Deep Space Mars Microprobes

The DS-2 mission was the second 'Deep Space' mission in NASA's New Millennium technology validation programme (Smrekar et al., 1999). It was to demonstrate miniaturized penetrators to enable subsurface and network science. The spacecraft that flew were radically smaller - by two orders of magnitude -than anything NASA had previously flown to the planets. The project cost a remarkably modest $29.6 million.

The original concept anticipated deployment at low latitude on Mars, and a payload including a microseismometer. As the mission evolved, and the delivery opportunity as a 'piggyback' payload on the Mars Polar Lander emerged, the mission concept had to change. In particular, the low-temperature environment at high latitudes on Mars reduced the expected energy capacity of the batteries (and thus the penetrators' lifetime) to the point where it was no longer likely that worthwhile seismic data would be acquired.

The new payload therefore centred on measuring the volatile content of the high-latitude soil. The same thermal environment that eroded the energy capability of the mission also made it likely that water might be trapped as ice in the soil.

Entry performance was driven by the entry conditions (at 6.9 km s-1 with a flight path angle of -13.1°, as for MPL) and the allowed flight parameters (velocity, angle of incidence) at impact (Braun et al. 1999b). A significant and unusual aspect was that since the probes were delivered as a secondary payload, no orientation or spin-up was provided to ensure any given orientation at entry (see Figure 25.1 for a view of the mounting structure/separation mechanism). (It may be noted in this context that the Viking entry bodies were hypersonically stable flying backwards.) The result was first the use of a 45° half-angle cone, instead of the more usual 70°, which offers higher drag (Cd — 1.7, versus 1.05 for the 45° cone). The second aspect was the very tight requirement on having the centre of mass as far forward as possible. These two aspects gave the system a

Figure 25.1. A DS-2 Microprobe mounted in the structure that attached it to Mars Polar Lander (mounted in turn on mechanical ground support equipment). The probes were ejected with a relative speed of <0.3 m s"1 and no spin. Note the very sturdy-looking 'spider' fitting that holds the microprobe.

Figure 25.1. A DS-2 Microprobe mounted in the structure that attached it to Mars Polar Lander (mounted in turn on mechanical ground support equipment). The probes were ejected with a relative speed of <0.3 m s"1 and no spin. Note the very sturdy-looking 'spider' fitting that holds the microprobe.

strong 'weathercock' stability. The forward CoM requirement necessitated a novel two-part design, wherein the penetrating forebody was surrounded by, rather than mounted in front of, an aftbody that would remain on the surface to perform communications.

To satisfy the permitted range of impact speed (140-210 m s" ) the ballistic

coefficient fl was required to be in the range 18-49 kg m , while the penetration incidence angle constraint (i.e. the velocity relative to vertical) of <30° required fl < 44.5 kg m"2. The angle of attack (i.e. the orientation of the vehicle relative to the velocity vector) was to be less than 10° - exceeding the incidence or angle of attack limits would lead to 'skip', the aftbody not remaining embedded securely in the ground. Nominally, the 0.35 m diameter shell and as-built mass of 3.6 kg gave fl = 36.5 kg m-2, leading to a peak deceleration of 12.4 g at an altitude of 44 km and a peak stagnation point heating of 175 W cm-2, after about 100 and 80 s respectively; integrated stagnation point heating would be 8085 J cm-2 (Micheltree et al., 1998). Impact (at a nominal altitude of 6 km - the southern high latitudes are elevated terrain, well above the 6 mbar Martian datum) would be at 191 m s-1 (around Mach 0.8) and 20° incidence, around 270 s after crossing the entry threshold (defined as a radius of 3522.2 km, around 142 km above the surface). The aftbody had to tolerate decelerations of some 60 000 g. The fore-body, penetrating further than the aftbody, therefore had a longer stroke over which to decelerate, so its impact loads were specified at 30000 g. The 3-c landing ellipse determined from Monte-Carlo simulations was about 180 X 20 km in extent: these simulations gave a probability of 76% for each probe (or 94% for either) to satisfy the impact conditions.

Tests to achieve reliable sub-surface soil sampling passively with holes or blades on the forebody were unsuccessful, and an auger drill had to be included in the design. This 9 mm diameter drill had an 8.5 mm stroke, driven by a 1W motor modified for high-g impact loads. The sample drizzled during drilling into a small cup; after drill operation the sample was sealed inside the forebody by a small pyrotechnic door mechanism.

The soil water detector comprised a small cup (able to hold about 160 mm3 of sample) around which a nichrome heater wire was wound. Thermistors were mounted on the edge of the cup and in the centre. The temperature rise experienced by the sensors for a given applied heating current would depend on the amount and thermal properties of the material deposited in the cup. It therefore acts as a crude form of thermal analyser. In particular, a deviation from a smooth heating curve would be observed if significant amounts of ice were present in the sample.

More sensitive detection of water was accomplished with a small absorption cell. A tunable diode laser emits light in a narrow bandwidth (nominally around 1.37 mm) that is swept across a wavelength range by modulating the current to the diode. The light from this source passes through a small volume that may be filled with gas from the sample; when the laser is at wavelengths where water vapour absorbs strongly, the light received by a photodiode is attenuated. The frequency-sweeping approach allows a more sensitive and robust detection than would a fixed wavelength.

Temperature sensors (platinum resistance thermometers) were embedded in the walls of the forebody to monitor the secular cooling of the probe after its emplacement into the Martian soil. The cooldown curve would yield information on the thermal conductivity (and indeed the temperature) of the soil (Urquhart and Smrekar, 2000; Smrekar et al., 2001).

A pressure sensor, using a micromachined silicon membrane, was to be flown. An earlier concept, also robust enough to be deployed on a penetrator, used a small radioactive source (in fact the same source used in domestic smoke detectors) as an ionization gauge.

Entry and impact accelerometers were installed on the aftbody and forebody respectively. The former were off-the-shelf micromachined devices (Analog Devices ADXL250, often used in cars for airbag actuation), the latter an Endevco 7270 piezoresistive accelerometer.

As the mission development progressed, the scientific capability of the mission was eroded somewhat (science was always only a bonus - the principal goal of the mission was to demonstrate a safe delivery to the surface). In particular, the telecommunications system experienced severe development difficulties, and the original intent of fitting the entire system on a single hybrid chip was not realized. The replacement design, introduced only 12 months or so before launch, used discrete components requiring both more volume (or circuit-board space) and more electrical power. The electrical power requirement reduced the expected mission duration (again) and the energy available for sample heating, which was otherwise the dominant consumer of energy. The growth of board space required for the communications system meant that the pressure sensor could no longer be accommodated.

Unusually, mass (see Table 25.1) was not the tightest constraint on instrumentation. Volume was in general more at a premium than mass. The mass distribution of the probe was critical, however, in that the passive aerodynamic stabilisation of the entry shell required that the probe centre of mass be as far forward as possible (see Figure 25.2). This was achieved in part by the forebody-in-aftbody concentric design, and by introducing a tungsten nose to the fore-body. This hemispherical nose, around 200 g in mass (about 1/3 of the total forebody mass, and many times the mass of the instruments) exploits the extremely high density of tungsten.

Table 25.1. Mass budget (per probe) of the DS-2 Mars Microprobes

Aftbody 1780 (incl. ^50 telecom system, 320 battery)

Forebody 670 (incl. 200 tungsten nose, ~10 microcontroller)

Entry system 1165

Entry mass total 3610

Spacecraft interface 2920

Total 6530



Heat shield

Aeroshell back shell


Figure 25.2. Drawing showing the layout of the probe installed in its aeroshell. The location of the probe gives a forward centre of gravity and thus 'weathercock' stability.

Heat shield

Aeroshell back shell


Figure 25.2. Drawing showing the layout of the probe installed in its aeroshell. The location of the probe gives a forward centre of gravity and thus 'weathercock' stability.

Little wind tunnel testing was performed, the aerodynamic performance being assessed principally by computational fluid dynamic simulation. The simulations were validated by ballistic range testing, and one hypersonic wind tunnel test at a facility in Russia.

Impact testing was a laborious aspect of the programme. Electronic components were mounted on test projectiles and shot into the ground, and taken back to the lab for health checks. The flight batteries posed a significant hazard in that they could explode if damaged by the impact. A separate series of tests was performed to evaluate impact accelerometer performance (Lorenz et al., 2000). The aftbody was to remain on the surface to permit data transmission - its penetration depth was stated to be <10 cm, although in some tests on soft targets it did penetrate more than 30cm (Lorenz et al., 2000). The forebody was nominally to penetrate up to —1 m.

The aftbody (Figure 25.3) was partially independent, in that it contained the crucial systems, namely the batteries and the communication system. Even if the forebody failed, or the umbilical cable broke, the aftbody would continue to operate and transmit data. The aftbody telecommunications system included a 6502 microprocessor, running around 8000 lines of code at 10 MHz. The receiver would operate for 1 s every minute to detect the query tones of the MGS relay spacecraft. After the first successful downlink, the sample sequence would be run.

The forebody (Figure 25.4) contained a microcontroller (Figure 25.5). This unit was based on an 8051 architecture with 64K RAM and 128K EEPROM.

Figure 25.3. Exploded view of the aftbody. Electronics were mounted on a flat circuit board; the principal other elements were a solar-cell experiment, the antenna and the batteries.

The unit, which incorporated a 16-channel 12-bit analogue-to-digital converter, was designed for very low power (<50 mW at 1 MHz, with a 1 mW sleep mode) and low volume and mass (<8cm3, <90 g). The microcontroller supervised forebody operations with about 14000 lines of code (in 8051 assembler) including the impact accelerometer sampling and the drill and heater cup. The analogue-to-digital converter on the forebody demonstrated operation at —70 °C, although its accuracy degraded somewhat at that temperature.

Pinion gear -Jam nut.

Forebody electronics.

Pinion gear -Jam nut.

Drill stem assembly Door slot . Sample chamber

Forebody tube.

1 ^ Accelerometer cavity

Science block structure

Drill stem assembly Door slot . Sample chamber

Drill motor

Forebody nose

Figure 25.4. Cross-section of the forebody layout. Clearly this spacecraft structure, like that of a wristwatch, is much more tightly integrated with the other subsystems. The auger drill would emerge to the right, and soil would drizzle into the sample chamber.

Forebody nose

Figure 25.4. Cross-section of the forebody layout. Clearly this spacecraft structure, like that of a wristwatch, is much more tightly integrated with the other subsystems. The auger drill would emerge to the right, and soil would drizzle into the sample chamber.

The forebody transmitted data along the umbilical to the aftbody twice every hour.

The DS-2 structure used some rather exotic materials. The forebody comprised a high-strength superalloy tube (MP35N) with a tungsten nose. The 'science block', into which the drill motor, thermal cup and impact accelerometer were embedded was simply aluminium alloy.

The aftbody was made from magnesium and titanium alloys. Its shape was driven by the mass distribution constraints, together with the required penetration performance. In particular, at higher angles of attack, the aftbody had a tendency to 'bounce' (more correctly, some rolling was involved) off the target.

Figure 25.5. The extreme level of miniaturization applied is evident in this photograph of the forebody microcontroller. The bevel gear for the sample drill is visible at the top left.

Introduction of tines on the front face of the aftbody helped to alleviate this tendency.

Small steel wires ('whiskers') were added to the titanium antenna to improve radiated signal performance (acting in effect as a longer antenna), while remaining robust to impact. The antenna survival was demonstrated by firing a probe backwards through a sample of the aft shield.

The heat shield, to which the probe was attached via three titanium fixtures, was of a very novel design. The structural stiffness was provided by an inner shell made from 0.8 mm thick silicon-carbide ceramic. The outer thermal-protection layer was a porous silica-rich layer called SIRCA-SPLIT (silicone impregnated reusable ceramic ablator-secondary polymer layer-impregnated technique). This thermal protection material was 1 cm thick at the nose. The backshell (hemispherical, so pressure forces during entry act through the centre of mass and do not apply torques) was made from FRCI (fibrous refractory composite insulation). While stiff, this structure was brittle, requiring careful ground handling. The brittleness was by design, in that the shell would shatter on impact without impeding the penetration.

The umbilical cable was folded as a concertinaed ribbon. The flat cable was made by depositing conductive traces on a Kapton substrate. Originally, the umbilical was to be 2 m long. Concern arose that the cable should be shielded from electrical transients, and so a deposited metallic shield was attempted. This shield layer turned out to be quite brittle and stiff (Arakaki and D'Agostino, 1999), such that the cable required a larger storage volume. This problem led to a descope of the cable to a shorter length, supported by the fact that the fore-aftbody separation in tests exceeded one metre in only two out of around fifty tests. In the event, the brittleness difficulties led the cable to not have a shield after all. The umbilical technology used in principle also allows electronic components to be installed on the same substrate (in fact many modern consumer electronic items such as CD players incorporate circuits built on flexible substrates) - a prime example being temperature sensors, so that heat flow measurements could be performed.

The batteries had very tight requirements. First was the ability to provide useful current even at temperatures of -78 °C (where many electrolytes are frozen, and the diffusion of ions in any electrolyte is slowed). The second driving requirement was the ability to tolerate 80 000 g loads at impact after a 3-year shelf life. Specially developed half-D cells by Yardney used lithium thionyl chloride. These cells presented some safety concerns in that if damaged (e.g. by short circuits induced by impact deformation) they could explode, and ordinary alkaline batteries were used during impact tests. The batteries provided for a 6 W-hr energy budget, giving an expected lifetime of 1-3 days, the low temperature and possible damage to some cells on impact being likely limiting factors.

The communications at UHF (with an RF power of only —300 mW) permitted 7 kbps during a typical pass of MGS, which would last a little over 10 minutes. At the high latitude landing site, the polar-orbiting MGS would make several high passes per day. The receiver would detect the presence of a query tone transmitted from the orbiter (the tone frequency identifying which of the probes was to transmit) which would trigger the microprobe to transmit its data.

The DS-2 microprobes were installed on the Mars Polar Lander at KSC, and launched on 3 January 1999 on a Delta 7425. After a nominal cruise, during which there was no communication with DS-2, the commands to separate MPL from its cruise stage, and to deploy the DS-2 probes, were executed. Nothing further was heard from either MPL or the DS-2 probes.

There was never an end-to-end test of the DS-2 probes during which all systems were operated together, fired into the ground and demonstrated to operate thereafter. There was similarly no end-to-end test of the communications system, which relied on the Mars Balloon Relay on the Mars Global Surveyor orbiter. This French-supplied relay system was installed originally to support operation of a balloon to be carried on the Russian Mars-96 mission. The DS-2 communications hardware itself was delivered quite late and so was not extensively tested prior to launch.

Although the detailed failure investigation favoured separate problems, the simultaneous loss of MPL and DS-2 suggests a common failure mode. Although dual initiators are installed on the separation system, extreme temperatures or an undiscovered software error could still have prevented separation. Another common mode could be attributed to Mars itself, if the terrain were too soft or rough.

The failure mode favoured (e.g. Harland and Lorenz, 2005) for MPL is that a sensor transient when the landing legs locked into place may have been interpreted by the on-board computer as an indication of contact with the ground. This problem would have been detected in a ground test, had not an unrelated problem (subsequently fixed) occurred in that test. 'Thinking' it had landed, the vehicle would have shut down the descent engines at an altitude of above 50 m, and would have crashed onto the ground.

While perfectly plausible, this scenario does not account for the loss of both DS-2 probes. Possible failure modes include failure of the radio transmitters or batteries, perhaps by hitting rocks. Very soft terrain is another hazard, if the aftbody bearing the communications antenna were buried so deeply that the radio signals from the probes were attenuated.

Some months after the MPL/DS-2 loss, images from the Mars Global Surveyor discovered features that may indicate seepage and flow from subsurface aquifers, around 100-200 m below the ground. It was noted cynically by some science team members that the umbilical system had not been tested underwater - it would be ironic indeed if the microprobes' quest for water on Mars were fatally successful.

Whatever the technical causes of failure, the programmatic causes are all too clear - a rushed schedule, changing goals and inadequate testing.

0 0

Post a comment