Professional Documents
Culture Documents
SMART-1 Development and Lessons Learnt
SMART-1 Development and Lessons Learnt
www.elsevier.com/locate/actaastro
Abstract
SMART-1 is the first of the small missions for advanced research and technology as part of ESA’s science programme
“Cosmic vision”. It was successfully launched on September 27, 2003 and is presently traveling towards its destination, the
Moon. The main objective of the mission, to demonstrate solar electric primary propulsion for future Cornerstones (such as
Bepi-Colombo), has already been achieved. At the time of writing the electric propulsion system has been working already
for more than 3400 h and has provided a Delta-V to the spacecraft of more than 2500 m/s. The other technology objectives are
also being fulfilled by the verification of the proper functioning of such on-board experiments like the X-Ka band transponder,
the X-ray spectrometer, the near IR spectrometer, the laser link, etc. The scientific objectives are related to lunar science and
will be fulfilled once the spacecraft enters its operational lunar orbit, currently expected for January 2005. SMART-1 lunar
science investigations will include studies of the chemical composition of the Moon, of geophysical processes, environment
and high-resolution studies in preparation for future steps of lunar exploration.
SMART-1 has been an innovative mission in many aspects and we are now drawing some preliminary conclusions about
the lessons to be learnt. The paper describes the spacecraft and the technology elements with particular emphasis to the
technology nature of the mission. The on-board avionics employs many novel designs for spacecraft, including a serial
CAN bus for data communication, autonomous star trackers and extensive use of auto-code generation for implementing the
attitude control system and the failure, detection, isolation and recovery (FDIR). Finally, the orbital operation phase currently
ongoing, including the routine electric propulsion operations and the instrument commissioning, is providing a wealth of
data and lesson-learnt useful for future autonomous planetary missions.
© 2005 Published by Elsevier Ltd.
1. Introduction
∗ Corresponding author.
SMART-1 is the first European lunar mission. The
E-mail addresses: Peter.Rathsman@ssc.se (P. Rathsman), spacecraft was successfully launched on September
Joakim.Kugelberg@ssc.se (J. Kugelberg), Per.Bodin@ssc.se
(P. Bodin), Giuseppe.Racca@esa.int (G.D. Racca),
27, 2003 from Kourou as an auxiliary payload on an
Bernard.Foing@esa.int (B. Foing), Luca.Stagnaro@esa.int Ariane V launcher. The Swedish Space Corporation
(L. Stagnaro). (SSC) has been the prime contractor for the SMART-1
• Structure
• Mechanisms
• Avionics
• On-board software
• Attitude and orbit control
• Hydrazine
• Telemetry tracking and command
• Power
• Thermal
• Electrical propulsion
• Scientific payload
The EP and the scientific payload instruments were Fig. 4. EP thruster orientation mechanism.
provided as customer-furnished items by ESA. The
interfaces and the overall system performance was the
responsibility of the prime contractor (SSC). 3.2. Mechanisms
3.3. Avionics
Communication between all units is through a serial the ones handling the on-board autonomy:
CAN bus. All sensors and actuators are interfaced by
remote terminal units, which perform A/D conversions • System manager
and adapt the data formats to/from the CAN protocol. • Failure detection, isolation and recovery (FDIR)
Basically all FDIR decisions are taken by the OBSW. core
To protect the system against a failed controller or • EP manager
OBSW, the health of the OBSW and on-board space- • Power core
craft controller is continuously monitored by a “plat- • Thermal core
form supervisor” function within the power control • Data handling layer
and distribution unit (PCDU). If a periodical “health
message” from the OBSW is absent from the CAN bus The system manager implements error protections re-
for more than 60 s, the PCDU powers down the active lated to the controller and the CAN bus communi-
spacecraft controller and powers up the redundant one. cation, and also controls the initial start-up sequence
A feature with operational consequences is the im- including the autonomous deployment of the solar
plementation of a non-volatile configuration register arrays.
kept in the power subsystem which is not erased other The FDIR module governs, with the help of the
than in the event of a full S/C power cycling. The reg- other cores and the configuration register, the switch-
ister contains the necessary information on the health ing of on-board units. The general philosophy of the
status of the on-board units and is used by the FDIR FDIR is that if a single failure is detected and can be
software at controller power up. isolated, the FDIR reacts by switching to the redun-
dant unit and marks the failed unit as unhealthy in
3.4. On-board software the configuration register. This is, in most cases, done
without affecting the S/C mode, thereby allowing con-
The software architecture is shown below. For the tinued EP operation. In the event of critical failures,
purpose of this paper, the most interesting modules are e.g. a double failure of AOC actuators or sensors which
P. Rathsman et al. / Acta Astronautica 57 (2005) 455 – 468 459
invalid parent implies invalid child The data handling module implements a large subset
of the packet utilisation services (PUS):
ST
STSSBRTU SS • Telecommand verification
BAPTA
• Configuration of housekeeping telemetry reports
GP
System GRWRTU • Event reporting
CAN bus RW
TMTC
• Memory management
HTRTU HILV
PCDU • On-board scheduling (time tag queue)
EPMEL
EP
• On-board monitoring
CON
board TCRTU-A
• Large data transfer
TCRTU-B • Storage and retrieval management
AMIE • Connection test
D-CIXS
Payload EPDP The AOC, FDIR, thermal and power application soft-
CAN bus KATE ware has been developed in-house by the Swedish
SIR Space Corporation in a MATLAB/SIMULINK envi-
SPEDE ronment from which the code has been auto-generated.
unresolved child failure becomes parent failure Using MATLAB for flight-code development has
child power on implies parent power on proven to be a very efficient and reliable development
approach (Fig. 7).
Fig. 6. FDIR hierarchy with action directions.
3.5. Attitude and orbit control
Application Layer
BEAM
The Electric
CAT A propulsion subsystem
Thruster
(Anode) CAT B consists of the
following units:
• Xenon tank
• Bang-bang
FU XFC Pressure
Regulation Unit
• Xenon Flow
Controller
EPIRTU BPRU
• Power
Processing Unit
PPU
• Electrical Filter
Unit
Xenon
• Pressure
TANK
Regulation
Electronics
•
Power input Monitoring
(50 V DC) & Thruster
Control
to the thruster at the proper voltage and the electrical increasing the probability of collision with neutrals.
filter unit dampens out discharge current oscillations Collisions between drifting electrons and xenon gas
from the thruster. The thruster consumes 1220 W at create the plasma. The generated primary ions are ac-
maximum power and the dry weight of the subsys- celerated by the negative potential existing in the area
tem is 29 kg. At launch, the tank contained 82.5 kg of near the exit of the chamber due to the Hall-effect.
xenon, providing spacecraft with 4000 m/s of delta-V. The external cathode acts also as a neutralizer, in-
The EP system only contains limited redundancy. jecting electrons into the beam, in order to maintain
The cathodes, XFC and BPRU are doubled whereas zero-charge equilibrium in the thrust beam and on the
the rest of the system is single string. The SMART-1 spacecraft.
EP thruster is a stationary plasma thruster (PPS- The PPS-1350, shown in Fig. 11, has an exit di-
1350), which constitute a family of electric propul- ameter of 100 mm. Even though the thruster can
sion engines belonging to the category of Hall-effect operate at a maximum discharge power of 1500 W,
thrusters. it is also capable of start-up and functioning at a
In this type of thrusters, electrons from an external much reduced power (480 W). At beginning of life
cathode enter an annular ceramic discharge chamber, (BOL), the SMART-1 spacecraft was designed to
attracted by an anode piece. On their way to the anode, provide a maximum of 1190 W of nominal discharge
the electrons encounter a radial magnetic field created power, which produces a nominal thrust of 68 mN at
between inner and outer coils, causing cyclotron mo- a specific impulse of 1640 s. This family of thrusters
tion around the magnetic field lines and dramatically has a qualified lifetime of 7000 h in cycles at maxi-
P. Rathsman et al. / Acta Astronautica 57 (2005) 455 – 468 463
3200 Cum ON
3000 EP ON 60
2800 EP OFF
2600 50
2400
2200
2000 40
1800
Hours
1600
30
1400
1200
1000 20
800
600
10
400
200
0 0
18 Apr 04
20 Nov 03
01 Oct 03
28 Feb 04
09 Jan 04
07 Jun 04
27 Jul 04
Fig. 12. EP ON and OFF Times during first 11 months of mission.
Earth centred inertial projection on moon orbital plane Earth centred inertial projection on moon orbital plane
SPT100: launch by AR 5: 2002/09/21 -> 2003/12/29; 350 kg -> 295.764 kg 500000
5.0
400000
4.0
300000
3.0
200000
Y-coordinate (× 1E5 km)
Y-coordinate (km)
2.0
100000
1.0
0
0.0
-100000
-1.0
-200000
-2.0
-300000
-3.0
-400000
-4.0
-500000
-5.0 -500000 -200000 0 200000 500000
-5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 X-coordinate (km)
X-coordinate (× 1E5 km) (direction of moon ascending mode w.r.t. EQ2000)
(ascending mode of moon orbital plane)
Fig. 14. As-flown trajectory.
Fig. 13. Pre-flight predicted trajectory.
Moon distance km
90000
80000
70000
60000
50000
40000
30000
20000
10000
0
0 24 48 72 96 120 144 168 192 216 240
based on swept charge device detectors and micro- The objectives of the SMART-1 missions are both
collimators, measuring X-ray fluorescence from the technological and scientific. Clearly the full scien-
lunar surface discriminating the solar background by tific return will be achieved only during the scientific
means of the X-ray solar monitor. Several measure- operations in lunar orbit. Concerning the technologi-
ments have been performed of bright celestial sources, cal objective, the main purpose was the flight test of
like SCO-X1 and others. electric primary propulsion. This is considered largely
AMIE: Asteroid-Moon imaging experiment (2.2 kg, achieved because of the following considerations:
9 W). A 5.3◦ FOV miniaturised camera with a four-
band fixed filter (0.75, 0.9 and 0.95 m wide-band • The engine has been functioning in space for more
mineralogical filters and a 0.847 m narrow-band fil- than 3300 h with good performance.
ter for the laser-link). The camera is based on high- • The optimisation, control and navigation of the
density 3-D cube-packed multi-chip module electron- spacecraft during the powered operations have
ics. Several images of the Earth and the Moon have been successfully performed.
been taken during the cruise. An example is shown in • The EP has been monitored with the diagnostic
Fig. 17. package and its effects have been shown and the
P. Rathsman et al. / Acta Astronautica 57 (2005) 455 – 468 467
results are being analysed by the experts. So far, no that usually prevent a solid architectural design and
adverse effects have been observed on S/C subsys- changes in later stages. The end result was a coher-
tems as the result of EP operation. ent architecture that was maintained throughout the
• The second lunar resonance has been successfully project.
achieved and the remaining one will follow before • Extensive testing: Extensive testing together with
lunar capture. the capability of quickly recovering errors (software
• The innovative aspect of combining EP with “weak based spacecraft) replaced exhaustive pre-analysis
stability” gravity assists has been demonstrated. and lengthy trade-offs.
• Joint-team approach: Rather than performing a
Concerning the technology instruments they have all “surveillance” function derived from the detached
been commissioned, thereby completing their objec- role of customer to contractor, the ESA project
tive for the lunar transfer phase. team took an active part in design decisions as
well as providing a continuous and appreciated
peer-review of the prime’s work. The ESA in-
4. Lessons learnt volvement allowed a high degree of visibility in
all areas, allowing them to verify adherence to re-
4.1. EP operations quirements beyond the normal reviewing of formal
documentation.
For the EP operation, the original intention was to • Keep team small: During certain critical periods,
pre-programme up to 10 days of autonomous opera- it was necessary to increase the manpower. This
tions. While this has been accomplished once or twice, was achieved by increasing the work-load for the
in the general case, also very small variations in the existing team rather than taking on new staff. This
EP thrust level generates an increased divergence be- allowed the core team to remain small and therefore
tween the predicted and actual position in space. Over quickly respond to project needs.
a few days, this error propagates to a level that is un-
acceptable. The remedy is to recalculate the flight tra-
jectory prediction every few days. This, however, pre-
vents the long term planning that is desirable in order 5. Conclusions
to reduce ground operation costs. A possible solution
for future EP missions would therefore be to have on- The performance of SMART-1, ESA’s first mission
board an autonomous guidance and navigation system. to the Moon, has so far been excellent. Due to more
On SMART-1 such a system is run as a test package, available on-board power than predicted, combined
but with the control loop closed on the ground. with good ground-based navigational techniques by
Another major aspect that we have learned in fly- the flight control team, the travel time to the Moon has
ing SMART-1 is the advantage of easily being able to been shortened by approximately 2 months compared
enhance on-board autonomous functions, and the ca- to initial predictions. The lunar capture is scheduled
pability to programme new functions (e.g. EP restart for mid-November 2004, with scientific observations
after SET flame-outs). starting once the final lunar orbit has been achieved in
mid-January 2005. In-flight commissioning of the sci-
4.2. Techniques to reduce development costs entific instruments has successfully been performed,
leading to hopes for a very rewarding scientific phase
• The Prime retained a large part of the develop- of the mission in 2005. SMART-1 is regarded as ESA-
ment functions: On SMART-1, the prime (SSC) re- s first “faster, better, cheaper” mission within the sci-
tained a significant part of the new developments. ence programme. As part of the lessons learnt activity,
This included the on-board avionics, the attitude the joint-team approach of the prime contractor and
control system and the application software includ- ESA is regarded as an important element in the suc-
ing FDIR. This allowed an evolution of the design cessful realization of a technologically advanced low-
at low cost by removing the contractual barriers cost mission.
468 P. Rathsman et al. / Acta Astronautica 57 (2005) 455 – 468