The Design and Engineering of Curiosity

by Emily Lakdawalla

  After the Preliminary Design Review, the sample handling approach changed again. A drill would simultaneously penetrate into and powder the rock, augering it into a sample chamber. Then the drill would transfer the material from the sample chamber into a device on the arm that could sieve, portion, and deliver the right kind of sample to the waiting science instruments. The approach was simpler than the two-arm core-crushing solution, but the development effort started very late. Also, the switch from coring to percussive drilling and the addition of sample handling hardware to the end of the arm increased the weight of the turret from 15 to 34 kilograms.29 The 2-meter-long arm and its 5 motors would need to be much more robust than planned to support all of that weight.

  1.5.2 Motor problems

  MSL’s design included a total of 31 motors, not counting the ones in the science instruments: 6 to rotate the wheels, 4 to steer them, 2 for the high-gain antenna, 3 for the mast, 3 for the instrument inlet covers, 5 for the robotic arm, and 8 for the drilling, sampling, and dusting hardware in the arm turret. The MSL mission had sought to reduce the rover’s power demands by using dry-lubricated motors that could operate at very low temperatures without being heated for all but 4 of these. The effort had started off well: JPL successfully developed titanium-geared motors lubricated with powdered molybdenum disulfide that operated perfectly under Martian conditions, down to a minimum temperature of –135°C, colder than the Mars minimum of –127°C (the freezing point of carbon dioxide).30 But the motors failed later lifetime tests that checked how they would cope with the demands of operating under Martian conditions for millions and millions of revolutions. Over time, the titanium gears fatigued and cracked, their teeth falling out.

  With schedule pressing, they had no choice but to return to the old way of doing things: steel motors with wet lubricant, based as much as possible on the titanium motor designs that had been worked on to date. These motors could not operate under most Martian conditions without heaters. Mars Exploration Rover motors had such heaters, thin strips that were elegantly incorporated into flat surfaces planned for that purpose in the motor design. MSL motors had been designed without these interior surfaces, so there was no choice but to add them to the exteriors of the motors. Furthermore, the MSL motors were dramatically larger than the Mars Exploration Rovers’, and took correspondingly longer to heat. In order to heat them up in a reasonable 1 to 2 hours, they would require larger heaters, consuming more power, than they would have if they had been planned into the design from the start. In fact, they would consume so much power that they would leave the rover’s battery depleted, with little power remaining to drive or do science.31

  Steel motors were also heavier than the titanium ones would have been. The mass of the robotic arm increased by 14%, sending engineers back to the drawing board to make sure all the parts would be up to the stresses of landing and driving at their new masses.32 The motors were supplied by one vendor; the arm was being manufactured by a different vendor. Wherever the two components interacted, it took a tremendous amount of planning and work to coordinate the design efforts of the two companies. Engineers tried to minimize the mass of the heavier motors, but that meant repeatedly changing their specifications, delaying their delivery. Arm development was delayed. Meanwhile, the late redesign of the sample handling system meant that requirements for the motors in the arm kept changing even as design of the new motors had already begun.33

  The switch to wet-lubricated motors created new constraints for landing site selection. Although the rover could theoretically operate at high altitudes and high latitudes, severe winter cold would leave the rover without sufficient power to warm up its motors and drive. The Mars Exploration Rovers mostly parked in winter because shorter days without overhead sunlight limited their power. MSL could potentially suffer the same problem, if sent to a landing site more than 15° south of the equator.34 (Southern winters are harsher than northern winters, due to the combination of higher average elevations, seasonal timing, and elliptical shape of Mars’ orbit.) Winter immobility was acceptable for Spirit and Opportunity because the rovers had never been intended to last into the Martian winter anyway. But for MSL, which was supposed to be able to land nearly anywhere on Mars and operate for a full Martian year, cold-weather inactivity could prevent mission success.

  Even the science instruments were affected. Power limitations meant that activities requiring rover motion would need to be planned for the warmest part of the Martian day. But some instruments, like ChemCam, had detectors that worked best when cold. It looked like ChemCam was stuck between a rock and a hard place, as principal investigator Roger Wiens explained in a memoir:

  With the change in motor lubrication, normal operations became restricted to between 10 am and 4 pm local time to avoid excessive energy use. MSL was short on system engineering help, but eventually the combined impact of these decisions was noticed. There was an MSL leadership meeting in late 2007 in which the issue was addressed. A chart was displayed showing in one color the times of day during which ChemCam’s detectors would be cool enough to meet its science requirements, with another color indicating the times of day that the rover’s mast would be warm enough to point to targets. I was told that the room erupted in laughter. The two colors never overlapped. ChemCam was cool enough only at night and the mast gimbals were warm enough only in the middle of the day. It was a perfect mismatch.35

  Despite a massive effort to adjust designs to cool the ChemCam detectors, it looked like the instrument might only be usable for half an hour at each end of an operational day.

  1.5.3 Heat shield failure

  The other disaster of 2007 involved the mission’s heat shield. All NASA Mars missions to date had used a material for the heat shield called Super Lightweight Ablator, or SLA. The material consisted of a mixture of corkwood, epoxy, and gas-filled silica-glass spheres that is packed into a honeycomb structure pre-shaped into the blunt cone of the heat shield, allowing SLA to be used for heat shields of any size and shape. So MSL’s larger size didn’t present a manufacturing problem. But MSL was more massive and it would be arriving at Mars at a higher speed than previous spacecraft, imposing higher pressure and temperature on it, and causing air to flow around it turbulently rather than smoothly. And the heat shield would be tilted at an angle as the spacecraft performed its guided entry, heating it asymmetrically. The first great challenge of developing the heat shield was to develop new ways to test all these more-extreme conditions.

  The heat shield failed these new tests “drastically.”36 Instead of receding slowly, the material emptied out of the honeycomb cells catastrophically, in mere seconds. The engineers couldn’t figure out why, so they couldn’t even attempt to modify the design to prevent failure. They had to start from scratch – and had only 18 months to find a new solution.

  NASA’s human exploration program came to the rescue. An alternative heat shield material had recently been developed and tested for use in a large heat shield. The material, called Phenolic Impregnated Carbon Ablator (PICA), was first used successfully on the Stardust sample return capsule, which returned to Earth in January 2006. The same year, the Orion human exploration program adopted PICA for their capsule, and subjected it to intensive tests at even more extreme conditions than MSL faced. To support the Orion effort, the manufacturer worked to double their manufacturing capacity, bringing the new capability online exactly on time to produce a heat shield for MSL.

  Lockheed Martin started developing a new PICA heat shield in October 2007 on a fast-tracked schedule that had a Preliminary Design Review on February 7 and Critical Design Review on June 13, 2008. A major challenge was that PICA could not be built in large custom-shaped pieces. Individual pieces were limited to the 0.81-meter diameter of the Stardust sample return capsule. MSL’s heat shield was 4.5 meters across. It would have to be tiled, like the Space Shuttle’s ceramic exterior, in Stardust-capsule-sized pieces.

  1.5.4 Critical Design Review

  The MSL proj
ect had had its Critical Design Review in June 2007, before either the actuator or heat shield problems had fully come to light. The project passed, but the review panel noted two very serious problems. The first was that the sample handling mechanism design was not nearly mature enough. The second was that schedule pressure and development problems likely meant that the mission would need a budget increase of about $75 million.

  1.5.5 Stern descopes

  The mission requested more money at an unfortunate time. There was a new Associate Administrator of the Science Mission Directorate at NASA: Alan Stern, a planetary astronomer and aerospace engineer best known for being the principal investigator of the New Horizons mission to Pluto and the Kuiper Belt. Stern had already publicly expressed frustration with cost overruns on some NASA missions harming others. In a period of 5 years, he noted, a total of $5 billion worth of cost overruns had diverted funds from research programs and caused opportunities for other missions to be lost. Also, Stern shared with other members of the science community a concern that NASA was focusing too much of its limited resources on Mars, to the detriment of all the other compelling destinations in the solar system.

  Stern told the MSL project that they could not have their requested budget increase. Rather, they had to descope their mission – remove capabilities in order to keep the mission within its original budget. Whatever savings could not be realized with descopes, they would have to take out of other missions within NASA’s Mars Exploration Program. Stern asked JPL for suggestions of what could be cut.

  There were not many options. Most of the rover’s planned hardware was exactly what was needed to get the spacecraft safely to the surface. They cut a planned rock-grinding tool, replacing it with a simpler brushing tool. Richard Cook had no choice but to put science instruments on the list of things to be cut. The mission drew up a list and sent it to NASA: ChemCam, Mastcam, and MARDI. NASA accepted the cutting of ChemCam and MARDI, but pushed back on Mastcam, knowing that it would be hard to justify to the public the removal of all color photo capability from the rover. Could Mastcam be built cheaper? JPL knew that the electronics were already complete, but the optics were not, so they offered up the Mastcam optics, descoping the zoom and focus capability, leaving the rover with cameras of two different, fixed, focal lengths, and fixed focus.37

  The descopes were announced in a press release from NASA on September 17, 2007:

  The MSL project required some focused and prudent reductions in scope in order to better ensure project success. Furthermore, because all of the funds MSL requested were not available in the Mars Exploration Program reserves pool, and because SMD did not want to impact other current or future science missions to fund these new costs, the Science Mission Directorate at NASA Headquarters has been working closely with the MSL project and the science community to identify mission scope reductions to minimize the project’s need for funds, while minimizing both technical risk and impacts to the mission’s science return.

  As a result of this careful process, a combination of low-impact mission scope reductions and some new funding from the Mars Program’s reserves pool, has been agreed upon. Together these measures effectively resolve the MSL cost increase issues identified at its [Critical Design Review].

  Engineering changes to the mission include some reductions in design complexity, reductions in planned spares, some simplifications of flight software, and some ground test program changes. These changes were selected largely to help reduce mission risks. Changes in mission science content were limited to removal of the Mars Descent Imager (MARDI), the MASTCAM zoom capability from the mission, and a change from a rock grinding tool to a rock brushing tool. As noted by the science input NASA received, most of MARDI’s capability can be provided by the Mars Reconnaissance Orbiter’s HiRise camera now in orbit and working successfully. Furthermore, NASA has directed that the project expend no additional funds on ChemCam, and cost-cap SAM and CheMin at their current budgets. Future budget requests for these instruments cannot be funded.

  In total, the cuts would theoretically free up about $26 million to augment mission reserves, although that number failed to account for the termination costs of the contracts with industry partners. The impact on MSL’s science capability was severe. The most painful loss was ChemCam, which could not be included on the rover without further expenditure of funds. The laser instrument represented the rover’s only capability to measure rock composition from a distance. Without ChemCam, the rover would need to drive up and touch with APXS every rock it might want to explore in situ, at enormous cost to operational efficiency. The loss of zoom capability on Mastcam would be detrimental to its usefulness for studying the landscape at various spatial scales. And the budget axe loomed over SAM and CheMin. The Mars science community decried the cuts as being penny-wise and pound-foolish.

  At the same time, Stern directed that MSL actually add something to its design: a sample cache, a simple basket designed to hold Mars rock samples. The sample cache was intended as a first step towards Mars sample return. Stern specified that funds would come from within the Science Mission Directorate, but outside the Mars Exploration Program. Even so, members of the Mars community were angry that Stern had cut science instruments from MSL to save relatively small amounts of money, while spending for a sample cache that few people believed would ever be picked up by a future sample return mission.

  Representatives of the Mars Exploration Program Analysis Group (MEPAG), led by Jack Mustard (who was not a Mars Science Laboratory science team member), met with Stern and other Headquarters personnel on September 24, 2007. In a summary of the meeting, Mustard wrote: “Almost all comments relayed to the MEPAG chair preceding the meeting were related to ChemCam, noting the loss of remote geochemical capability and the significant investments of the French science and technology communities. MEPAG relayed concerns regarding the international implications of the stop funding order for ChemCam.” The group also expressed their reservations about the proposed sample cache: “The MEPAG group re-iterated the need for appropriate samples for [Mars Sample Return], and that poorly documented rock fragments in an open sieve basket will not meet the criteria for science as outlined in numerous National Academy and MEPAG reports.”

  1.5.6 Second site selection workshop

  The landing site selection committee convened the second community workshop on October 23, 2007. The mood at the second meeting was anxious, to say the least. Scientists were still reeling from the instrument descopes. Many scientists learned for the first time about the failures in the heat shield and motor development process. The motor problems threatened to take many favorite southern hemisphere sites out of consideration.

  Engineers Mike Watkins and Adam Steltzner told the scientists that higher-elevation landing sites were substantially riskier than lower-elevation sites, another blow to southern hemisphere fans.38 In fact, the highest-elevation sites were so risky that Watkins and Steltzner requested that scientists proposing high-elevation sites develop a related “safe haven” site at similar latitude but much lower elevation. At the end of the workshop, the list of possible sites had formally narrowed to 11, but there was a great deal of uncertainty about whether MSL would be capable of landing at any of them, and what the quality of its science would be without ChemCam and the other cut instruments.

  Just weeks later, ChemCam and MARDI were saved. The good news was announced in a letter from Alan Stern and Jim Green on November 8:

  Malin Space Science Systems has agreed that there will be no additional costs to NASA for the completion of the Mars Descent Imager (MARDI). Furthermore, funds returned to the Mars Exploration Program from the unfortunate elimination of MARDI operations on Phoenix will be used to support MARDI integration on MSL. In the case of ChemCam, LANL, the French Space Agency (CNES), and even other MSL instrument team members have developed a series of descopes and support arrangements to allow instrument completion, reducing the development cost-to-go by a little over 80%; i.e., from $2.5M to
about $400K. As a result, ChemCam will be funded another $400K by the Mars Exploration Program, allowing them to complete development.

  1.5.7 MARDI wheeling and dealing

  MARDI was returned to MSL at its principal investigator Mike Malin’s personal expense, with the help of a behind-the-scenes agreement with the University of Arizona’s Peter Smith, the principal investigator of the recently-launched Phoenix Mars lander. At the moment that it was descoped, Malin says, NASA saved about $80,000 by doing so. This was such a relatively tiny number that unusual funding sources might work to meet the shortfall. The Planetary Society, a nongovernmental organization, discussed with Malin running a fundraiser among its members to complete it. In the end, Malin decided to put up the money himself.39

  So MARDI was finished, but there was still no money to put it on the rover. That money came from Phoenix. MSL was not the first spacecraft with a MARDI. There had been one on Mars Polar Lander, and consequently there was one on the backup lander hardware that later became the Phoenix mission. Peter Smith had invited Malin to operate MARDI on the Phoenix mission. But there was a problem with the interface between MARDI and Phoenix’ main computer, and ultimately Phoenix MARDI was not used at all during the mission. Smith offered to give to JPL the money that he would have paid Malin to support Phoenix MARDI operations. It was enough to pay to integrate MARDI on the rover. JPL agreed, but said that the integration had to happen immediately.