Surveyor Soil Mechanics Surface Sampler

NASA. Public domain (NASA / US government work).

The problem

A lander arm that tests soil has to act on a surface of unknown strength with a mechanism whose loads can be reconstructed, from a fixed base, through a command link that allows one motion at a time followed by a television picture. The environment adds no atmosphere to carry heat and a lunar night that the hardware must survive or be lost [9], [10]. Before Surveyor, strength estimates were passive: footpad penetration, crushable block imprints and the tracks of rolling fragments [11], plus optical, thermal, radio and radar inference of a low-density, highly porous surface of under 1 g/cm3 that Surveyor 1 footpad data and television particle counts overturned [11], [8].

What has flown

The Soil Mechanics Surface Sampler was a motorized scoop on an extending arm, flown on Surveyor 3 and Surveyor 7 and operated successfully on both [3], [10]. The program flew five successful landers between June 1966 and January 1968, and only these two carried the sampler; Surveyor 4 was lost before landing [3], [12], [13].

Surveyor 3 (20 April 1967). The lander came to rest tilted 12.5 degrees in the southeast of Oceanus Procellarum, a mare site [3]. The sampler ran for 18 h 22 min and performed 8 bearing tests, 14 impact tests, 4 trenches and the handling and dumping of material onto a footpad [9]. The touchdown telemetry anomaly invalidated the motor-current channel, so television was the only source and forces were bounded by deliberately stalling motors against preflight calibration [11]. The spacecraft never responded after its first lunar night, with compartments cooled to 0 F.

Surveyor 7 (10 January 1968). The lander sat on the ejecta blanket north of Tycho, in the southern highlands [10]. The mechanism was identical to the Surveyor 3 unit except for two horseshoe magnets in the scoop door bearing plate, a 30 degree rotation of the mounting, and an interface that let the sampler pick up the deployed alpha-scattering sensor head [3]. The current was sampled at a much higher rate across several commutator frames, which resolved the current pulse and gave a more accurate force [10]. The motors operated normally from 180 F down to minus 167 F after sunset, and the mechanism survived a lunar night across two lunar days. When the alpha-scattering head jammed in its background position, the sampler released it and relocated it to a rock and later to subsurface material, the first recovery of a failed instrument deployment by a robotic arm on another world. A rock was weighed at a density between 2.4 and 3.1 g/cm3 and another was broken by a moderately hard blow [10].

Specifications.

Parameter	Value	Source
Configuration	scoop on a lazy-tongs extendable linkage with elevation, azimuth, extension and door axes, each driven by a small DC motor commanded by timed pulses	[3]
Tong extension	64 in (1.63 m) with 54 degrees of vertical and 112 degrees of azimuth motion, as reported for Surveyor 3; 148 cm on the Surveyor 7 calibration	[9], [10]
Mass	20.05 lb on Surveyor 3, 21.20 lb on Surveyor 7	[3]
Force measurement	none fitted; applied force recovered from motor current	[3], [10]
Position measurement	position potentiometers, strain gages and accelerometers deleted	[3]

The force-measuring instruments and potentiometers were deleted so that the sampler could replace the approach television camera on an unmodified spacecraft; two flight units were authorized in August 1966 and a spare was added to Surveyor VII in May 1967 [3]. The Hughes-built unit had no onboard computation: motion used timed motor pulses, a single motion needed several commands, and television could not be taken while the arm moved [3], [9], [10]. The two reach figures differ, 1.63 m against 1.48 m, and the sources do not reconcile them [9], [10].

Later lunar samplers and soil probes. The Apollo crews used a self-recording penetrometer, core tubes, a drill and the Lunar Roving Vehicle as soil probes, and the S-200 synthesis from those data replaced the Surveyor values as the reference model [5], [6]. Apollo 15 track depth averaged about 1.25 cm at a wheel load of about 290 N, and a trafficability parameter set with 35 degree friction reproduced measured rover energy consumption on Apollo 15, 16 and 17 [5], [6]. Chang’e-5 and Chang’e-6 scooped and returned samples with a four-degree-of-freedom arm [17], [18]. Lunar PlanetVac, a pneumatic sampler, acquired regolith on the Firefly Blue Ghost lander at Mare Crisium on 3 March 2025 in a five-second sequence, and the head sat in an off-nominal position [14], [15]. Yutu-2 gave the first rover-derived terramechanical characterization of farside regolith in 2024 [26].

What is being tried now

Soil properties from returned and in situ material (flown, analyzed on the ground). The Chang’e-6 scooped farside regolith has D60 of 48.4 um, volume-averaged sphericity of 0.58 against 0.69 for Apollo 11 and 14, and static and dynamic angles of repose of 52.9 and 70.4 degrees, above any Chang’e-5 or Apollo simulant measured, so it is markedly more cohesive than nearside soil [16]. Physical, mineralogical and geochemical characterization of those samples was compared against Chang’e-5 and Apollo material [18]. From the Yutu-2 wheel tracks, sinkage averaged about 8 mm, bearing capacity was near 4 kPa against 2 to 3 kPa along the Lunokhod-1 traverse, and cohesion was 520 to 3154 Pa with friction angle 21.5 to 42.0 degrees [26]. These replace an inference from one scoop with inversions from wheel motion and returned grains.

Autonomous sampling (flown). Chang’e-6 scooping selected sample points from a binocular point cloud against slope, relief and obstacle criteria, and replaced the manual fine-tuning used on Chang’e-5 with a 6D pose estimate by point cloud registration [17]. The four-degree-of-freedom arm alone had about 30 mm and 2 degrees of absolute end-effector error from flexible deformation, and soil sink under the touching disc of 5 to 20 mm was compensated separately [17]. Surveyor had no such loop: every step was an Earth-evaluated television picture [10].

Dedicated in situ strength instruments (ground test). The SPARTA toolkit combines a vane shear tester, cone penetrometer, thermal conductivity probe and dielectric probe, and has completed three phases of testing [22]. It fills the role the Surveyor sampler filled by improvisation, with force measured directly.

Strength in reduced gravity (ground test). Static cone penetration at 1/6 g, 1 g and 2 g on a magnetic levitation rig found that resistance at 1/6 g is about 29.1 percent below 1 g at 40 percent relative density but only about 8.8 percent below at 76 percent, so the lunar reduction at depth is likely smaller than assumed [19]. A drop tower shear and compression cell was built to obtain cohesion, friction angle and tensile strength at lunar and small-body acceleration [20]. Granular flow in a drop tower jammed at 0.19 g where it flowed freely at 1 g, because cohesion becomes comparable to weight [32]. Gravity offload by mass reduction gives overoptimistic trafficability because the grains are still under Earth gravity, and single-wheel drawbar pull measured in parabolic flight shows reduced-weight testing overstates mobility [33].

Simulants and databases (ground test). Cohesion of simulants grows exponentially with bulk density and spans 0.185 to 1.672 kPa for lunar types [21]. Particle size distribution alone does not set fidelity: simulants within one standard deviation of the lunar average differ substantially in shear strength, and the selected Mines testbed simulant has cohesion about 355 percent above the highland estimate [23]. NASA’s guide scores fourteen simulants and concludes no simulant reproduces every aspect of lunar regolith [24]. An open database consolidates lunar and simulant properties by measurement method [25]. Facility simulants are characterized for abrasion and density [29], a dust-suppressed cohesive simulant exists [31], and a DEM scaling study of a grousered wheel in FJS-1 simulant targets slip, sinkage and power at lunar gravity from Earth-gravity runs [30].

Simulation and gravity scaling (ground test). A smoothed-particle continuum model inside the Chrono multibody simulator reproduces discrete-element and physical single-wheel tractive force and torque curves at lower cost and was applied to a six-wheel Yutu model under lunar gravity [28]. Vertical vibration kept lunar simulant flowing through hoppers at lunar gravity in vacuum across 60 parabolas, where it otherwise arches and stops [34]. None of these has flown. The same boulder-track bearing technique that Apollo used has been applied to asteroid Didymos, giving a mean bearing capacity of 70 N/m2 [27].

The numbers

Surveyor bearing strength profile.

Depth	Bearing strength	Evidence	Source
Uppermost millimeter	less than 0.1 N/cm2	imprints of small rolling fragments	[11]
1 to 2 mm	0.2 N/cm2	alpha-scattering sensor head imprint	[11]
1 to 2 cm	1.8 N/cm2	crushable block imprints	[11]
averaged to 6 cm	4 to 6 N/cm2	footpad penetration	[11]

Strength rises by nearly two orders of magnitude in the first few centimeters [11]. Footpad-derived static bearing strength at the Surveyor 7 site averaged 3.4 N/cm2, similar to Surveyors 1, 3 and 5 [10].

Soil parameters by source.

Quantity	Value	Source
Surveyor cohesion	0.035 to 0.05 N/cm2	[11]
Surveyor friction angle	35 to 37 degrees	[11]
Surveyor density	about 1.5 g/cm3, porosity 0.4 to 0.5	[11]
Surveyor 7 Tycho soil	37 to 39 degrees friction, about 1.5 g/cm3	[10]
Apollo density, 0 to 15 cm	1.50 g/cm3	[6]
Apollo density, 30 to 60 cm	1.74 g/cm3	[6]
Apollo friction angle	35 to 50 degrees	[6]
Apollo cohesion	0.1 to 1.0 kN/m2	[6]
Reference shear, 0 to 15 cm	42 degrees, 0.52 kPa	[24]
Reference shear, 30 to 60 cm	54 degrees, 3.0 kPa	[24]
Compiled database range	density 0.7 to 2.3 g/cm3, friction 20 to 55 degrees, cohesion up to about 5 kPa	[25]
Yutu-2 cohesion	520 to 3154 Pa	[26]

Surface debris from lander television was fitted to cumulative particle counts from 1 mm to 10 cm and beyond, with a stated factor of two on K and 0.2 on the exponent, fits made by eye to preliminary data [8]. The surface at 45 degrees latitude runs from a mean of 350 K at noon to 89 K before sunrise [2]. Mare and highland sites differ in failure mode rather than strength: Surveyor 3 soil cracked into relatively large chunks, Surveyor 7 material was more deformable and less brittle, and soil depth near Surveyor 7 varied from 1 to at least several inches over rock [10], [11].

What is not established

The Surveyor 3 forces rest on image interpretation and stall calibration, not on a recorded current history, and the program report presents bearing and cohesion values as brackets rather than values with uncertainties [9], [11]. Every Surveyor number is inferred from an action not designed as a soil test, so it carries the assumptions of its reduction model [11]. Reach figures are not reconciled. The Apollo model’s density and porosity conversions rest on few specific-gravity and void-ratio measurements and its trafficability parameters were fitted to energy consumption, so agreement with Surveyor is a consistency check, not a validation [6]. Farside soil is more cohesive than any nearside sample, so the nearside values do not transfer [16], [26]. The reduced-gravity effect on strength depends on relative density and is not settled for the lunar subsurface [19]. All in situ data are from equatorial and mid-latitude sites; polar regolith strength has no in situ reference measurement.

References