NASA JPL Space Simulators

NASA/JPL-Caltech. Public domain (NASA / US government work).

The problem

A robotic arm, rover or rotorcraft bound for Mars or deep space must survive combined vacuum, solar heating and cold that no single Earth environment produces on its own, and the qualification article is usually the vehicle that flies. NASA JPL’s environmental test organization answers with two thermal vacuum chambers built decades apart. The 25-Foot Space Simulator in Building 150 is a 1961 chamber with off-axis xenon solar simulation, built to reproduce cruise heating and deep-space cold on a whole spacecraft [1][5]. The 10-Foot Vertical Space Simulator has a hydraulic-lift floor so that a tightly packed configuration of rocks, tools and an arm stays reachable at Mars pressure and temperature, and it has no solar simulation [2]. Both hold Mars conditions on flight hardware that cannot drive; driving is done outdoors at the MarsYard, which holds neither pressure nor temperature [37]. The 25-Foot chamber is NASA-owned and available to other government agencies and to private industry [1][2].

The two designs split because accurate solar heating and full mechanical access are hard to build into one machine. The solar simulator was the harder problem from the start: the operator’s 1966 account of the first two years of use, testing Ranger, Mariner and Surveyor hardware, spends most of its length on the light source, and names it the largest source of trouble, above outgassed cable insulation contaminating test items and a stopped mechanical pump that let diffusion pumps backstream oil onto a spacecraft and arc its electronics [5]. A 1964 to 1965 spectral characterization of the beam from 133 mercury-xenon arc lamps found it two to three times too strong in some visible bands and only a third of solar intensity in others, although the integrated near-infrared matched the solar spectrum to within half a percent [6]. An absorptance computed against the real solar spectrum is therefore wrong for a coating whose absorption concentrates in the mismatched bands, so thermal balance plans are built around the beam JPL has. The optical chain (xenon lamps, integrating lens unit, large parabolic collimating mirror) also constrains upkeep: the mirror was pulled, stripped, re-plated and re-aluminized inside the chamber in 1994, using the chamber as the coating vessel [7].

What has flown

Ranger, Mariner and Surveyor, 1963 to 1964. The 25-Foot chamber’s first two years ran a standard progression from solar panel to temperature control model to proof test model to flight spacecraft [5]. They produced the contamination and backstreaming incidents above and a beam map of 184 W/ft2 nominal with a 6 percent spread over the central 4 ft. The real Sun’s angular divergence of plus or minus 0.267 degrees against a measured simulator divergence of plus or minus 5.3 degrees forced test engineers to redesign sun shades and shadowing hardware [5].

Mars Exploration Rover, November 2002 to April 2003. Both flight spacecraft went through cruise thermal-vacuum testing in the 25-Foot chamber, suspended on steel cables inside the aeroshell, under the xenon solar simulator on the cruise array and quartz lamp arrays on off-sun surfaces, across eleven thermal cases per spacecraft including an entry-descent-landing case with the Heat Rejection System off [9]. Each rover then went to the 10-Foot chamber for an 8 torr nitrogen surface thermal test with no solar simulation, standing up and deploying through hot and cold diurnal cycles. Margins held in every case; the tightest was a 3.8 C battery cell-to-cell gradient against a 5 C requirement [9]. The one new technique was a controlled re-pressurization of the 25-Foot chamber from 1e-1 Torr to 8 Torr in six minutes to simulate telecom operation during descent, never attempted there before [8].

Phoenix, 2008. The arm flew without a chamber test of the combination that later failed it [4]. Insolation on ice-bearing regolith in the aluminum scoop warmed the sample enough that it adhered to the scoop wall and would not fall through the TEGA inlet on the first two attempts; a third sample obtained by scraping, with less ice, did not congeal and was delivered [4]. The University of Arizona Payload Interoperability Testbed reproduced the delivery geometry closely enough that overlaid Robotic Arm Camera images from Mars and from the testbed were used to correct scoop positioning, but not the thermal combination.

Mars Science Laboratory, 2009 and 2011. The January 2009 launch [18] and cruise solar thermal vacuum test in Building 150 was prepared with rehearsed failures, called Gremlin Sessions, injected over the comm net to both shifts. A loss-of-building-power scenario exposed a faulty transfer switch and mis-assigned UPS circuits before flight hardware entered the chamber, and several rehearsed failures recurred during the 5.5 day test: a magnitude 5.0 earthquake during final pumpdown, loss of commanding, solar simulator lamp coolant leaks and a flight computer reset [18]. In 2011 the 900 kg Curiosity rover ran 16 days in simulated cruise under vacuum and in simulated Mars surface conditions under 8 torr nitrogen at mission-extreme hot and cold boundaries [3]. All thermal hardware performed nominally and the Rover Heat Rejection System performed better than predicted; the data correlated the thermal models that predicted Gale Crater performance across the 669 sol surface mission [3].

Mars Helicopter, 2018 and after. The Ingenuity flight model went through a 40 C bakeout, steady state thermal balance in vacuum and at Mars pressure with gaseous-nitrogen backfill, a -90 C cold cycle, and cold functional testing at -50 C shroud including blade spin-ups at 50 and 1800 rpm [11]. Gas conduction across internal gaps, not radiation, was the largest driver of survival energy at Mars pressure; the fix was geometric, enlarging the fuselage by 25 mm in one direction and 15 mm in another, plus a low emissivity inner blanket. The tested article carried ground support cabling that added 50 to 60 percent more heat loss than flight, so the survival energy budget closes only in the correlated model [11].

Mars 2020, 2016 to 2021. A 2016 instrument test in the 25-Foot chamber landed midway through a control-system replacement, so staff ran pump-down, cooldown and CO2 backfill by hand against an 81-step procedure with analog gauges in every cabinet and field operators on noise-canceling radios, completing 38 runs; redundant Hand/Off/Auto switches are now standard on refurbished chamber control systems [10]. The May 2019 system thermal vacuum test ran six days (cut short by an anomaly) on a surrogate chassis and a qualification MMRTG in place of the unavailable flight rover. It validated cruise thermal design and heat rejection system leak tightness, found reversed thermostat wiring on one instrument and reversed catalyst-bed PRT wiring on landing engines 3 through 8, and showed that a build-to-print rebuild of an MSL-heritage design still shifted zone temperature ranges by 2 to 13 C [14]. The 10-Foot chamber then hosted a two year sampling and caching qualification series at about 7 torr and -70 C, nominally -65 C cold with stress cases to -115 C, on a floor of geoanalogue rocks including deliberately chosen challenge rocks [2]. Six qualification coring bits took 66 cores, one abrading bit made 34 patches, one regolith bit collected regolith twice and 17 seals were activated; flight software itself, not a test build, ran in the test environment [2].

What is being tried now

Every approach below has been exercised on the ground; none is a flight demonstration of a chamber technique.

Closing the combined-environment gap by splitting cases. The Mars Helicopter reference states that even JPL’s largest chamber cannot combine Mars air density, Mars-relevant temperature and true 1/3 g, so free-flight tests ran at Mars density and ambient temperature with a gravity offload rig, and thermal qualification ran density and temperature cases separately from the flight dynamics cases [12]. Test status: ground-tested and flown as qualification practice [11][12].

Rehearsed failure and manual fallback. Gremlin Sessions in the 2009 MSL test [18] and the 2016 hand-operated campaign [10] treat the facility and its crew as part of the system under test. Status: ground practice, with the transferable result being the switch standard in [10].

Photogrammetry under solar and cryo-vacuum load. In September 2008 a remotely actuated camera on a mobility cart measured thermo-elastic deformation of a 5 m deployable reflector to 0.025 mm RMS over the aperture across a 36-hour solar cycle test, which the program credited with raising the reflector’s deformation analysis from TRL 3 to TRL 6 [13]. Status: ground-tested on one article; ice forming on targets during cold soak was the one failure mode [13].

Simulation-first arm and drill validation. Target rocks are scanned into a point cloud and loaded into the Rover Sequencing and Visualization Program Hyperdrive simulator, and arm and drilling operations are validated in simulation before execution in the chamber [2]. Because the limited chamber space kept flight-like cameras from the rover’s true positions, the 3D environment model came from a fixed CAD model of the chamber plus rocks on keyed baseplates [2]. Status: ground-tested over the 2019 to 2021 campaign.

A ladder of cheaper venues beneath the chamber. The Mars 2020 testbed team ranks venues on hardware, avionics and flight software, and environmental fidelity: the Hexapod tilts a fixed chassis without rover kinematics, Scarecrow matches rover weight but not flight hardware, and the Vehicle System Testbed twin, OPTIMISM, covers system-level verification, dry runs, operations training and anomaly debug [20]. JPL’s Guidance and Control Spacecraft Autonomy Testbed combines a Simulink and F’ software testbed with hardware-in-the-loop platforms in the Robodome [35]. Mobility is established outside the chambers: the MER variable terrain tilt platform, a 16 foot square table run over 200 trials, gave slip that rose nonlinearly above about 10 degrees and reached 91 percent driving directly up a 20 degree slope on sand, and drives at Meridiani planned against those curves landed within 10 percent of the destination [21]. The outdoor MarsYard III covers 66 m by 36 m with 7 variable angle and cohesion slopes [37], and the Planetary Robotics Laboratory offers about 139 m2 of floor plus two indoor sand pits [22]. NASA’s consolidated best practices for planetary rover and tire mobility testing cover drawbar pull, slope and obstacle tests and list the absence of standard south pole simulants among open gaps [34]. Status: all in routine ground use.

Chamber repurposing. In twelve months the 10-Foot chamber was converted to an ion engine test chamber for the Prometheus/JIMO program, with six nude cryopumps rated 57,000 L/s of nitrogen each, a graphite liner and water-cooled shroud circuits, and was later converted back [19]. Arcing at a 10 kV feedthrough, floor shroud leaks and window coating were logged, and more than 20 shroud leaks were attributed to chlorine in supplemental city water attacking welds in shrouds that had run 40 years [19]. No achieved vacuum performance is reported.

Chambers and rigs at other centers. ESA’s Large Space Simulator at ESTEC is 15 m high by 10 m wide, about 2,300 m3, with a horizontal auxiliary chamber for the solar collimation optics [25]; its published envelope is a typical 5e-6 mbar (3e-7 mbar best), shrouds below 100 K in liquid nitrogen mode at 170 kW capacity or 150 to 350 K in gaseous mode, and a 6 m beam adjustable from 70 to 2600 W/m2 at plus or minus 4 percent [31]. It sits in a 6000 m2 cleanroom complex with Phenix and VTC 1.5 thermal vacuum chambers, shakers and an acoustic facility [32]. The ESTEC Mars Yard is an 8 by 8 m indoor bed of sand, gravel and rocks for locomotion, navigation and arm placement [26]. NASA Glenn built the Glenn Extreme Environment Rig, a 3 ft by 4 ft chamber of over 28 cubic feet intended for roughly 500 C and 100 bar with nine gas streams, because existing Venus chambers held about 1 cubic foot and reproduced only CO2 and N2; the paper predates commissioning [27], and a later briefing reports over 90 days of cumulative operation at Venus conditions without test data [28]. Glenn’s Planetary Exploration facilities add soil tanks, a single-wheel dynamometer, an excavation lab and the VOiD thermal-vacuum dust chamber [33]. The Lunar Dust Adhesion Bell Jar stirs simulant during pumpdown, bakes at 200 C at 1e-6 torr, thermal cycles to about -150 C and plasma-treats the surface, with no demonstration that the result resembles lunar regolith [29]. For entry heating, the Ames Thermophysics Facilities Branch operates arc jets, a ballistic range and an electric arc shock tube [30], with arc jet power of 10 to 60 MW and surface pressure of 0.01 to 75 kPa [36]. NASA’s Johnson Space Center Chambers A and B, built for Apollo and Gemini on a comparable scale, serve crewed rather than robotic programs [15]. JPL’s 1990s telerobotics and rover overviews describe tools these chambers later qualified [17]. A survey of flown and proposed lunar and Mars rovers compiles the shared temperature, vacuum and regolith envelope independently of any facility [16]. Radiation qualification is separate: two Cobalt-60 irradiators span 0.001 to 1 and 0.1 to 500 rad(Si)/s [23], and the Guidance and Control Section describes in-house testbeds as a capability overview [24].

The numbers

Parameter	Value
Operator	NASA JPL environmental test organization [1]
Location	Pasadena, California, United States
Commissioned	25-Foot Space Simulator, 1961; 10-Foot chamber date not published [1]
Type	Thermal vacuum chambers, one with off-axis solar simulation [1][2]
Simulant or terrain	No simulant bed. Geoanalogue rocks on keyed baseplates [2]
Instrumentation	600 thermocouple channels and 40 RTD channels to the article [1]
Ground truth	Kendall cavity radiometer on the solar beam; six ion gauges [1]
Fidelity limits	Chamber conditions are hard to vary between tests and are held fixed [2]
Access	Available to other government agencies and to private industry [1]
Cited by	Curiosity [3], Phoenix arm [4]

25-Foot Space Simulator [1]

Parameter	Value
Working volume	6.1 m diameter x 7.6 m high with the solar simulator; about 21.3 m high without
Test article limits	Wall hard points rated 10,000 lb each; door 4.6 x 7.6 m
Vacuum	5 x 10^-5 torr in about 3 hours, into the 1 x 10^-6 torr range on cryopumps
Temperature	-185 C on liquid nitrogen; -125 to +100 C on gaseous nitrogen
Illumination	2.0 to 11.0 solar constants, beam 8.5 to 18.5 ft, 37 xenon arc lamps
Gravity offload	None. Articles are hard mounted or suspended at 1 g
Instrumentation	600 thermocouple, 40 RTD and 60 heater control channels to the article

The chamber is 25.9 m high and 8.2 m in diameter [1]. Access is at ground level through a 4.6 by 7.6 m door, and articles enter on a monorail crane [1]. Support is either hard mounting to stanchions through the floor shroud onto an isolated seismic mass, or suspension from wall hard points [1]. An earlier facility description gives a 7.6 m diameter by 25.7 m tall chamber and a 5.8 m collimated beam to 2.3 solar constants at plus or minus 4 percent uniformity [8]. Pumping runs in four stages, with cryopumps and turbopumps taking it into the 1 x 10^-6 torr range [1]; a 2016 account gives 1e-6 Torr in about 90 minutes [10], consistent with pumpdown depending on loading and outgassing. A 2008 test reached about 1e-7 torr after facility preparation [13]. Mars surface conditions are a gas fill: about 8 torr of nitrogen for the MSL rover test [3], and 8 to 10 torr with shrouds between 20 C and -130 C for the MER surface test run in the 10-Foot chamber [8][9].

Thirty-seven xenon arc lamps feed an integrating lens unit, through a fused quartz window and off a 23 ft collimating mirror [1]. Two lens units and two mirrors give four beam combinations spanning 8.5 to 18.5 ft and 2.0 to 11.0 solar constants. The 18.5 ft beam at 2.0 solar constants has a 1.1 degree collimation half angle with 96.2 percent of energy inside 1 degree and uniformity of +0.6 to -1.4 percent; the 8.5 ft beam at 11.0 solar constants has a 2 degree half angle and plus or minus 5 percent [1]. The mirror was re-plated to 381 micrometers of electroless nickel, polished to 64 percent reflectivity and re-aluminized to about 89 percent after conditioning [7]. Facility instrumentation adds 140 thermocouple channels on shrouds, mirror and structure and three temperature-controlled quartz crystal microbalances [1].

10-Foot Vertical Space Simulator [2]

Parameter	Value
Vacuum	About 7 torr for Mars surface work; base pressure not published
Temperature	Nominal cold testing at -65 C, stress cases to -115 C
Illumination	No solar simulator
Dimensions	3.0 m by 13.7 m per the earlier description [8]
Layout	End bell on a hydraulic lift: first-floor level for access, second-floor level when sealed

Comparison with other chambers

Facility	Figure	Marker
JPL 25-Foot, solar beam	2.0 to 11.0 solar constants	[1]
ESTEC Large Space Simulator, solar beam	70 to 2600 W/m2, 6 m	[31]
ESTEC Large Space Simulator, shroud	below 100 K, 170 kW (LN2)	[31]
Ames arc jet complex, power	10 to 60 MW	[36]
MarsYard III, area	66 m by 36 m, 7 slopes	[37]
JPL Radiation Effects Group, Co-60	0.001 to 1 and 0.1 to 500 rad(Si)/s	[23]

What is not established

The 1964 to 1965 spectral mismatch has not been shown superseded by a later remeasurement; the beam has been adjusted in intensity since (dimmed toward Mars solar constants without a reported change in relative spectral emission) but not requalified against the solar spectrum band by band [6]. Floor area and base pressure for the 10-Foot chamber are not published in any cited source, only the approximately 7 torr Mars operating point [2]. Chamber conditions were difficult to vary from test to test in the Mars 2020 campaign, so temperature and pressure for a configuration were generally held unchanged, and the test plan was built as a design of experiments around that [2]. Neither chamber offloads gravity or offers a driving bed [1][2]. No cited source describes a combined test holding Mars pressure, Mars-relevant temperature and representative solar flux on a moving tool; every campaign trades at least one away, the gap the Phoenix icy-soil delivery anomaly fell into [4][11]. The ESA Large Space Simulator description is qualitative for solar and vacuum envelope beyond the figures in [31], and the Glenn Venus rig and arc jet entries are capability statements, not campaign results [27][36].