Viking Surface Sampler

Source: brewbooks, via Wikimedia Commons [13]; CC BY-SA 2.0.

Overview

A stationary lander on Mars cannot drive to a sample, and a lightbulb-length radio delay rules out driving a manipulator from Earth in real time. Whatever reached the surface material had to be commanded once, execute a whole acquisition and delivery cycle on its own, and survive doing so with no operator watching until the telemetry came back. The Viking Surface Sampler Subsystem was the first machine built to meet that problem and, in doing so, became the first machine to acquire and process extraterrestrial samples for onboard analysis on another planet [8], [1], [13]. Built by Martin Marietta Aerospace under NASA LaRC, one flew on each of Viking Lander 1 and Viking Lander 2, landing in 1976 [8], [9].

The subsystem is not a manipulator alone. It comprises the acquisition assembly, a furlable articulated boom carrying a collector head; the biology processing and distribution assembly; the gas chromatograph mass spectrometer processing and distribution assembly; and the electronics control assembly [10]. It interfaces with the biology instrument, the GCMS and the X-ray fluorescence spectrometer, and it also carries the hardware for the physical properties and magnetic properties investigations: mirrors on the side of the acquisition assembly, magnet arrays on the collector head backhoe, a magnet cleaning brush and a magnet magnification mirror.

Specifications

Parameter	Value	Source
Boom type	furlable tube boom, flattened on the drum, forms a column when extended	[12]
Boom stowed length on drum	about 12 ft (3.7 m) wound on a 6 in (152 mm) drum
Boom extension	10 ft (3.0 m)	[9]
Acquisition assembly envelope	about 14 in high by 8.5 in wide by 24 in long (0.36 by 0.22 by 0.61 m)	[12]
Acquisition assembly mass	about 28 lb (12.7 kg)
Primary sample field	120 degree arc, 3 ft minimum to 10 ft maximum radius, about 90 sq ft (8.4 m2)
Collector head	clamshell, motor-rotated lower jaw housing, solenoid-actuated upper jaw	[12]
Head rotation	180 degrees, motor driven
Jaw vibration	solenoid driven by an 8.8 Hz square wave
Collector head sieve	2000 micrometer holes in the lid, with disaggregation teeth inside
Scooping force	collector head extended forward with 30 lbf (133 N)
Backhoe furrow	about 3 in wide by 1 ft long (76 mm by 0.3 m)
Backhoe instrumentation	two samarium-cobalt magnet arrays plus a magnifying mirror
Jaw instrumentation	temperature probe on the lower jaw surface
Boom position sensing	potentiometers inside the boom, 9-bit commanded position word
Command interface	16-bit digital commands from the lander guidance, control and sequencing computer

Mission profile

Parameter	Value	Source
Landers carrying the subsystem	2	[9]
Planned deliveries, 90 sol mission	4 to the biology PDA, 3 to the GCMS PDA	[12]
Viking Lander 1 first deliveries	sol 8, to biology, GCMS and XRFS	[9]
Boom anomalies, Viking Lander 1	2
Boom anomalies, Viking Lander 2	1

Acquisition assembly

About 12 ft of furlable boom is wound on a single 6 in diameter drum [12]. The element is forced flat as it is wound and forms a structural column when driven out through guides [12]. On the flight vehicle the boom extends 10 ft from the lander [9]. The collector head at the tip was designed to dig or scoop. The rotation motor sits in a cylindrical section immediately behind the jaw; the lower jaw housing is the largest single piece and the primary structure; the upper jaw is opened, closed and vibrated by a solenoid attached to the housing directly behind it.

The vibration mechanism is dual purpose and self-diagnosing. The solenoid is driven by an 8.8 Hz square wave [12], and a switch adjacent to the jaw monitors jaw movement and chops the drive signal, which produces a short sharp vibration suited to moving material. If the jaw hangs open because of excess load or material sticking, the switch does not chop and the full square wave is applied, giving much more violent jaw motion [12]. The backhoe on the reverse face was designed to acquire subsurface material and can also cut trenches, which were used to study wind effects on soil and angles of repose. It carries two samarium-cobalt magnet arrays viewable directly from behind by the lander cameras, and a magnifying mirror giving a close view of the front of the arrays. A temperature probe on the lower jaw surface records jaw temperature, and a grid pattern painted on the lander top deck provides a reference against which deposited material can be assessed for color, size, angle of repose and wind action.

Deployment and acquisition sequence

Deployment is a fixed opening sequence. Pyrotechnics release the covers on both processing and distribution assemblies and the wind spoilers deploy automatically. The furlable boom is then extended about 1.5 in to release the boom mechanism from its support post, the mechanism elevates about 45 degrees and rotates about 180 degrees in azimuth, the collector head rotates 45 degrees inside its protective shroud, and the boom extends approximately a further 4 in, which mechanically releases the shroud so that four springs propel it about eight feet clear through the thin atmosphere [12].

A surface acquisition then runs as a strictly serial sequence: azimuth rotation to a predetermined position, elevation to approximately horizontal, extension to about nine feet. Each motion is performed singly rather than simultaneously, which the designers adopted to improve system accuracy and increase safety [12]. The boom then elevates down until the ground contact switch in the collector head gimbal actuates, the jaw opens, and the head is driven forward with 30 lbf, filling the scoop. For a subsurface sample the boom retracts a short distance on ground contact [12], the backhoe is positioned and cuts a furrow about 3 in wide and a foot long, and the head then opens and moves forward to take material from the bottom of the furrow. The reachable field on the flight landers was about 130 sq ft in front of and between legs 2 and 3 [9]. Delivery inverts the head so that material lodged in recesses falls out, then rotates in azimuth over the target processing assembly, extends to center over its screen, and lowers to a height that allows half an inch of head vibration while remaining shielded from the wind by the spoilers atop the processor. The head is then vibrated [12]: the lid is pulsed in short strokes at 8.8 Hz, sharp teeth inside the cover break up loosely bound clods, and the 2000 micrometer holes sift material onto the primary screen. A level sensor in the metering tube reports whether enough material has arrived; if not, the head vibrates again and the level sensor rechecks after time is allowed for grinding and sieving.

Sample processing

The GCMS processing and distribution assembly is a grinder and metering chain. An agitator, a spring wire shaped like a question mark, rubs the underside of the 2000 micrometer primary screen while its stem runs down the funnel into the grinder, where a lobe on the grinder agitates it, keeping material flowing and preventing compaction in the funnel [10]. Material passes into a comminutor that acts as an auger and grinder combined; free to position itself within its housing, its eccentric action grinds while the flutes reject material, in the manner of a drill forcing chips away from the cutting face, so material circulates until it is fine enough to fall through the 600 micrometer gap around the periphery of the auger. It then falls into a plenum chamber and is stirred over a 300 micrometer sieve covering a 1 cm3 metering tube [10]. When the level sensor reports the tube full, the tube is translated in a sliding shuttle to transfer material to the GCMS, then to a dump position where the shuttle vibrator runs and the comminutor is reversed; a linear cam lowers the auger to open the gap to 4000 micrometers so the flutes force oversized material through the dump tube into a self-contained dump box.

The biology assembly is functionally similar but has no grinder. It uses material directly as sifted through a 1000 micrometer primary screen, processes 7 cm3, is unpressurized, and breathes through a bio-filter in its upper cover, with its own integral dump box [10].

Organic cleanliness drove the manufacturing process. The biology assembly was cleaned to 1 ppm total organics at the Martin Marietta plant; the GCMS assembly and the collector head and shroud assemblies were cleaned at the NASA White Sands Test Facility to less than 1 nanogram per square centimeter using procedures developed for Apollo [10]. Assemblies were broken down as far as possible, each part cleaned, flushed, reassembled and vacuum baked. The parts finished so clean that a screw would not thread into an insert, so new screws with silver plating as lubricant were used for reassembly, and motors were hermetically sealed to retain their lubricants. Assemblies were then pressurized to exclude contaminants and stayed pressurized through testing, transport and spacecraft integration, relying on their own sealed construction to hold cleanliness until landing. Organic cleanliness of the sample path was a mission-level requirement because the GCMS was the instrument searching for organic material [9].

Control and avionics

The Surface Sampler Control Assembly sits inside the lander body and receives 16-bit digital commands from the lander guidance, control and sequencing computer, decoding each and activating the appropriate motor or solenoid [10]. For boom movements the command word carries 9 bits of position information. Potentiometers inside the boom return analog position to the control assembly, which digitizes and compares commanded against actual position, removing power from the boom when they agree. Each command is timed to assure completion before the next is issued. Because of the Earth-Mars transmission delay, every sampler operation is executed by onboard sequences and logic rather than by real-time control. The first sequence was stored in the lander computer before launch and could be wholly replaced by ground update. The software is organized as tables of detailed digital commands for specific operations, with decision elements tied to the level detectors and the flexibility to deliver a sample to any investigation or combination of them.

Qualification

Component-level qualification, subsystem-level tests and integration into a lander System Test Bed were completed before flight. Beyond the normal dynamic and thermal environments, sampler hardware was flown on the NASA and USAF KC-135 to reproduce the effects of reduced Martian gravity on the mechanism [10]. Development also ran late enough that the surface sampler boom motor appeared on the Viking project’s list of top problems tracked by program management [9].

Flight performance and anomalies

The Viking Lander 1 sampler produced three separate operational problems, all resolved by resequencing rather than by hardware workaround [9].

On sol 2 a boom no-go occurred because the boom was not commanded to extend far enough and a locking pin, part of the shroud latching system, did not drop free [9]. The pin was intended to fall to the Martian surface during boom extension, and the commanded extension in the stored sequence was too short to release it [12]. The failure was reproduced on the science test lander at JPL, and analysts calculated that extending the boom to about 35 cm would let the pin fall. Boom retraction was also constrained in the recovery plan, because at a certain point the boom extraction motor deliberately clutches and shuts itself off to avoid motor damage. New commands were issued on sol 5, the boom extended far enough, the pin fell free, and samples were delivered on sol 8 to the biology instrument, the GCMS and the XRFS.

The GCMS then did not indicate a full sample and its analysis was automatically deferred by the lander computer. A second acquisition was ordered, and another boom no-go occurred before delivery. Analysis established that this one was caused by commanding two successive retract sequences [9]. The GCMS analysis was started on the material already delivered and returned a successful result; the boom was subsequently exercised in extension and continued without further problems.

Viking Lander 2 flew with sequence changes derived from these events and did not repeat them, but had one of its own: on sol 8, after delivery to the biology experiment and before delivery to the XRFS, a boom no-go occurred [9]. Analysis and further testing concluded that a switch sensing collector head rotation had malfunctioned. Subsequent boom sequences were modified to remove the need for that signal, and no further sampler anomalies occurred; all remaining deliveries were made as scheduled.

Results returned

No Viking lander experiment was designed to measure physical properties, so the mechanical properties of the surface materials were inferred from lander data: touchdown dynamics, engine-exhaust erosion, sample trenches, surface-bearing tests, backhoe touchdowns and motor currents in the comminutor [11]. Three soil-like materials were distinguished. Drift material is fine grained with local planes of weakness, and is consistent with an angle of internal friction of about 18 degrees, cohesion of 0.7 to 3.0 kPa and a bulk density of 1200 kg/m³ [11]. Blocky material is consistent with about 30 degrees, cohesions of 1.5 to 16 kPa and 1600 kg/m³. Crusty to cloddy material is variable: about 35 degrees and 0.5 to 5.2 kPa where chiefly crusty to cloddy, about 31 degrees and 0.2 to 2.3 kPa where fines and crusts are mixed, with 1400 kg/m³ plausible for the Lander 2 site [11]. The 18 degree figure for drift material was itself treated as a puzzle in the report, since the trench estimates average 18.2 ± 2.6 degrees and few materials other than smooth glass spheres are that low; the authors raise the possibility that it is an apparent angle produced by pore gas pressure rather than a true one. Landing provided an independent measurement [11]: both landers touched down at about 2 m/s, and on Lander 1 footpad 2 penetrated drift material 0.165 m while footpad 3 penetrated blocky material 0.036 m. The environment those numbers describe, and the design allowances against it, are the ones later Mars surface missions were built to [2].

Technologies developed and later heritage

The furlable flattened boom is the mechanism the program is remembered for: a 3 m reach and a 12 m2 workspace obtained from a boom stowed on a drum 152 mm in diameter [12], [9]. The collector head established the pattern later Mars samplers followed, a scoop with a secondary cutting element on the reverse face for harder material, a vibratory sieve integrated into the tool rather than placed downstream, and instrumentation carried on the tool itself for a separate investigation. The self-diagnosing vibrator, where a jaw-motion switch chops the drive waveform and a hung jaw automatically receives a more violent full-wave drive, is a mechanically implemented fault response with no software involvement. The organic cleanliness chain [12], cleaning to 1 ng/cm2 at White Sands, silver-plated fasteners as the only permitted lubricant on reassembly, hermetically sealed motors and pressurized storage through to landing, is the earliest flown implementation of an organically controlled sample path on another planet. The operational lesson was that a subsystem executing stored sequences with no real-time control needs its recovery paths in the sequence design: all four flight anomalies across the two landers were cured by changing commands, twice by extending a motion further than originally planned and once by ceasing to depend on a failed switch [12].

Every Mars surface sampler flown since has kept the two-jointed idea of a scoop for loose material and a harder-material tool riding beside or behind it, even where the mechanism inside changed completely. The Phoenix Robotic Arm replaced the furlable boom with a four-degree-of-freedom rigid arm and traded the backhoe for a rasp built into the back of the scoop, so a single tool could both dig icy soil and grind a harder ice-cemented layer before scooping the cuttings, and it used its own trenching and scraping motions as a soil mechanics instrument in the way the Viking backhoe had done [4], [5]. Mars Science Laboratory went further still, mounting a rotary-percussive drill and a hard-material scoop on the same turret so that either tool could be selected and then processed through an onboard sieving and portioning chain, CHIMRA, that plays the same role the Viking comminutor and metering tube played, reducing acquired material to a metered portion before delivery to an instrument [6], [7]. The Icebreaker drill concept, aimed at depths no flown Mars sampler has reached, still traces its own sample-acquisition history back to the Viking scoops as the first excavators put on the planet’s surface [3].

What is not established

The Viking sampler literature is a performance record, not an analysis. martin1977viking reports what the sampler did sol by sol but tabulates almost none of the underlying engineering numbers, force, torque, boom deflection, in extractable form, and its account of the sol 2 boom jamming is the design team’s reconstruction from telemetry and a matching ground failure rather than a reproduced fault [9]. The soil mechanics properties derived from the sampler and the rest of the lander were never measured directly by any instrument; Moore et al. state plainly that there is no sensible way to determine undisturbed drift density, that the widely quoted 1200 kg/m³ figure is only as good an estimate as any other, and that the anomalously low 18 degree friction angle for drift material has no settled explanation among pore-pressure, mixed shear failure and buried-obstruction hypotheses [11]. Both landing sites were in the northern plains, and Lander 2’s footpads both struck rocks, so no footpad penetration property exists for that site, and no rock strength was measured anywhere because nothing the sampler carried was built to chip or scratch rock [11]. Seger’s pre-launch account of the qualification program predates flight by two years and says only that component and subsystem testing was complete and the design appeared sound, without stating how well the flight units performed [12]. No source in this record ties a specific numeric fault tolerance or design margin to the sampler’s mechanisms; the anomaly record is a narrative of what sequence changes fixed each no-go, not an engineering post-mortem with root-cause margins attached.