by Rod Pyle
Of the many improvements made to Phoenix over the failed Mars Polar Lander, the method used to shut down the descent engines was perhaps most important. MPL had used a “shock sensor” that was supposed to have triggered the engine shutdown when the lander set down. Unfortunately, an unrelated shock felt during descent (probably the deployment of the landing legs), while MPL was still over one hundred feet up, shut down the rockets prematurely, and the lander crashed. Phoenix would return to a simpler method, similar to one used as far back as the Apollo lunar lander: switches on the footpads would signal actual touchdown. This was one of a number of improvements made in an attempt to ensure a successful mission.
The Mars Phoenix mission was unfamiliar ground for NASA. It would be the first mission in the space agency's history to be led and operated by an academic institution, the University of Arizona. This was worrisome to an organization used to complete control.1 To make things more complex, a number of foreign universities would contribute instruments and expertise to the mission, including institutions in Canada, Germany, the United Kingdom, Finland, Denmark, and Switzerland. The result was a potpourri of scientific and technical input. It is important to note that while the University of Arizona operated the spacecraft, the navigation and landing were controlled by JPL in Pasadena.
The probe landed successfully in May 2008. It was the first lander to travel to the polar regions of any planet. And Phoenix was unusual in other ways. It was small in comparison to any other Mars lander except perhaps Pathfinder, weighing in at just under eight hundred pounds. About five feet across (eighteen feet with solar panels deployed), it was almost the same size as an individual Mars Exploration Rover and would look downright puny next to Viking. And after the bouncy-ball landings of Pathfinder and the Mars Exploration Rovers, Phoenix signaled a return to the rigors of a powered descent, where rockets must handle the final moments of a soft, pinpoint landing. It was not a mission for the faint of heart, especially with the fifteen-minute delay from Earth to Mars for communication. At times, add an hour or two for the orbiting Mars-probe relay, and you have a real issues. Planning was key.
As a largely recycled spacecraft, Phoenix was a compromise mission. It was designed to be a low-cost attempt at a Mars lander (in this case, less than $500 million). This had a number of real impacts on the mission design. And the stakes were high, as this was one of the first planetary missions since the twin losses of Mars Climate Orbiter and Mars Polar Lander; JPL could not afford another failure. Perhaps for this reason, as much as any other, the design and control of the mission followed an unusual path.
While built by Lockheed Martin, Phoenix was designed by a consortium of academic and NASA/JPL personnel. The mission proposal originated from the University of Arizona, and parts of the spacecraft (notably the camera) were actually built there. An ad hoc mission-control center was also sited at the campus, looking more like a low-end software company than a deep-space mission-control room. Many of the staffing needs were met by hiring young and inexpensive talent; some were even fulfilled via the use of grad students. Not surprisingly, many of the key players were from the Mars Pathfinder mission, itself a departure from traditional JPL methods. This was not your father's Mars mission.
Even the software intended to run it was recycled from programming originally designed to operate a Mars orbiter. It had to be rewritten and repurposed for a lander, resulting in some last-minute sweat and angst as the software team tested and retested the command structure to make sure it would meet the demands of short-term surface operations. Much of it was left open-ended; changes and new instructions would be regularly uploaded from the University of Arizona controllers and JPL via the Deep Space Network. This recycling of existing software created many sleepless nights for the coding team.
As with previous missions, the processing power of the flight computer was not particularly robust by consumer standards. In fact, the flash memory was only one hundred megabytes; far below off-the-shelf flash drives that were offering over four gigabytes. But given the expense of flight-rated, radiation-hardened components, it was what the budget could support. As with many of JPL's spacecraft, the CPU was the venerable RAD 6000 chip manufactured by IBM, which had been extensively proven in spaceflight, having powered the computers of the Mars Exploration Rovers, Mars Pathfinder, and Mars Odyssey, among others. This was a proven design, and there were many available engineers and programmers who knew how to squeeze the maximum performance out of the chip. It ran at a nonblistering thirty-three megahertz and cost upward of $250,000. It was a far cry from the two-gigahertz Mac G5 available down the street for under $2,000.2
The instrumentation on the Phoenix lander was a compromise of light weight, low cost, and high reliability. Deciding what to include was, as always, a study in creative compromise.
Paramount for a lander of this type and purpose was a robotic arm. Unlike Viking's spring-steel “wind-up” arm, it utilized a more traditional and simple hinged “elbow.” Designed to extend almost eight feet from the lander, it was capable of digging about eighteen inches deep. Besides a scoop, there was a drill-like rotating rasp for shaving ice samples (a true drill would be too heavy). A small camera was affixed to the end of the arm. As powered landings were still rare at the time (as opposed to the more passive beach-ball approach) another camera, the descent imager (complete with a microphone) was installed to transmit video of the landing. Unfortunately, at the last minute an electronic flaw prevented the use of the camera or microphone. Important data and a wonderful PR opportunity were lost.
The primary camera, the surface stereo imager (SSI), was similar in design to that on the top of Pathfinder's rover, Sojourner—largely because it was built by the same team and had worked well on that mission. It would extend vertically on a mast from the lander.
To search for organics (precursors to life) in the soil, a set of small, high-temperature ovens called the Thermal and Evolved Gas Analyzer (TEGA) would use a spectroscope to analyze gasses baked out of soil samples. The TEGA consisted of eight tiny, one-use oven compartments, each about the size of a pen. The instrument would look for water, carbon dioxide, and organic elements such as methane. It shared a more-than-passing resemblance to the ovens found in the Viking lander thirty years previous.3
Phoenix was packed with other instrumentation. The Microscopy, Electrochemistry, and Conductivity Analyzer (MECA) device would examine soil particles on a microscopic level, using a so-called wet chemistry lab or WCL. The WCL had test chambers where purified water would be added to soil samples and sensors would measure ionic activity, looking for biological compatibility of the sample (by earthly standards). This would give some idea of how able the soil would be to harbor microbial life. This was an important distinction from the Viking missions: rather than searching for life, this would search for the ability to support life.
Also inside the MECA was the Thermal and Electrical Conductivity Probe (TECP). This would measure soil temperature, humidity, thermal (temperature) conductivity, electrical conductivity, and other properties of the soil fed to the MECA. An optical microscope was included to take pictures of these samples at an extreme magnification, using arrays of multicolored LEDs for different results under different colors of illumination. Each color would reveal different properties in the magnified samples. The final component of MECA was an atomic force microscope (AFM) that shared space with the optical microscope. Small silicon crystal tips would brush across the sample, measuring repulsion from the sampled soil to analyze its composition at the atomic scale.
Atop Phoenix, a meteorological station would provide ground-level measurements of the Martian day and night. The technology used ranged from a simple telltale (not unlike a wind sock used at airports) to a highly innovative and complex LIDAR (laser-powered radar) that measured dust, ice particles, and moisture in the air. Of course, temperature, humidity, and air pressure were measured as well.
As always with a lander, the decision where to set down was a long and
arduous process. The vastly improved images available from the Mars Reconnaissance Orbiter and Mars Odyssey helped a lot; rocks as small as twenty inches across could be seen now. This was a far cry from even the Pathfinder mission, when the landing zones were selected by a combination of medium-resolution imagery, intuition, and luck. And, to bolster confidence further, Phoenix had fourteen inches of ground clearance, as opposed to Viking's 8.5 inches. Nevertheless, the final selection would not be made until after Phoenix had left the launch pad.
Phoenix set down in an area called Green Valley in Vastitas Borealis (“Northern Waste”), not far from the north Martian pole, on May 25, 2008. It was the first successful powered landing since Viking 2 in 1976. This region was finally selected for smoothness and other safety considerations, and it was also where the largest concentration of water ice (besides the pole itself) had been found to date.
Among its first duties were to transmit data about its orientation and status back to JPL. As it turned out, things had gone as well as they could have dreamed; the lander was just about level. Next, the giant solar panels were unfolded to give the hungry batteries the power they needed. As quickly as possible, team members needed to establish how close they had gotten to their point of aim for the landing. When the result came in, they were astonished. They were right on target. The accuracy of the landing impressed people all the way to the top of the NASA food chain, one of whom characterized the feat as making a hole in one with a golf ball launched in Washington, DC, toward a moving target in Australia. The pinpoint arrival was confirmed by an image snapped by the Mars Reconnaissance Orbiter, which spotted not only the lander but also the parachute resting nearby.
Soon it was time for the first look at the surrounding terrain—and what a delight that turned out to be. All around the lander were fascinating polygonal shapes etched into the permafrost. The cracks in the soil appeared to be fresh—older ones would have filled in or eroded away. This indicated ongoing changes within the soil, thawing and refreezing as the ambient temperatures swung from one extreme to another. Of course, this also highlighted the urgency of getting Phoenix's work started as quickly as possible, for these polygons were a stark reminder of temperature extremes what would kill a lander. Already, the onboard thermometers were beginning to measure temperatures that would range from -22°F to -122°F. And while Phoenix was designed to operate within this range, and colder, the specter of the advancing north polar winter, about three months away, weighed heavily on everyone involved. Phoenix would most likely not survive the long Martian polar night.
Then, just as things were getting interesting, Phoenix went silent. The link between the lander and the MRO orbiting overhead failed, plunging the mission into hours of tense data darkness. A day later, the issues had been ironed out, and the Mars Odyssey orbiter, still operating after almost a decade, was also pressed into relay service. But another snag occurred almost immediately. There was an excruciating delay while issues with the robotic arm were worked out. The low-cost approach to building and operating Phoenix seemed to be manifesting gremlins. Ultimately, the arm was unable to touch Martian soil until May 31, almost a week after touchdown.
Without further delay a sample was scooped up from the frozen soil and then the arm folded back and dumped it into the funnel leading to the TEGA analyzer. At least, that's what they thought it had done. The telemetry was less convincing—it appeared that none of the soil retrieved had actually gotten into the sample container below. Images of the entry area of the unit showed the soil sitting atop the screen that was supposed to sift the dirt before it dropped into the TEGA's oven. There the sample sat, comfy as a fat cat on a pillow. Hypotheses were formulated: the dirt was too wet, or to clumpy. It might be a clay of fine, sticky particles—not dry, loose particles. Or the screen was clogged. Or the Great Galactic Ghoul had stepped in. Nobody knew.
So controllers turned on a vibrator attached to the screen to shake some of the soil down into the TEGA. Nothing happened. They shook it some more…and some more. Ultimately, they shook the screen far longer than it had been designed to do, almost an hour, but still, no dirt.
There was a photocell—an electronic component that measured light in each of these small containers. If the sample had gone in, it should have blocked the photocell, at least partially. And at this point, the cell was picking up a nice, fat signal from the LED-generated light just a fraction of an inch away, across the empty container. So, no dirt.
This was one of those sweaty moments in a robotic mission: planners had to decide how to proceed with a compromised result on the lander and a collection of unattractive options. They could keep trying to get the sample down into the TEGA, but it was almost certainly dried-out and less interesting by now. Another option was to simply button up the oven and bake whatever might be in there, but if it was empty, this would waste an oven—they were strictly a one-shot deal. Or an attempt could be made to dump another sample on top of the one that was not entering the oven, but that might not work either. It was a vexing problem.
A parallel issue was an ever-expanding discussion about which area near the lander to sample next, assuming the sampling procedures could be worked out. Some scientists wanted to dig from the same trench, some wanted to try another area. And on top of it all was the communication delay and programming time—things thought up on Thursday were not able to be executed till Friday or later.
It was like swimming through a molasses of variables. And, at ten days into a ninety-day mission, time was of critical and finite quantity. Since this lander was near the frigid pole, the usual number of mission extensions was unlikely. Engineers and specialists swarmed the robotic twin of Phoenix at the control center at the University of Arizona, seeking answers to their questions. One plan that met approval was the idea of using a “sprinkle” technique instead of a “dump” technique. Yes, it gets that precise. Rather than simply dropping the soil sample on top of the entry to the ovens, they would hold the arm's dirt scoop over the funnel and run the small rasp motor attached to it. The vibration from this should cause the fine, silt-like particles that appeared in the pictures to rattle down a small depression in the shovel itself. The result should be that rather than simply dumping the entire scoop of soil onto the grating, only the finer particles would be sent in and assured to enter (and fill) the oven. Or so they hoped.
The idea was tested first on Earth, then on Mars. A small amount of soil was sprinkled onto the deck of the lander…and success! It appeared that the sprinkle technique would separate out the finer grains from the gunky soil. The maneuver was repeated to get a sample into another one of the TEGA ovens, and things were back on track. The shake-then-bake was under way. The oven would heat the soil for days, one step at a time, ending up at about 1800°F. Then the TEGA itself would sniff for compounds in the resultant gasses.
One problem down. Meanwhile, the Phoenix team was still seeking consensus on the next place to sample. There were now a variety of trenches dug by the arm to pick from: Dodo, Goldilocks, and Wonderland were just three of the names used. It's interesting to compare this naming scheme to Pathfinder, when a rock could be named after Scooby-Doo® or any other critter that struck a collective funny bone. No more. One morning, someone at the NASA legal office woke up and realized that they could potentially be sued for using names of copyrighted cartoon (and other) characters. So the edict went out: use names and titles in the public domain only. Which means they have to be out of copyright. Which means waiting for seventy years after the author or creator of the work has died. Which means…oh, just pick a very old book. Alice in Wonderland will certainly do.
As the TEGA did its bake, the Phoenix team was looking at these trenches in the ground. One of the most attractive targets were bits of “white material”—which is the term the scientists used to make it clear that these were not yet identified as water ice, the holy grail—seen in some trenches. These were monitored with care to see if they would shrink over time, which would indicate meltin
g (or more properly sublimating—going right from ice to vapor, skipping the liquid water phase). Meanwhile, the TEGA results came in: carbon dioxide, check. Water vapor, check. But no ice. This was not a deal breaker, as the sample that was finally baked had sat for some time in the scoop before being ingested into the TEGA oven, and any water ice would have vanished in that time. So the jury was still out on water ice.
Then day 23 (or more properly, sol 23, as this indicates the longer Martian day) comes along, and with it, a by-now-familiar problem in the Mars spacecraft family. Phoenix drops out of contact—the lander goes into safe mode. Data gathered in the last twenty-four hours is gone, and the lander is waiting for binary CPR. Corrections are swiftly made to the software and uplinked to Phoenix at the next opportunity; things seem fine for now. Back to the “white stuff.”
After a few more days of observation, the white chunks have vanished. It was not Phoenix's doing; the lander was napping for part of the time. Bottom line: it must be ice. The announcement is made at the next press briefing.
About this time, the first results of the MECA experiment come out: Martian soil could, in fact, support some earthly crops, at least in theory. It is very alkaline, and might support asparagus but not strawberries. But that's enough.