual engines did not correlate, it was determined that offsetting the alignment of the booster engines could counteract this roll motion and minimize the roll tendency at liftoff. After Schirra's Mercury flight did experience momentary roll problems early in the launch, the change was incorporated into Gordon Cooper's booster on MA-9.