The Design and Engineering of Curiosity

by Emily Lakdawalla

  MSL waited until its inertial measurement unit registered a speed of only about 400 meters per second and then changed its configuration again to prepare to deploy its parachute. MSL prepared for parachute deployment with the “straighten up and fly right” maneuver. The falling spacecraft threw away six 25-kilogram entry balance masses in pairs at two-second intervals. (You can see the entry balance masses on the backshell in Figure 2.10.) The release of the entry balance masses counteracted the off-center weight distribution that had been imparted by the release of the cruise balance masses.

  The aeroshell tipped up, aligning its angle of attack to within 5° of its descent trajectory. At the same time, the reaction control system rolled the spacecraft 180° (a maneuver referred to as the “victory roll”) to the desired bank angle for later radar operation purposes. Straighten up and fly right took a total of 14 seconds.36 The work of the descent stage reaction control thruster system was complete. Throughout entry and descent, the reaction control thrusters had performed a total of 2256 thrust pulses, operating for a total of 110.725 seconds (Figure 2.18).37 The aeroshell had dissipated 99.6% of the vehicle’s kinetic energy through friction with the atmosphere.38 The spacecraft was now ready to deploy its parachute. The balance masses continued along their ballistic trajectories, later impacting the ground at the northern edge of Mount Sharp, beyond the landing site to the east (Figure 2.7).

  Figure 2.18. Descent reaction control system thruster use. Odd-numbered thrusters were used for all pulses; even-numbered thrusters were secondary, used only when more thrust was needed. SUFR = Straighten Up and Fly Right. Modified from Baker et al ( 2014 ).

  2.3.5 The parachute

  MSL’s parachute had the same shape as the Vikings’, but with a diameter of 21.35 meters it was 33% larger (Figure 2.19). Another crucial difference was the distance between the backshell and parachute: Viking’s parachute trailed by 8.5 times the parachute diameter, but MSL’s lines were longer, to separate it by 10.32 times the diameter. This increased separation was designed to reduce “area oscillations” of the parachute. The parachute was composed of orange and white ripstop nylon, except for the crown, which was made of a heavier ripstop polyester. The suspension lines were made of Technora and Kevlar, both synthetic fibers with high strength and heat resistance.39

  Figure 2.19. The MSL parachute. Top: dimensions, from Cruz et al ( 2014 ). The bottom two images were taken during April 2009 testing of parachute deployment in a wind tunnel at NASA Ames Research Center. NASA/JPL-Caltech releases PIA11992 and PIA11993, annotated by Emily Lakdawalla.

  2.3.6 The descent stage

  The descent stage was a complicated spacecraft all on its own, with a mind-boggling number of systems crammed into its open structure (Figure 2.20 and Figure 2.21). At its heart was the most sophisticated propulsion system JPL had ever built. It was actually two distinct propulsion systems that had to share components to conserve mass and volume. The descent stage also served as the structural link between all other spacecraft components, with six separation nuts each connecting the top hexagon of the descent stage to the cruise stage and backshell, and three connecting the bottom of the descent stage structure to the top deck of the rover. It weighed 1068 kilograms, of which 397 was fuel.40

  Although the rover’s main computer ultimately commanded the descent stage, the descent stage contained numerous avionics of its own, including a computer to control the thruster systems; the descent inertial measurement unit, with gyroscopes that facilitated the precision flying of the guided-entry phase; an X-band radio system that was used throughout cruise, entry, descent, and landing; the Terminal Descent Sensor radar system used to measure altitude and velocity; and the bridle umbilical device used to lower the “rover-on-a-rope”.

  Figure 2.20. Descent stage parts (part 1). Top photo taken at JPL in early October 2008, bottom photo around November 2008. NASA/JPL-Caltech release PIA11425, annotated by Emily Lakdawalla.

  Figure 2.21. Descent stage parts (part 2). Top photo taken at JPL around November 2008. Bottom photo taken at Kennedy Space Center around November 2011. NASA/JPL-Caltech releases PIA11808 and PIA15020, annotated by Emily Lakdawalla.

  The Descent Reaction Control System (DRCS) that steered the aeroshell throughout entry and descent consisted of eight 250-newton thrusters in four pairs, one primary and one secondary. Holes cut into the backshell allowed these thrusters to protrude. MSL used the primary (odd-numbered) thrusters for small pulses; the secondary (even-numbered) thrusters came into play for larger pulses. These eight thrusters drew fuel from only one of the descent stage’s three fuel tanks, the one mounted toward the rover’s front. The rover’s computer updated commands to the thrusters every 125 milliseconds, commanding thrusts in increments of 15.625 milliseconds.

  The eight downward-pointing Mars Lander Engines (MLE) were much larger than the upward-pointing Reaction Control System thrusters, at 3300 newtons as compared to 250. The landing used only about two-thirds of the descent engines’ thrust capability because the low altitude of the landing site gave MSL ample time to decelerate. They drew on three propellant tanks using a flow regulator that had been launched into Earth orbit multiple times as part of the Space Shuttle Discovery before being rebuilt for MSL.41 Four of the engines were canted at 5° outboard from the rover, and four were canted at 22.5°. When the descent stage was connected to the rover, the nozzles of the engines projected beyond the rover’s belly pan, keeping exhaust clear of the rover (Figure 2.22).

  Figure 2.22. Descent stage mated to the rover and inside the backshell. The red caps on descent stage rockets, sliver wrap on rover wheels, and yellow covers on MARDI camera (square) and terminal descent sensors (round) were removed before flight. Photo taken at Kennedy Space Center in in October 2011. NASA/JPL-Caltech release PIA14756.

  The Terminal Descent Sensor sensed the ground with six radar beams. One beam pointed directly downward; three pointed at an elevation of 20° in different directions (one toward the rear and one each to left and right); and two, called the “headlight” beams, pointed forward and slightly left and right at elevations of 50° (Figure 2.23). Unlike other landing radar systems, MSL’s was “memoryless” – measurements of range and velocity were essentially instantaneous, not relying on previous measurements or even on sharing of information between beams. It was computationally intensive, but a “bad lock” didn’t propagate error forward in time, allowing the system to be robust to spurious signals.

  Figure 2.23. Terminal Descent Sensor beam pattern. Emily Lakdawalla after Pollard ( 2012 ).

  2.3.7 Descent under parachute: 259 to 375 seconds

  Still traveling at Mach 1.7, MSL fired the explosive sabot that deployed the parachute 259 seconds after hitting the entry interface, at 5:15:05 Spacecraft Event Time. The parachute filled with air, stretching its suspension lines in 1.135 seconds and fully inflating in under two seconds. In those two seconds, Mars’ gravity was still accelerating the spacecraft; it sped up by 0.743 meters per second. The parachute was qualified to survive deployment at up to Mach 2.3 and able to withstand an opening force of 289 kilonewtons. In the event, it experienced only 153.8 kilonewtons. The reaction control system thrusters remained ready to work to cancel out any spinning or rocking motions, but MSL’s descent was stable enough for them not to be needed.42 With the parachute inflated, the MEDLI instrument suite shut down. Only 20 seconds after the parachute deployed, MSL had slowed to subsonic speeds, so it dropped the heat shield, exposing the rover and descent stage to the Martian air at 5:15:24.43

  About 6 seconds before the heat shield separated, the Mars Descent Imager (MARDI) had switched on and begun taking images at an average 3.88 frames per second. The first 26 MARDI photos were black; the next 622 documented the final 2.5 minutes of landing. As the heat shield fell away, a white-balance target on the inside of the heat shield helped MARDI’s autoexposure algorithm to adjust quickly from the pitch-black interior of the capsule to the brightly lit Martian day (Figure 2.24).

  Figur
e 2.24. MARDI image of the heat shield taken at a spacecraft clock time of 397501995, one second after the release of the heat shield. MARDI image 0000MD0000000000100033E01. NASA/JPL-Caltech/MSSS.

  Angled to the east along the descent path and with a field of view of 70-by-55°, MARDI’s first images encompassed much of the eastern half of the landing ellipse. The heat shield can be clearly tracked through the first 250 of the images, and MARDI even documented the moment of its impact onto the Martian surface in image number 345, taken at 05:16:48 on the spacecraft’s clock (Figure 2.25). (Note: Time stamps in MARDI image files appear to be 3 seconds later than the spacecraft clock times in the same files. The given spacecraft clock times correctly correspond to the landing timeline in Table 2.2 within a fraction of a second.)

  Figure 2.25. Series of MARDI images documenting the impact of the heat shield onto the Martian surface. A plume of material spread for several seconds after the impact before the MARDI field of view no longer encompassed the impact site. MARDI images 0000MD0000000000100344E01 to 0000MD0000000000100358E01, taken between 5:16:45 and 5:16:49. NASA/JPL-Caltech/MSSS.

  After dropping the heat shield, MSL waited three seconds, ready to use its thrusters to cancel any rocking motion caused by the separation, but the spacecraft was steady and needed no correction. After another two seconds, it activated the Terminal Descent Sensor radar system. The five-second delay after heat shield separation was necessary to prevent the radar system from confusing the nearby heat shield with the ground. The Terminal Descent Sensor achieved radar lock about 20 seconds after heat shield jettison. One radar beam showed a spurious measurement of the “ground” at a range of 1003.66 meters and a velocity of –47.76 meters per second about 30 seconds after dropping the heat shield. This was probably a detection of the heat shield falling toward the ground!44

  Although the Terminal Descent Sensor provided information on the distance to the ground, the instantaneous altitude of the spacecraft is not necessarily the same as its altitude relative to its final landing site some distance away; it was mainly for this reason that the landing site needed to be flat throughout the landing ellipse.45

  Shortly after dropping the heat shield, direct-to-Earth transmission of the X-band signal ceased. The MSL team was now entirely dependent upon Mars Odyssey for real-time information on the status of the landing. Odyssey performed well throughout the landing, delivering continuous updates on MSL’s health.

  As MSL descended, Mars Reconnaissance Orbiter sped toward it from the south (see Figure 2.14 and look for 5:14 UTC on all three ground tracks, then follow the time forward). Less than a second after the heat shield dropped, Mars Reconnaissance Orbiter’s HiRISE camera began acquiring an image of the landing site. Like most Mars orbiting cameras, HiRISE is a “pushbroom” instrument that sweeps a long, skinny detector across the surface, taking advantage of spacecraft motion to build up an image swath about 10,000 pixels wide by as many as 126,000 pixels long. It can take as many as 100 seconds to capture a single image. Ordinarily, not much changes on the Martian surface during such a short period of time, but things were happening fast as MSL descended. HiRISE’s beam swept across the heat shield at 353 seconds after entry; it caught the backshell and then the parachute about 3 seconds later.46 So the amazing HiRISE image actually captures different moments in time for the two pieces of hardware (Figure 2.26).

  Figure 2.26. HiRISE’s amazing image of MSL under parachute, its heat shield significantly below it. HiRISE image ESP_028256_9022. NASA/JPL-Caltech/UA.

  At an altitude of 3000 meters, the rover prepared its descent stage for powered descent. When the data from the Terminal Descent Sensor indicated that the spacecraft had reached a speed of 79 meters per second and an altitude of 1671 meters, 117 seconds and 10.4 kilometers’ altitude after deploying the parachute, it primed the descent stage engines, flowing fuel to them at 1% throttle, and abruptly cut the connection to the backshell and parachute.47 For two seconds, the spacecraft plummeted, making room between it and the backshell. The parachute remained attached to the backshell, and both fell together. Lacking rockets to slow their descent further, they landed before the rover did, to the west-southwest of the rover’s landing site.

  2.3.8 Powered descent: 378 to 412 seconds

  The Mars Lander Engines throttled up, beginning the “powered approach” phase, at 5:17:04. The rockets worked to smoothly zero out the spacecraft’s horizontal motion while bringing the vertical descent rate to 32 meters per second. Figure 2.27 summarizes the work of the landing engines. At the beginning of powered approach, the descent stage also performed a divert maneuver, shifting the spacecraft’s position 300 meters to the left of the entry trajectory, a distance sufficient to ensure that the rover’s eventual landing site would not be directly on top of the already-landed backshell and parachute.48

  Figure 2.27. Mars Lander Engine (MLE) thruster operation during the spacecraft’s final descent. The position of a propellant injection device, called a pintle, in the throat of the rocket controlled the amount of thrust. Emily Lakdawalla after Baker et al ( 2014 ).

  Following powered approach, the spacecraft was finally directly above its eventual landing site. For the first time, the terminal descent sensor’s altitude readings directly measured the remaining distance to the surface: 247.9 meters. A brief descent phase called the “constant velocity accordion” saw the rover continuing to descend at 32 meters per second. The constant velocity accordion was intended to accommodate any mismatch between the terminal descent sensor’s measured altitude at the beginning of powered approach and the altitude of the actual landing site now beneath the rover. The constant velocity accordion could have accommodated as many as 100 meters of altitude difference; in fact, there were only 5.5 meters of altitude difference between the estimated and actual altitude.49 With that out of the way, at an altitude of 142 meters, MSL entered the constant deceleration phase, smoothly slowing the spacecraft from a descent rate of 32 meters per second to 0.75 meters per second.50 It was time to deploy the landing gear.

  2.3.9 The lander

  Before the MSL mission could rove Mars, its rover had to perform the functions of a lander, deploying landing legs and coming to a stable halt. Once on Mars, the landing gear had to transform into the rover’s mobility system. The rocker-bogie suspension system uses a number of passive pivots to balance out rough terrain and keep the rover body as level as possible. But during landing, with the wheels not yet touching ground, the interconnected levers of the mobility system needed to be carefully restrained until the last possible moment, to keep the six wheels as close to flat as possible upon touchdown.51 The mobility system was restrained at five points: at the four corners of the rover, connecting the rockers and bogies to the rover body, and also in the center of the rover’s back, holding the differential arm still. During flight, the long rocker arm connecting the front wheel to the rear bogie was folded nearly at a right angle in order to fit the mobility system within the cramped space of the aeroshell (Figure 2.28 and Figure 1.​7).

  Figure 2.28. Rover mobility system in stowed configuration. Photo taken at Kennedy Space Center in November 2011. NASA/JPL-Caltech release PIA15021.

  Many of the devices now visible on the top deck of the rover are related to cruise, entry, descent, and landing, and several are not used in the surface mission (Figure 2.29).

  Figure 2.29. Parts of the rover relevant to cruise, entry, descent, and landing. The base image is a self-portrait taken with the Mastcam on sol 1197 (19 December 2015). NASA/JPL-Caltech/MSSS/Emily Lakdawalla.

  2.3.10 Sky crane and landing: 412 to 432 seconds

  The descent stage switched from decelerating at a constant rate to descending at a constant rate of 0.75 meters per second, so required less rocket power. Out of concern that the descent stage rocket exhaust could impinge on the rover, the four engines canted at only 5° were throttled down to 1%, the other four throttling up to compensate. The descent stage wobbled a bit in response to the sudden change in the descent en
gines’ activity; the spacecraft allowed 2.5 seconds for those wobbles to settle out before proceeding, of which it needed only 1.25 seconds.

  At 5:17:38, at an altitude of about 21 meters, with the descent stage stable and descending at 0.75 meters per second, three pyros fired to separate the rover from the descent stage. The weight of the rover pulled on three nylon/Vectran cords wrapped across a confluence point pulley and then around a spool attached to the descent stage, called the bridle umbilical device (Figure 2.30). A brake within the spool controlled the rate of descent. The rover had pulled the cords to their full length of about 7.5 meters in 5 seconds (Figure 2.31). Along with the three strings of the bridle, the bridle umbilical device also deployed an umbilical cable that allowed commands to be passed from the rover computer to the descent stage. (An artist’s concept of the extended bridle and umbilical can be found in Figure 1.​21.) The tapered shape of the spool made it spin at a higher angular rate as the rover descended, and the faster it spun, the more the brake resisted the motion; this controlled the rate at which the rover descended under Mars’ gravitational acceleration.52

  Figure 2.30. The bridle umbilical device connection to the rover. Emily Lakdawalla after Gallon ( 2012 ).