The Design and Engineering of Curiosity
Page 10
Rocker deploy
5:17:39
953.59
413.59
397502128.59
S2014
Powered descent
Rover reaches end of bridle
5:17:43
957.89
417.89
397502132.89
S2014
Powered descent
Bogie release
5:17:44
958.89
418.89
397502133.89
S2014
Powered descent
Ready for touchdown
5:17:47
961.86
421.86
397502136.86
K2014
Powered descent
HiRISE image end
5:17:57
970.9
430.9
397502145.90
S2016
Landing
Touchdown sensed
5:17:57
971.52
431.52
397502146.52
K2014
Landing
Flyaway
5:17:58
972.31
432.31
397502147.31
K2014
Landing
First rear Hazcam image
5:18:40
1014.068662
474.07
397502189.07
L2016
Landing
First front Hazcam image
5:20:37
1131.501662
591.50
397502306.50
L2016
Landing
Mars Odyssey LOS
5:23:53
1327
787
397502502.00
W2013
Landing
Mars Reconnaissance Orbiter LOS
5:23:54
1328
788
397502503.00
W2013
*Sources for the data in the table are: K2014: Karlgaard and Kutty (2014); L2016: calculated by me from MARDI image time stamps; MC2014: Mendeck and Craig McGrew (2014); N2016: Novak et al (2016); S2014: Sell et al (2014); S2016: Christian Schaller, personal communication, email dated February 17, 2016; Sc2014: Schratz et al (2014); W2013: Way et al (2013).
Figure 2.7. Entry, descent, and landing trajectory. Base image is from Viking global mosaic, trajectories from JPL Horizons.
2.3.1 Telecommunications during landing
Ever since the loss of Mars Polar Lander, NASA has required Mars landers to be in constant communication with Earth during the dramatic and risky events of entry, descent, and landing. On the day of MSL’s landing, Mars and Earth were separated by about 248.2 million kilometers, or 828.0 light-seconds. The entire entry, descent, and landing took only 432 seconds from start to finish. There is nothing that anyone on Earth could do to rescue a mission should something go wrong; instead, any telemetry received would serve to help engineers determine the cause of a landing failure, with the goal of preventing a future one. MSL was required to transmit all highest-priority data within 3 seconds of the event it recorded. That way, some information could be salvaged from a catastrophic accident to improve future missions.19
Ideally, the spacecraft would use a single radio configuration for communications throughout entry, descent, and landing. But the MSL landing sequence had the spacecraft reconfiguring itself multiple times, throwing away hardware on which antennas were mounted. To handle communications, MSL had to switch among multiple radio systems and antennas.
There were two primary X-band radio systems for communicating directly with Earth, one within the rover (still used now for surface operations) and one attached to the descent stage. During cruise, the descent stage X-band system handled communications through a medium-gain antenna mounted to the cruise stage. During entry and descent, the descent stage communicated with Earth through two low-gain antennas mounted on the backshell’s parachute cone, beginning with the parachute low-gain antenna and later switching to the tilted low-gain antenna (Figure 2.10). The signal from such small antennas, 112 million kilometers from Earth, was weak, so they transmitted no telemetry. Rather, they broadcast 11-second-long “tones,” signals with frequencies slightly offset from the main carrier frequency, with different frequencies signifying different events. Events during the cruise, approach, and guided-entry phases were separated far enough in time that 11-second-long signals were good enough to communicate the spacecraft’s status, but after that, MSL needed higher-rate communications. When events did overlap in time, complicated logic governed which tone would play first:
If additional tone events occurred while one tone was playing, the new event was queued until the currently playing tone had completed. Then, the queued tone played. Each tone had a defined priority. Nominal tones were generally prioritized lower than tones indicating faults or specific critical events during EDL. Of the available tone events, a small subset were labeled as “stomping tones,” which interrupted a currently playing tone, causing the interrupted tone to be re-queued by flight software to replay when time permits. In the event of multiple queued tones, the highest priority tones were played first. In the event of tones with the same priority, the most recent tones were played first, because the newest information during EDL was generally favored over stale information. Because of this software logic, the time between a tone event occurring and when it actually was radiated varied by several seconds, and some tones appeared to play out of order. Although this made real-time operations more complicated, it was the preferred strategy to enhance the probability of receiving indications of off-nominal behavior in the event of a fault…Parachute deploy and touchdown tones [were] carrier-only tone, where no subcarrier modulation is used.20
For transmitting telemetry during descent and landing, MSL used the Electra-Lite UHF radio within the rover, broadcasting to receivers on Mars orbiters. It transmitted through three different low-gain UHF antennas at different times: one on the parachute cone of the backshell, one mounted to the top of the descent stage, and finally the rover’s helix antenna, the one that it still uses for surface operations (see section 4.5). Amazingly, during entry, descent, and landing, the Parkes radio telescope back on Earth, in Australia, was able to pick up the signal from MSL’s UHF antennas. While the signal was too weak for Parkes to decode any telemetry, it did provide Doppler information as the spacecraft decelerated toward landing.
Telemetry arrived on Earth with the assistance of orbital relays. Odyssey provided the primary conduit for real-time communication. It served as a bent-pipe relay: it received MSL’s signals, demodulated them, and immediately sent the decoded telemetry to Earth. The Odyssey relay arrived only seconds slower than the direct-to-Earth tones. Mars Reconnaissance Orbiter recorded MSL’s UHF signals in an “open-loop” mode, without demodulating them. This recording wasn’t available on Earth until hours after landing, but would have provided more data if an anomaly happened that caused Odyssey to lose lock on the signal. The European Space Agency’s Mars Express also listened for MSL’s signal. It operated in an open-loop carrier-only detection mode, which didn’t record telemetry but provided an alternate angle for Doppler tracking relative to that recorded from Earth.
2.3.2 The aeroshell and MEDLI
MSL’s was the most challenging Mars atmospheric entry in history, for two main reasons. MSL was dramatically larger than any previous landed Mars spacecraft, and its goal was a much more precise landing than previously attempted (Figure 2.8). The aeroshell was 4.5 meters in diameter, the heat shield a 70° cone (Figure 2.9). The heat shield’s shape was the same as for all previous Mars landers, but Curiosity’s aeroshell was a meter larger and more than three times heavier than any previous one.21 In a throwback to Viking, the aeroshell was able to generate lift. Parts of the backshell and heat shield are labeled in Figure 2.10 and Figure 2.11. The backshell with parachute and balance masses weighed 576.6 kilograms.22 The heat shie
ld weighed 440.7 kilograms.
Figure 2.8. Comparison of NASA Mars aeroshells. Emily Lakdawalla after Edquist et al (2009) and Wallace ( 2012 ).
Figure 2.9. Top: Aeroshell dimensions. The aeroshell consists of the backshell and heat shield. Bottom: geometry of the aeroshell during guided flight. Images: NASA/JPL-Caltech. Top diagram based on Karlgaard et al ( 2014 ). Bottom based on Steltzner et al ( 2010 ).
Figure 2.10. Parts of the MSL backshell. NASA/KSC image releases KSC-2011-4526 and KSC-2011-7183, annotated by Emily Lakdawalla.
Figure 2.11. Parts of the MSL heat shield. Top: exterior surface of the heat shield, with locations of the MEDLI Integrated Sensor Plugs marked. Bottom: interior surface of the heat shield. NASA/JPL-Caltech/Lockheed Martin image release PIA14128, taken at Lockheed Martin Space Systems, Denver, in April 2011, annotated by Emily Lakdawalla.
The heat shield gathered an unprecedented amount of information about the descent through the Martian atmosphere, thanks to the MSL Entry, Descent, and Landing Instrumentation (MEDLI) sensors embedded within it. MEDLI had two kinds of sensors. Seven transducers of the Mars Entry Atmospheric Data System (MEADS) measured atmospheric pressure by tiny 2.5-millimeter through-holes in the shield. Seven MEDLI Integrated Sensor Plugs (MISP) were embedded in the heatshield within 33-millimeter-diameter plugs. They consisted of thermocouples to measure temperature and recession sensors to document how the PICA material weathered entry. Locations of the MEDLI sensors are shown in Figure 2.11 and Figure 2.12.23
Figure 2.12. Locations of the MEDLI MEADS (orange) and MISP (white) sensors on the MSL heat shield. Flow lines show the direction of expected air flow. Based on Little et al ( 2013 ) and Beck et al ( 2010 ).
2.3.3 Final approach
Ten minutes before entry, at 5:00:46 Spacecraft Event Time on August 6, 2012, the cruise stage separated, its work complete. Later images from Mars Reconnaissance Orbiter HiRISE and CTX instruments show numerous impact craters from the cruise stage scattered over a strewnfield 12 kilometers long, indicating that the cruise stage – unprotected by an aeroshell – broke up in the atmosphere (Figure 2.13).24 The cruise stage took with it MSL’s star trackers. From that point on, the rover computer would maintain its sense of its own orientation by dead reckoning. MEDLI began to acquire data from the heat shield.
Figure 2.13. Impact sites of the cruise balance masses and fragments of the cruise stage. At about 4 meters in diameter, the two largest craters are probably the cruise balance mass impact sites. All the other, smaller impacts are likely from fragments of the cruise stage. HiRISE images ESP_029245_1755 and ESP_029601_1755. NASA/JPL-Caltech/UA.
Nine minutes prior to entry, the guidance, navigation, and control system activated, and the rover computer fed it the navigators’ best estimate of the spacecraft’s position and velocity. (Many publications about the landing refer to this moment, 540 seconds before entry, or 397501174.997338 seconds on the spacecraft clock, as “t 0 ” for the landing phase, while others use the moment of entry as the zero point.) The spacecraft stilled its rotation and oriented to the correct angle for hitting the top of the atmosphere. It ejected two 75-kilogram blocks of tungsten, the cruise balance masses, which went on to impact the surface close to the cruise stage (Figure 2.13). The sudden loss of 150 kilograms of mass offset the capsule’s center of mass away from its centerline. Once the capsule was in the atmosphere, this offset gave it a 16° angle of attack. The capsule was ready to fly in the Martian air.
MSL switched X-band antennas, now broadcasting tones from the tilted low-gain antenna, which was pointed 17.5° away from the aeroshell’s axis of symmetry (Figure 2.10). The switch of antennas caused only a very brief loss of communication with the spacecraft.
As MSL approached Mars, Mars Reconnaissance Orbiter approached the equator from the south, while Mars Odyssey approached from the north (Figure 2.14). Mars Reconnaissance Orbiter’s path took it across the westernmost rim of Gale crater, carrying it nearly overhead during landing, while Odyssey passed considerably to the east. That geometry would allow Odyssey to have a second communications pass with MSL later on landing day, passing to the west about two hours after landing.25 Mars Reconnaissance Orbiter began its “open-loop” recording of the MSL signal at 8 minutes 7 seconds before entry.26
Figure 2.14. Geometry of MSL and orbiter ground tracks during entry, descent, and landing. Base image is from Viking Orbiter; spacecraft positions retrieved from JPL Horizons. ODY = 2001 Mars Odyssey; MRO = Mars Reconnaissance Orbiter; LOS = loss of signal.
2.3.4 Entry: 0 to 259 seconds
MSL entered the Martian atmosphere at 05:10:46 at an altitude of 125 kilometers. Traveling at a relative speed of 5.8 kilometers per second, it shed all of that velocity within the next 7 minutes. Within one minute, it had plunged to only 40 kilometers’ altitude, broadcasting tones to keep Earth updated. Watching the X-band tones arrive on Earth, Allen Chen had a moment of sheer terror: a tone had arrived that indicated that the vehicle orientation was out of control, suggesting that the loss of the spacecraft could be imminent. Fortunately, it turned out to be a calibration issue with the MEADS sensors, not an actual anomaly, and the rest of the landing events happened as expected.27
At 46 seconds after entry, the descent stage inertial measurement unit had begun to sense the atmosphere as a drag force of 0.2 gees, beginning the range control phase of guided entry. This was earlier than expected, because the navigation team’s atmospheric model had overpredicted the temperatures there, underpredicting the pressures, although the pressures and temperatures that MSL measured were consistent with those reported by Mars Climate Sounder.28 The mismatch between prediction and reality had little effect on the landing process.
During the range control phase, the rover computer predicted the downrange distance it would fly and adjusted lift as necessary in order to shoot for the correct range. Unlike an airplane, MSL had no flaps or elevators to change its angle of attack, so the way that the spacecraft adjusted its range was to perform a series of banking turns, rotating its center of gravity around the axis of its blunt nose. Its initial entry point was biased to the left (north) of the intended landing site, so it began with a banking turn to the right. The computer monitored the spacecraft’s cross-range drift, and commanded a bank reversal when the drift passed a threshold. It reversed its bank angle to the left, then right, then left again. Figure 2.15 shows how the velocity, altitude, and bank angle varied with time. The first, commanded bank was at very nearly 90° (resulting in no lift being generated), so the spacecraft descended on an almost ballistic path. By the time of the first bank reversal, it had slowed dramatically and the spacecraft commanded less bank angle. The capsule truly began to fly in the Martian atmosphere.29
Figure 2.15. Best estimate of entry trajectory, based on spacecraft telemetry. Modified from Mendeck and Craig McGrew ( 2014 ).
All this time, the heat shield was doing its job. Initially, the spacecraft continued to lose altitude at a rate of a kilometer per second. The hypersonic entry pressurized the air in front of the capsule, creating a shock wave with temperatures as high as 4000 kelvins (Figure 2.16). At 65 seconds after atmospheric entry, the atmosphere had become thick enough that the flow of air across the heat shield abruptly transitioned from laminar (smooth) to turbulent.30 The heat shield’s temperature increased rapidly. At 85 seconds after atmospheric entry, the surface of the heat shield reached its peak temperature, of around 1300 kelvins (Figure 2.17). MEDLI data showed that peak heating happened at a different location and lower temperature than had been predicted during heat shield development, possibly because the flow of air over the heat shield became turbulent earlier than predicted. The PICA heat shield material withstood these forces easily, with little of it receding away: every single MEDLI thermocouple survived entry, even though some were installed just 2.54 millimeters underneath the surface.31
Figure 2.16. Artist’s concept of the MSL aeroshell creating a shock wave during entry. NASA/JPL-Caltech release PIA14835.
> Figure 2.17. MEDLI MISP temperature flight data (solid lines) compared to preflight predictions (dashed lines), from Bose et al ( 2014 ). Each MISP sensor has four thermocouples at different depths. The colorful pattern on the heat shield shows the preflight predictions. Peak heating actually occurred closer to sensor T7 near the center of the heat shield, not at the most leeward sensor T3 as predicted.
With every second of entry, the spacecraft flew into denser air. It reached peak deceleration 80 seconds after entry, the pressure of the air decelerating it at 12.5 gees. As the spacecraft began its first bank reversal, dropping below 20 kilometers altitude, those forces began to wane, and the flying saucer entered a period of nearly level flight for two full minutes. It flew with a tailwind of about 20 meters per second, but the spacecraft’s reckoning of its downrange target depended on an inertial measurement unit that wasn’t affected by the wind, and the spacecraft stayed on course. The final bank reversal left it with about 1 kilometer of downrange error, well within tolerances.32
At an altitude of 14 kilometers and speed of 1.1 kilometers per second, the spacecraft transitioned into the “heading alignment” phase of guided entry.33 The spacecraft banked left to correct its cross-range heading, probably to compensate for a 10-meter-per-second crosswind.34 It flew downrange for 100 seconds at a near-constant altitude, steering lightly to arrive at the optimal location for parachute deployment. It was during heading alignment, at 222 seconds after entry, when Mars Odyssey achieved lock on MSL’s UHF signal and began relaying telemetry directly to Earth at a rate of 8 kilobits per second through the Deep Space Network station in Canberra, Australia.35 Back on Earth, engineers applauded the news; the landing would occur just the same with or without Odyssey communications, but only Odyssey could give Earth real-time telemetry. “Real” time being 13.8 minutes after the events on Mars, thanks to the distance separating Mars and Earth.