OSIRIS - The scientific camera system onboard Rosetta

Abstract

The Optical, Spectroscopic, and Infrared Remote Imaging System OSIRIS is the scientific camera system onboard the Rosetta spacecraft (Figure 1). The advanced high performance imaging system will be pivotal for the success of the Rosetta mission. OSIRIS will detect 67P/Churyumov-Gerasimenko from a distance of more than 106 km, characterise the comet shape and volume, its rotational state and find a suitable landing spot for Philae, the Rosetta lander. OSIRIS will observe the nucleus, its activity and surroundings down to a scale of ~2 cm px−1. The observations will begin well before the onset of cometary activity and will extend over months until the comet reaches perihelion. During the rendezvous episode of the Rosetta mission, OSIRIS will provide key information about the nature of cometary nuclei and reveal the physics of cometary activity that leads to the gas and dust coma.
OSIRIS comprises a high resolution Narrow Angle Camera (NAC) unit and a Wide Angle Camera (WAC) unit accompanied by three electronics boxes. The NAC is designed to obtain high resolution images of the surface of comet 67P/Churyumov-Gerasimenko through 12 discrete filters over the wavelength range 250–1000 nm at an angular resolution of 18.6 μrad px−1. The WAC is optimised to provide images of the near-nucleus environment in 14 discrete filters at an angular resolution of 101 μrad px−1. The two units use identical shutter, filter wheel, front door, and detector systems. They are operated by a common Data Processing Unit. The OSIRIS instrument has a total mass of 35 kg and is provided by institutes from six European countries.

Full text

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA
H. U. KELLER1,∗, C. BARBIERI2, P. LAMY3, H. RICKMAN4, R. RODRIGO5,
K.-P. WENZEL6, H. SIERKS1, M. F. A’HEARN7, F. ANGRILLI2, M. ANGULO8,
M. E. BAILEY9, P. BARTHOL1, M. A. BARUCCI10, J.-L. BERTAUX11,
G. BIANCHINI2, J.-L. BOIT3,V.BROWN
5, J. A. BURNS12,I.B¨
UTTNER1,
J. M. CASTRO5, G. CREMONESE2,20, W. CURDT1, V. DA DEPPO2,22,
S. DEBEI2, M. DE CECCO2,23, K. DOHLEN3, S. FORNASIER2, M. FULLE13,
D. GERMEROTT1, F. GLIEM14, G. P. GUIZZO2,21, S. F. HVIID1, W.-H. IP15,
L. JORDA3, D. KOSCHNY6, J. R. KRAMM1,E.K¨
UHRT16,M.K¨
UPPERS1,
L. M. LARA5, A. LLEBARIA3,A.L´
OPEZ8,A.L´
OPEZ-JIMENEZ5,
J. L ´
OPEZ-MORENO5, R. MELLER1, H. MICHALIK14, M. D. MICHELENA8,
R. M ¨
ULLER1, G. NALETTO2, A. ORIGN´
E3, G. PARZIANELLO2, M. PERTILE2,
C. QUINTANA8, R. RAGAZZONI2,20, P. RAMOUS2, K.-U. REICHE14, M. REINA8,
J. RODR´
IGUEZ5, G. ROUSSET3,L.SABAU
8, A. SANZ17, J.-P. SIVAN18 ,
K. ST ¨
OCKNER14, J. TABERO8, U. TELLJOHANN6, N. THOMAS19, V. TIMON8,
G. TOMASCH1, T. WITTROCK14 and M. ZACCARIOTTO2
1Max-Planck-Institut f¨
ur Sonnensystemforschung, 2, 37191 Katlenburg-Lindau, Germany
2CISAS, University of Padova, Via Venezia 1, 35131 Padova, Italy
3Laboratoire d’Astrophysique de Marseille, 13376 Marseille, France
4Department of Astronomy and Space Physics, 75120 Uppsala, Sweden
5Instituto de Astrof´
ısica de Andaluc´
ıa – CSIC, 18080 Granada, Spain
6Research and Scientific Support Department, ESTEC, 2200 AG Noordwijk, The Netherlands
7Department of Astronomy, University of Maryland, MD, 20742-2421, USA
8Instituto Nacional de T´
ecnica Aeroespacial, 28850 Torrejon de Ardoz, Spain
9Armagh Observatory, College Hill, Armagh BT61 9DG, Northern Ireland
10Observatoire de Paris – Meudon, 92195 Meudon, France
11Service d’A´
eronomie du CNRS, 91371 Verri`
ere-le-Buisson, France
12Cornell University, Ithaca, NY, 14853-6801, USA
13Osservatorio Astronomico de Trieste, 34014 Trieste, Italy
14Institut f¨
ur Datentechnik und Kommunikationsnetze, 38106 Braunschweig, Germany
15Institute of Space Science, National Central University, Chung Li, Taiwan
16Institut f¨
ur Planetenforschung, DLR, 12489 Berlin-Adlershof, Germany
17Universidad Polit´
ecnica de Madrid, 28040 Madrid, Spain
18Observatoire de Haute-Provence, 04870 Saint Michel l’Observatoire, France
19Physikalisches Institut der Universit¨
at Bern, Sidlerstraße 5, 3012 Bern, Switzerland
20INAF, Osservatorio Astronomico, Vic. Osservatorio 5, 35122 Padova, Italy
21Carlo Gavazzi Space, Via Gallarate 150, 20151 Milano, Italy
22CNR – INFM Luxor, Via Gradenigo 6/B, 35131 Padova, Italy
23DIMS, University of Trento, Via Mesiano 77, 38050 Trento, Italy
(∗Author for correspondence: E-mail: keller@mps.mpg.de)
(Received 27 March 2006; Accepted in final form 22 November 2006)
Abstract. The Optical, Spectroscopic, and Infrared Remote Imaging System OSIRIS is the sci-
entific camera system onboard the Rosetta spacecraft (Figure 1). The advanced high perfor-
mance imaging system will be pivotal for the success of the Rosetta mission. OSIRIS will detect
Space Science Reviews (2007) 128: 433–506
DOI: 10.1007/s11214-006-9128-4 C
Springer 2007

434 H. U. KELLER ET AL.
67P/Churyumov-Gerasimenko from a distance of more than 106km, characterise the comet shape
and volume, its rotational state and find a suitable landing spot for Philae, the Rosetta lander. OSIRIS
will observe the nucleus, its activity and surroundings down to a scale of ∼2cmpx
−1. The observa-
tions will begin well before the onset of cometary activity and will extend over months until the comet
reaches perihelion. During the rendezvous episode of the Rosetta mission, OSIRIS will provide key
information about the nature of cometary nuclei and reveal the physics of cometary activity that leads
to the gas and dust coma.
OSIRIS comprises a high resolution Narrow Angle Camera (NAC) unit and a Wide Angle Camera
(WAC) unit accompanied by three electronics boxes. The NAC is designed to obtain high resolution
images of the surface of comet 67P/Churyumov-Gerasimenko through 12 discrete filters over the
wavelength range 250–1000nm at an angular resolution of 18.6 μrad px−1. The WAC is optimised
to provide images of the near-nucleus environment in 14 discrete filters at an angular resolution of
101 μrad px−1. The two units use identical shutter, filter wheel, front door, and detector systems. They
are operated by a common Data Processing Unit. The OSIRIS instrument has a total mass of 35 kg
and is provided by institutes from six European countries.
Keywords: Rosetta, OSIRIS, camera, imaging system, spectroscopic, cometary activity, 67P/
Churyumov-Gerasimenko, Narrow Angle Camera, Wide Angle Camera
1. Introduction
1.1. HISTORY OF THE INSTRUMENT
On March 14th 1986 at 00:03 Universal Time, the European Space Agency’s (ESA)
spacecraft Giotto made its closest approach to comet 1P/Halley. The only remote
sensing instrument onboard the spacecraft was the Halley Multicolour Camera
(HMC), which was designed to image the nucleus and innermost coma of the
comet from the spinning spacecraft. The instrument development was led by the
Max-Planck-Institut f¨ur Aeronomie (now Max-Planck-Institut f¨ur Sonnensystem-
forschung, MPS) with the participation of several other major institutes in Europe
(Keller et al., 1995).
HMC was by far the most complex instrument onboard Giotto and a remarkable
success (Figure 2). After the International Rosetta Mission (hereafter ‘Rosetta’)
was selected as the 3rd Cornerstone Mission of ESA’s Horizon 2000 programme,
it was natural for a significant part of the HMC team to come together again to
build the imaging system for the main spacecraft. Groups from MPS, the Labo-
ratoire d’Astronomie Spatiale in Marseille (now Laboratoire d’Astrophysique de
Marseille, LAM), the Osservatorio Astronomico di Padova (UPD), the Belgian In-
stitute for Space Aeronomy (BISA), the Rutherford Appleton Laboratory (RAL)
and the Deutsches Zentrum f¨ur Luft- und Raumfahrt (DLR) started working to-
gether in 1995 to study a modern imaging system that would be powerful enough
to maintain Europe’s lead in the remote sensing of cometary nuclei. The result-
ing proposal for the Optical, Spectroscopic, and Infrared Remote Imaging System
OSIRIS was the only experiment proposed to ESA as the main imaging system on
the Rosetta spacecraft in response to ESA’s Announcement of Opportunity (AO).

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 435
Figure 1. The two OSIRIS cameras units (top left, with white radiators) with VIRTIS and the Philae
lander mounted on the –xpanel of Rosetta.
The proposal included two cameras units (one narrow angle, one wide angle) with
an infrared imager incorporated into the narrow angle system (Thomas et al., 1998).
There was also the possibility to include a UV spectrometer to cover the wavelength
range from 200 to 400 nm. The instrument was extremely ambitious.
The Rosetta mission definition study, or ‘Red Report’, which outlined the goals
and implementation of the mission, included a dedicated scientific imaging system
as part of the strawman payload. However, funding problems led to considerable
uncertainty as to whether the ESA Member States could fund such an ambitious
imaging system. These problems were resolved about one year after the selection
of the rest of the payload when a descoped version of OSIRIS was finally approved.
The descoped version eliminated the IR imaging element of the cameras (the main
interest of the Belgian and UK partners, BISA and RAL). However, additional
support was offered by a group of Spanish laboratories led by the Instituto de As-
trof´ısica de Andaluc´ıa (IAA), by ESA’s Space Science Department (now Research
and Scientific Support Department, RSSD) and by the Astronomical Observatory
of Uppsala (now Department of Astronomy and Space Physics, DASP) in Sweden.
The contributions from the different institutes finally involved in the OSIRIS in-
strument development are listed in Table I.
OSIRIS was delivered to ESA and integrated on the Rosetta spacecraft in
2002. The launch of Rosetta, originally foreseen for January 2003, was deferred

436 H. U. KELLER ET AL.
Figure 2. The nucleus of comet 1P/Halley as observed on March 14th, 1986, by the Halley Multi-
colour Camera onboard the Giotto spacecraft.
to early 2004, changing the target comet from 46P/Wirtanen to 67P/Churyumov-
Gerasimenko. OSIRIS was successfully commissioned in-flight during the months
after the exciting launch on March 2nd 2004 and in the meantime has been used for
scientific measurements of comet 9P/Tempel 1 in the course of the Deep Impact
mission (Keller et al., 2005; K¨uppers et al., 2005).
1.2. THE OSIRIS NAME AND SYMBOL
The name, OSIRIS, standing for Optical, Spectroscopic, and Infrared Remote Imag-
ing System, was selected at the time of the first instrument proposal, which included
infrared imaging capability and the possibility of an ultraviolet spectrometer. Al-
though several aspects of the original instrument were descoped, the name was
retained.
Osiris was the Egyptian god of the underworld and of vegetation. He was the
brother and husband of Isis who gave birth to their son, Horus, after his death. He
was killed by the rival god, Seth. As legendary ruler of predynastic Egypt and god
of the underworld, he symbolised the creative forces of nature and the indestruc-
tibility of life. The name was selected for the imaging system because Osiris is

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 437
TABLE I
Tasks of the OSIRIS consortium.
Responsible
Task institute
Overall responsibility and project management, system engineering, interfaces,
Focal Plane Assemblies, CCDs and Readout Boards, HK Boards, integration
& qualification of E-Boxes, harnesses, system integration, high-level
software, NAC & WAC system calibration, QA, mission operations
MPS
NAC telescope, camera integration and qualification LAM
WAC optical bench, camera integration and qualification, shutter mechanisms
and shutter electronics, Front Door Mechanisms (mechanisms for NAC and
WAC )
UPD
Mechanism Controller Board IAA
Filter Wheel Mechanisms, E-Box Power Converter Module, NAC & WAC CRB
Power Converter Modules
INTA
Data Processing Unit RSSD
Mass memory, low-level software and data compression IDA
NAC & WAC Filters DASP
Thermal and structural analysis, NAC MLI, WAC FPA MLI UPM
MPS – Max-Planck-Institut f¨ur Sonnensystemforschung (Germany), LAM – Laboratoire
d’Astrophysique de Marseille (France), UPD – University of Padova (Italy), IAA – Instituto de
Astrof´ısica de Andaluc´ıa (Spain), INTA – Instituto Nacional de T´ecnica Aeroespacia (Spain), RSSD
– Research and Scientific Support Department (The Netherlands), IDA – Institut f¨ur Datentech-
nik und Kommunikationsnetze (Germany), – DASP Department of Astronomy and Space Physics
(Sweden), UPM – Universidad Polit´ecnica de Madrid (Spain).
identified with the ‘all-seeing eye’ that is depicted in the hieroglyph of his name
(Figure 3).
1.3. FORTHCOMING SECTIONS
In Section 2, an overview of the key questions in cometary physics is presented. This
is followed by a short section that describes the dual camera concept under which
OSIRIS was developed. In Section 4, the detailed scientific rationale and objectives
of the instrument are described. The subsequent sections describe the hardware in
detail. We begin with the optical active elements (Sections 5–7), followed by the
filter wheel mechanisms (Section 8), the shutter systems (Section 9) and the front
door mechanism (Section 10). In Section 11 we deal with the image acquisition
system. In Sections 12 to 16, we describe the overall control electronics, the dig-
ital interfaces, the onboard software, the EGSE and the telemetry. The calibration
and operations are described in Sections 17 and 18. A conclusion completes the
paper.

438 H. U. KELLER ET AL.
Figure 3. The hieroglyph of Osiris from the tomb of Nefertari, Thebes, nineteenth dynasty.
2. The Origin of Comets and Solar System Formation
Cometary missions such as Rosetta derive their greatest intellectual excitement
from their potential to address questions about the origin of the Solar System. In
order to apply data acquired by spacecraft missions to our understanding of these
questions, it is necessary to understand in detail the physical and chemical processes
that might occur in, on, and near the nucleus.
Some of the key problems of the cosmogony of comets and the Solar System
include the nature of the accretion process in the protoplanetary disc, the physical
and chemical conditions (temperature, pressure, molecular composition) that pre-
vailed there, the relationship between the original interstellar composition (both
gaseous and solid) and the disk composition, and the variation of its properties with
both time and heliocentric distance. To derive the maximum scientific return, the
camera system on Rosetta was designed to address as many of these questions as
possible.
The size distribution of planetesimals and the degree to which they come from
different parts of the protoplanetary disc can be studied directly by images from
which the heterogeneity of a cometary nucleus at all scales can be determined.
Images can show the chemical heterogeneity both on the surface and in the material
released from the interior, the structural heterogeneity as seen in activity and in
topography and its changes with erosion, and porosity and its variations as seen
in the bulk density and moments of inertia. Heterogeneity at the largest scales,
from comet to comet, is then studied by comparison of the results from Rosetta

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 439
Figure 4. Conceptual models of the structure of the cometary nucleus: (a) Whipple’s icy conglom-
erate, (b) fluffy aggregate, (c) primordial rubble pile, (d) icy-glue (Weissman, 1986; Donn, 1991).
with results for other comets (such as 1P/Halley, Borelly, Wild 2, and Tempel 1).
This comparison will show whether or not phenomena such as resonances and
instabilities in the protoplanetary disc are important in creating a characteristic
size for planetesimals rather than a broad distribution of sizes characteristic of
agglomeration and collisional phenomena. Our lack of knowledge of the structure
of cometary nuclei is illustrated by the competing models shown in Figure 4. Note
particularly the differences in the scales of the inhomogeneities. These models
are further distinguished by the way in which the building blocks adhere to one
another. This can be studied by determining the relationship between outgassing
and structural inhomogeneities and by analysing the changes in topography and
structure as the comet goes from a nearly inert state to a very active state. It can
also be addressed both by measuring the degree of mixing between refractories and
solids on the surface of the nucleus and by analysing the material released from
the nucleus. Species could be mixed at the microscopic level, at macroscopic levels
that are still small compared to the size of the nucleus, or at scales comparable to
the size of the nucleus. We need to know the scale of mixing in a cometary nucleus
as this can tell us, for example, whether large sub-nuclei with different histories
were brought together in the nucleus.
The physical and chemical composition of the protoplanetary disc can be stud-
ied with calibrated images that provide abundances of species that are sensitive
to those conditions (such as the OH/NH ratio and various mineralogical ratios).
Questions of the nature of physical and chemical variations within the disc can be
addressed by comparisons, both among the components of comet 67P/Churyumov-
Gerasimenko nucleus and among comets formed in different parts of the disc (e.g.
by comparing the properties of a Jupiter-family comet from the Kuiper belt, like

440 H. U. KELLER ET AL.
67P/Churyumov-Gerasimenko, with the properties of a Halley-family comet, like
1P/Halley itself, originally from the Uranus-Neptune region).
It is also necessary to understand the evolution of comets, since the changes
that have occurred over a comet’s active lifetime will have affected the observable
properties of the nucleus. Do comets disappear by gradually shrinking in size as the
ices sublime, do they disintegrate because of the activity, or do they become inert
by choking off the sublimation? Are the intrinsic changes important compared to
the extrinsic changes (collisions, perturbations that dramatically change the orbit,
etc.)? How do comets contribute to the population of interplanetary dust, and how
do they contribute to the population of near-Earth objects? The Rosetta mission and
OSIRIS, in particular, are well suited to study the evolution over a large fraction of
an orbit and to determine the actual contribution per orbital period to interplanetary
dust. They are also well suited to study the evolution of the surface or mantle of
the comet in order to address, for example, the question whether devolatilisation
is more or less important than simple loss of the surface layers. Data obtained by
Rosetta will be compared to those of missions to Near-Earth Asteroids.
Our understanding of the nature and origin of comets, and our use of them as
probes of the early Solar System, is critically dependent upon understanding the
cometary sublimation processes, because this knowledge is needed before we can
relate results from Earth bound remote sensing to the nature of cometary nuclei.
Although many processes in the outer coma, beyond about 100km, are well under-
stood already, the processes at the surface of the nucleus and in the near-nucleus
portion of the coma, closer than a few cometary radii, are poorly understood and
in some cases simply unknown. We need to understand the process by which ma-
terial leaves the nucleus. Are observed variations in the ‘dust-to-gas’ ratio caused
by intrinsic differences in the bulk ratio of refractories to ice, or are the variations
dominated by properties and processes near the surface such as gas flow and struc-
tural strength? Does the size distribution of the particles change in the near-nucleus
region because of either vaporisation or fragmentation or both? What fraction of
the volatiles is released directly from the nucleus and what fraction is released sub-
sequently from particles in the inner coma? Is the gas released from vaporisation
at the surface or at some depth below the surface? Do periodic variations in the
properties of the mantle occur and do they lead to variations in the coma that are,
in fact, unrelated to the bulk properties of the nucleus?
OSIRIS will directly determine the outflow of gas and dust from different re-
gions of the nucleus and will compare those variations with variations in surface
mineralogy, in topography, and in local insolation. This will provide the context
in which to interpret the results from the Rosetta lander (Philae). The unique
strength of OSIRIS is the coverage of the whole nucleus and its immediate en-
vironment with excellent spatial and temporal resolution and spectral sensitivity
across the whole reflected solar continuum up to the onset of thermal emission. In
the next section, we briefly describe the imaging concept of OSIRIS. In the subse-
quent sections, we will address the many, detailed observational programmes to be

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 441
carried out by OSIRIS and how they bear on the fundamental questions outlined
above.
3. NAC and WAC – a Complementary System
During the proposal phase, it was immediately obvious that the scientific objectives
of the camera system on Rosetta would best be served by a combination of a Narrow
Angle Camera (NAC) and a Wide Angle Camera (WAC). The NAC would be a
system with high spatial resolution that would allow an initial detection of the
nucleus, study its structure and rotation from relatively great distances (typically
104km), investigate the mineralogy of the surface, and study the dust ejection
processes. The WAC would have much lower spatial resolution but, accordingly, a
much wider field of view. This would allow observations of the 3-dimensional flow-
field of dust and gas near the nucleus and, in addition, would provide a synoptic view
of the whole nucleus. In summary, the WAC would provide long-term monitoring of
the entire nucleus from close distances, while the NAC would study the details. The
two camera units have therefore been designed as a complementary pair, which,
on the one hand, addresses the study of the nucleus surface, and on the other,
investigates the dynamics of the sublimation process. The resulting cameras have
the basic parameters shown in Table II.
Optical designs with central obscuration are notorious for their stray light prob-
lems. Therefore, off-axis designs with no central obscuration were selected for both
TABLE II
Basic parameters of the NAC and WAC units.
NAC WAC
Optical design 3-mirror off-axis 2-mirror off-axis
Detector type 2k×2kCCD 2k×2kCCD
Angular resolution (μrad px−1) 18.6 101
Focal length (mm) 717.4 140 (sag)/131 (tan)
Mass (kg) 13.2 9.48
Field of view (◦) 2.20 ×2.22 11.35 ×12.11
F-number 8 5.6
Spatial scale from 1 km (cmpx−1) 1.86 10.1
Typical filter bandpass (nm) 40 5
Wavelength range (nm) 250–1000 240–720
Number of filters 12 14
Estimated detection threshold (mV) 21–22 18

442 H. U. KELLER ET AL.
systems. These provide maximum contrast between the nucleus and the dust. The
internal baffle of the cameras was optimised for stray light suppression.
The NAC angular resolution was chosen as a compromise between requests for a
high resolution required for investigation of unknown scale lengths on the nucleus
surface, the need to maintain the nucleus in the FOV of the WAC when only a
few nucleus radii above the surface, and the mass requirements for a longer focal
length system. A spatial resolution of ∼2cmpx
−1was favoured, corresponding to
an angular resolution of ∼20 μrad px−1at a distance of 1 km. This value is also
well adjusted to the limited data volume that can be transmitted back to Earth.
The NAC focal ratio (F-number) was set at 8, which is a compromise between
speed (required at high heliocentric distance, rh) and mass. Extensive calculations
were performed to compute the motion of the image footprint over the surface during
the mapping phase taking into account the orbit of the spacecraft and the rotation
of the target, which would produce image smear. The calculations indicate that
exposure times shorter than 50 ms are probably not required, given the resolution of
the NAC. The WAC observations of the dust and gas environment require narrower
filter bandwidths. Therefore the WAC exposure times are significantly longer.
The major considerations for the CCDs were:
r‘full well’ signal-to-noise ratio (in order to optimise the dynamic range of the
instrument)
rUV response (to give good signal-to-noise ratio for gas species)
rhigh Quantum Efficiency (QE) in the range 800 to 1000 nm (information on
olivine and pyroxene bands).
A2k×2kbackside illuminated detector with a UV optimised anti-reflection
coating was selected. This type of device has high QE over an extended wavelength
range. Full-well dynamic range for these devices is of the order of 2 ×104.Over-
exposure control is needed to allow saturation on the nucleus while acquiring high
signal-to-noise information on the dust and gas. Custom CCDs with lateral anti-
blooming were developed for OSIRIS. For cost reasons, identical devices are used
in the two cameras.
While the two cameras have different scientific objectives, the similar nature of
the instruments naturally led to our seeking cost reduction through development of
identical subsystems. Hence, the mechanical design was adjusted so that identical
Focal Plane Assemblies (FPA) could be used. The large format CCD necessitated
the use of a mechanical shuttering of the exposure. Here again, identical subsystems
were designed.
The requirement to determine the chemical and physical structure of the nucleus
and the inner coma suggested the use of an extensive filter set. Identical filter wheels
were used in the two cameras although each camera had its own filter complement
adapted specifically for its own science goals.
Both cameras need protection from dust impacts when not operating. Hence, they
have doors which can be opened and closed on command. Although the apertures

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 443
ELECTRONICS-BOX
H5
H3
Harness
H1
H2
H6
R
M
M
R
Prim.
Power
S/C I/F
PCM
Controller
PCM
Controller
R
M
DC/DC
Converters
DC/DC
Converters
J3
J4
Processing
Element
Processing
Element
Mass
Memory
Board
DPU I/F
Board
R
M
TC/TM
S/C I/F
M
R
J1
J2
Data
Processing
Unit
Mechanism
Controller
Mechanism
Controller
Mechanism
Controller Board
Power
Conv.
Module
Filter
Wheel 1
Motor
Filter
Wheel 1
Motor
Filter
Wheel 2
Motor
Filter
Wheel 2
Motor
I/F Plate
& FD
Heater
I/F Plate
& FD
Heater
I/F Plate
Non-Op.
Heater
I/F Plate
Non-Op.
Heater
Calibration
Lamps
Calibration
Lamps
NARROW ANGLE CAMERA
R
R
M
M
Front
Door
Motor
(& Fail Sa ve)
Front
Door
Motor
(& Fail Save) R
M
Focal
Plane
Electronics
Shutter
Actuators
& Fail Save
CCD
Operat. &
Annealing
Heater
FPA
MR
R
M
MR
CCD
Non-Op.
Heater
Filter
Wheel 1
Motor
Filter
Wheel 1
Motor
Filter
Wheel 2
Motor
Filter
Wheel 2
Motor
Structure
Heater
1 & 2
Structure
Heater
1 & 2
Structure
Non-Op.
Heater
Structure
Non-Op.
Heater
Calibration
Lamps
Calibration
Lamps
WIDE ANGLE CAMERA
R
R
M
M
Front
Door
Motor
& Fail Save
Front
Door
Motor
& Fail Save R
M
Focal
Plane
Electronics
Shutter
Actuators
& Fail Save
CCD
Operat. &
Annealing
Heater
FPA
MR
R
M
MR
CCD
Non-Op.
Heater
CCD
Readout
Board
CRB
House-
keeping
CRB
Power
Converter
Shutter
Electronics
NAC CRB-BOX
CCD
Readout
Board
CRB
House-
keeping
CRB
Power
Converter
Shutter
Electronics
WAC CRB-BOX
H4
H7
H8
MR
R
M
Figure 5. Block diagram of the OSIRIS functional blocks (M: main, R: redundant).
(and therefore the doors themselves) are different, the drive mechanism is the same
in both cases. In addition, the doors can be used to reflect light from calibration
lamps mounted inside the baffles. The lamps in the NAC and the WAC are identical.
The modular concept of OSIRIS functional blocks, mechanisms, and electronics
subsystems can be seen in Figure 5.
The selection of identical subsystems in both cameras reduced the management
effort, cost, and overall complexity considerably, although interface definition and
specification to accommodate these subsystems was more difficult throughout the
project and required additional spacecraft resources (mass). The flight OSIRIS in-
strument consists of two camera units and 3 electronics boxes with related harnesses,
a total of 22 subsystems, with a total mass of 35 kg and an average operational power
consumption of 34 W.
4. Scientific Objectives
4.1. THE COMETARY NUCLEUS
The imaging systems on the Giotto, Vega, DS-1, Stardust, and Deep Impact space-
crafts were remarkably successful in providing our first glimpses of cometary nuclei
and their immediate environments. Reviews of the results of these investigations
can be found, e.g. in Keller et al. (1995, 2004), Tsou et al. (2004), and A’Hearn

444 H. U. KELLER ET AL.
et al. (2005). Despite this success, the imaging results were limited and many ques-
tions were left unanswered, and additional questions arose, many of which will be
addressed by OSIRIS. We describe here the goals of our nucleus observations.
4.1.1. Position and Size of the Nucleus
The first goal of OSIRIS will be to localise the cometary nucleus and to estimate
its size and shape as quickly as possible for mission planning purposes. These
properties must be coarsely known well before the mapping phase commences.
This determination should be performed near the end of the approach phase when
the spacecraft is between 103and 104km from the nucleus. Determination of the
radius to an accuracy of 10% from 104km can be performed with the NAC and will
immediately yield an estimate of the nucleus volume (and mass for an assumed
density) accurate to about a factor of 2.
4.1.2. Rotational State
Another goal of OSIRIS is to determine the rotational properties of the comet
including the periods of rotation about three principal axes, the total angular mo-
mentum vector L, the changing total spin vector and the characteristics of any
precessional behaviour. Measurements of these quantities will constrain the range
of possible inhomogeneities of the nucleus and will also permit the development
of time-dependent templates over which other data sets may be laid. The sec-
ondary, more ambitious goal is to use OSIRIS to monitor the rotational properties
throughout the entire mission to search for secular evolution in response to the
torques acting on the nucleus caused by the onset of jet activity as the comet ap-
proaches perihelion. Model calculations indicate that for a small nucleus, such
as 67P/Churyumov-Gerasimenko, torques could force re-analysis of the rotational
properties of the nucleus on timescales of days (Guti´errez et al., 2005). The mea-
sured precession rate, along with an estimate of the average reaction force (from the
non-gravitational acceleration of the nucleus) and an estimate of the torque caused
by outgassing, may allow an estimate of the absolute value of the nucleus moment
of inertia. This, in turn, would give clues to the internal density distribution, espe-
cially when combined with the gravity field determination (see P¨atzold et al., this
volume), allowing us to distinguish between a lumpy, a smoothly varying, and a
homogeneous nucleus (see also Kofman et al., this volume). The structural inho-
mogeneity would provide an important clue for the size distribution of the forming
planetesimals.
4.1.3. Shape, Volume, and Density
The concept of comets as uniformly shrinking spherical ice balls was shattered
by the Giotto results. The nucleus is expected to be highly irregular on all scales
as a consequence of cratering, outgassing, and non-uniform sublimation (Keller
et al., 1988). However, it is not clear whether these irregular-shaped bodies reflect

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 445
the shape of the nuclei at their formation, or are the result of splitting during their
evolution, or are caused by non-uniform sublimation.
To accurately model such a craggy shape, techniques developed at Cornell Uni-
versity (Simonelli et al., 1993) can be used. Although OSIRIS has no stereo capa-
bility per se, the motion of Rosetta relative to the nucleus can be used to produce
stereo pairs. The shape model will be based upon these stereogrammetric mea-
surements in addition to limb and terminator observations. Once the shape model
is available, it can be used to determine the surface gravity field and moments of
inertia and will also be used to reproject and mosaic digital images, as well as to
develop surface maps. This technique has yielded accurate shapes for the Martian
moons and Galileo’s asteroid targets, 951 Gaspra and 243 Ida (Thomas et al., 1995,
1994).
To look for internal inhomogeneities of say 30% implies that differences in the
geometrical and dynamical moments of inertia need to be known to better than
10%. We therefore need to measure both to better than 1%. Thus, the topography
must be characterised over the entire nucleus to an accuracy of ±20 m.
4.1.4. Nucleus Formation and Surface Topography
On the smallest scales, the building blocks comprising the cometary nucleus may
be a heterogeneous mixture of interstellar and interplanetary dusts and ices, with a
structure and composition reflecting the physical conditions and chemistry of the
protoplanetary disc. The different accretion processes leading to the production of
first, grains, then, building blocks and, finally, cometary nuclei, are all expected
to have left their mark on a nucleus which has remained largely unaltered since
its formation. OSIRIS will therefore perform a detailed investigation of the entire
cometary surface over a range of spatial scales as wide as possible to identify the
hierarchy of cometary building blocks.
In addition to its implications for nucleus formation, the topography of the
surface determines the heat flow in the uppermost layers of the nucleus (Guti´errez
et al., 2000; Colwell, 1997). High resolution imaging will determine the normal to
the surface and hence provide input to surface heat flow calculations.
The Vega 2 TVS observations of jets were interpreted as showing a fan generated
from a few, kilometre-long, quasi-linear cracks (Smith et al., 1986; Sagdeev et al.,
1987). If fresh cracks appear on the surface during the aphelion passage, then
OSIRIS will be able to probe the inner layers of the nucleus where some stratification
is expected from the loss of volatiles near the surface.
4.1.5. Colour, Mineralogy, and Inhomogeneity
Inhomogeneity of mineral composition and colour could provide the most obvious
clues to the size of building blocks. The Vega and Giotto cameras were able to
determine only rough estimates of the broad-band (λ/λ =5) colour of the nucleus
of 1P/Halley. OSIRIS will allow a much more sophisticated study of the mineralogy

446 H. U. KELLER ET AL.
of the nucleus surface by recording images that span the entire wavelength range
from 250 to 1000 nm.
OSIRIS also has the opportunity to search for specific absorption bands associ-
ated with possible mineral constituents. The wavelengths of pyroxene absorptions
are highly dependent upon their exact structure (Adams, 1974). Hence, filters giv-
ing complete coverage of the 750nm to 1 μm regions at 60 nm resolution were
incorporated. Vilas (1994) suggested that the 3.0μm water of hydration absorption
feature of many low albedo (including C-class) asteroids strongly correlates with
the 700 nm Fe2+→Fe3+oxidised iron absorption feature. Given the spectral sim-
ilarity between C-class asteroids and 1P/Halley and the high water ice content in
comets, a search for the water of hydration feature at 700 nm will be made.
4.1.6. Surface Photometry
Due to the limited information from high-velocity fly-bys, little was learned of the
photometric properties of the surface of comet 1P/Halley. The correct determina-
tion of the phase function for comet 67P/Churyumov-Gerasimenko will provide
information on the surface roughness through application of, for example, Hapke’s
scattering laws (Hapke, 1993). The Philae observations will provide the parameters
necessary to validate the surface roughness models used to interpret global data
provided by OSIRIS.
4.1.7. Polarization Measurements
The properties that can be addressed by polarization measurements can be obtained
more accurately by observations of the surface from the Philae or by in situ analysis.
Implementation of polarization measurements in OSIRIS was thought costly in
terms of resources and calibration, and they were therefore not included.
4.1.8. Active and Inactive Regions
Modelling (K¨uhrt and Keller, 1994) suggests that debris from active regions will
not choke the gas and dust production in view of the highly variable terrain, the
extremely low gravity, and the lack of bonding between particles forming the debris.
Inactive regions can only arise if either the material comprising the regions formed
in the absence of volatiles or, alternatively, if the regions have become depleted in
volatiles without disrupting the surface.
To verify this picture, a comparison of active and inactive regions on comet
67P/Churyumov-Gerasimenko must be of high priority. If inactive regions are
merely volatile-depleted with respect to active regions, high signal-to-noise ob-
servations at several wavelengths may be required to differentiate between the two.
Imaging of the interface between active and inactive regions may provide evidence
of surface structures and tensile strength present in one type of region, but not in
the other.
As the observations of ‘filaments’ indicate (Thomas and Keller, 1987), there
is no reason to suppose that active regions are homogeneous. Activity may be

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 447
restricted within an active region (see for example the theoretical calculations of
Keller et al., 1994). For this reason, OSIRIS will identify the active fraction within
what we call an active region. Achieving this goal may lead to understanding how
cometary activity ceases, leaving an inert comet. Do inactive spots within active
regions spread to reduce cometary activity or does the infall of material from the
edge of the sublimation crater choke emission?
Directions of jets and locations of active spots are influenced by topography. The
distribution and orientation of near-nucleus jets can be used to infer topographic
features (Thomas et al., 1988; Huebner et al., 1988). OSIRIS will investigate these
correlations.
4.1.9. Physics of the Sublimation Process
The physical processes characterising the sublimation and erosion processes in
or above active regions depend on the physical structure of the surface and the
distribution of refractory and volatile material within the nucleus. Dust particles
have usually been treated as impurities in the ice (icy conglomerate). Starting with
the interpretation of the images of 1P/Halley (Keller, 1989), it has become clear that
the topography requires a matrix dominated by refractory material (K¨uppers et al.,
2005). The other extreme is the model of a friable sponge, where the refractory
material is intimately mixed with the ice and where the erosion process maintains a
balance between the ice and dust. How are dust particles lifted off the surface? The
excellent resolution of the OSIRIS NAC, which will be smaller than the mean free
path of the gas near the surface, will allow the detection and study of the relevant
macrophysical processes.
4.1.10. The Diurnal Cycle
OSIRIS will be able to monitor short-term changes in active regions very easily.
Changes are most likely when active regions cross the terminators. Cooling will lead
to decreased activity, but on what timescale? On the other hand, as the insolation
increases, will there be changes in the surface structure?
4.1.11. Outbursts
Outbursts (or rapid increases in the brightness of cometary comae) have frequently
been observed from the ground and recently also during the approach of the Deep
Impact spacecraft to comet 9P/Tempel 1. This implies some sudden increase or
even explosion of activity ripping the surface crust apart. OSIRIS, and in particular
the WAC, can be used to monitor autonomously the nucleus activity over many
months at various scales. The NAC can then be used to look in detail at the source
to determine how the site has altered topographically and spectrally.
4.1.12. Mass Loss Rate
The floor of the active regions will be lower by several metres on average after the
passage of comet 67P/Churyumov-Gerasimenko through its perihelion. It is clear

448 H. U. KELLER ET AL.
that if an active area can be monitored by OSIRIS at a resolution of ≈30 cm the
mass loss will be evident. If the density of the surface layer can be determined by
Philae or through joint OSIRIS/Radio Science investigations, this is potentially the
most accurate means to determine the total mass loss rate particularly if the mass
loss is dominated by infrequently emitted large particles.
4.1.13. Characterisation of the Landing Site
The NAC was designed to remain in focus down to 1 km above the nucleus surface.
Mapping at ∼2cmpx
−1will reveal inhomogeneities of the nucleus at scale lengths
comparable to the size of Philae. Homogeneous sites would provide no difficulties
in interpretation but heterogeneous sites may be scientifically more interesting. As
a result, OSIRIS needs to be able to characterise the landing site and to identify on
what types of terrain Philae has landed.
4.1.14. Observation of the Philae Touchdown
There is no guarantee that the orbiter will be able to observe Philae when it strikes the
surface. However, OSIRIS will provide valuable information on the impact velocity,
the result of the initial impact, and the final resting position and orientation.
Outgassing from the impacted site may also occur. If fresh ice is so close to
the surface that the lander can penetrate the crust, emission of gas and dust may
be fairly vigorous. If so, OSIRIS can quantify this emission with highest possible
spatial and temporal resolution.
4.2. NEAR-NUCLEUS DUST
The near-nucleus dust environment of a comet is remarkably complex and remains
poorly understood. Understanding the near-nucleus environment is necessary to
understanding the nucleus itself. OSIRIS can investigate global dust dynamics.
4.2.1. Detection of Emission at Rendezvous
OSIRIS will be used to place constraints on distant activity of the nucleus. It
is evident, however, that detection of dust in the vicinity of the nucleus will be
extremely difficult at high heliocentric distances. The dust production may decrease
as steeply as r−2.9
h(Schleicher et al., 1998), with a corresponding decrease in flux
proportional to r−4.9
h. At 3.25 AU, we estimate the ratio of the signal received from
the dust to that from the surface (Id/Is)≈4×10−4based on scaling of Giotto
measurements. Therefore, to quantify the total dust production rate, a dynamic
range of >2000 is required. Both the WAC and the NAC were designed with this
contrast requirement.
4.2.2. Temporal Evolution
4.2.2.1. Variation with Heliocentric Distance. HMC observations showed that
the dust production rate of comet 1P/Halley during the Giotto fly-by was remarkably

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 449
stable over the three hours of the encounter. Ground-based observations have shown,
however, that comets exhibit large and rapid changes in dust production. A key goal
of OSIRIS will be therefore to monitor the variation in the production rate and to
compare it to the rotational characteristics of the nucleus and the change in rh.
4.2.2.2. Variations with Rotation. The lack of significant variation in the dust
production rate with the rotation seen at comet 1P/Halley was not expected. How
does the production vary with the solar zenith angle? How long does an active region
take to switch on after sunrise? These phenomena are determined by the physical
properties (e.g. the thermal conductivity) of the surface layer. If sublimation occurs
below the surface then a period of warming may be required before dust emission
starts. The surface layer could act as a buffer to stabilise the activity. These questions
can be addressed using OSIRIS to monitor the active region during the first minutes
after it comes into sunlight.
4.2.2.3. Night Side Activity and Thermal Inertia. The inferred absence of night
side activity during the Giotto fly-by and the thermal map created from near-
infrared spectral scans of comet 9P/Tempel 1 during the recent Deep Impact mission
(A’Hearn et al., 2005) suggest that the thermal inertia must be low. Observations
of comet Hale–Bopp (C/1995 O1) also suggest that the thermal inertia of comets is
low (K¨uhrt, 2002). The high porosity of the surface and the resulting low thermal
conductivity suggest that the activity should decrease rapidly and stop when the
energy source is removed. Monitoring the dust emission as an active region crosses
the evening terminator can confirm this hypothesis.
4.2.2.4. Short-Term Variability. The dust emission from the nucleus of comet
1P/Halley showed no evidence for short-term (order of minutes) temporal varia-
tions. Because of the nature of the active regions one might expect, however, that
the emission should occasionally show an enhanced or reduced rate on a timescale
of perhaps a few seconds. A sudden burst offers the possibility of following the
emitted dust and using it to derive streamlines and velocities in the flow. This obser-
vation would provide strong constraints on the hydrodynamics of the flow and lead
to increased understanding of the dust-gas interaction a few metres above active
regions. If large enough, outbursts could also modify the flow field itself allowing us
to use OSIRIS to monitor the reaction of the inner coma to changes in the emission
rate.
4.2.3. Large Particles in Bound Orbits
It was shown that gravitationally-bound orbits around cometary nuclei are possible,
in theory, for relatively small particles even in the presence of radiation pressure
(Richter and Keller, 1995). In addition, evidence from radar measurements suggests
that large clouds of centimetre-sized objects accompany comets in their orbits
(Campbell et al., 1989). The high resolving power of OSIRIS combined with our

450 H. U. KELLER ET AL.
proximity to the nucleus will allow us to place constraints on the number density
of objects with a particle radius of a>5 mm. Since it is now widely believed that
most of the mass lost by comets is in the form of large particles (McDonnell et al.,
1991), observations of this phenomenon could prove very important in determining
the dust to gas ratio. Clearly, it would be a major discovery to find an extremely
large chunk which might be termed ‘a satellite’ of the nucleus. Active chunks, as
seen in comet Hyakutake (Rodionov et al., 1998), may also be evident.
4.2.4. How Inactive are ‘Inactive’ Regions?
The observations by HMC and more recent fly-bys (A’Hearn et al., 2005) were not
good enough to place firm constraints on the activity of so-called inactive regions.
Dust emission from the illuminated but apparently inactive regions could have been
up to 10% of the emission from active regions and remained undetected. This clearly
has implications for the evolution of the nucleus and for the flow field of gas and
dust emission about the nucleus.
4.2.5. Optical Properties of the Dust
The orbit of Rosetta and the broad-band filters in OSIRIS will allow observations of
dust at many phase angles (0◦–135◦) over a wide wavelength range. The phase curve
and colour are sensitive to particle size, composition, and roughness. Deduction of
these properties and their variation with rhwill be important for ground-based
observations of other comets since it will provide the single scattering albedo, the
phase function, and the characteristic particle size.
4.2.6. Eclipses
Eclipse measurements are extremely interesting for the innermost dust coma as they
would allow OSIRIS to determine the forward scattering peak of the dust phase
function, which provides the best information on the size distribution and nature of
the dust particles. The strong forward scattering peak also yields the most sensitive
measurement of the dust column density (e.g. Divine et al., 1986).
4.2.7. Acceleration and Fragmentation
Complications with the determination of local dust production rates arise if the
observations cover the dust acceleration region, if fragmentation is significant, or
if optical depth effects become important. Measurements of the acceleration will
quantify the drag coefficient of the gas-particle interaction and characterise the
near-surface Knudsen layer. The fulffiness of the cometary dust can be derived
from these observations.
A complementary approach is to measure the radius and velocity of large es-
caping dust agglomerates in dependence of heliocentric distance. By knowing the
gravitational forces, this would also provide information on the physics of the gas-
dust interaction (drag coefficient) at and near the surface (Knudsen layer), on the
cohesive forces, and on the density of the agglomerates.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 451
4.3. GAS EMISSIONS
Our current understanding of the composition of the nucleus and variations within
the nucleus is severely limited by our lack of knowledge about the processes in
the innermost coma. We know little about the variations of the composition of the
outgassing on any scale, although there are indications from Earth-based measure-
ments of large-scale heterogeneity (e.g. in 2P/Encke and in 1P/Halley). There are
distributed sources in the coma which produce some of the species in the coma, in-
cluding H2CO, CO, and CN. Because we cannot separate completely the extended
coma source from the nuclear source, we cannot determine reliably the amount of
ice in the nucleus. We therefore plan to make observations of the gas in order to
address some of the most crucial questions in relating abundances in the coma to
abundances in the nucleus.
4.3.1. Selected Species
In order to constrain the heterogeneity of other parent molecules, we will map
the release of certain daughter species in the vicinity of the nucleus. Dissociation
products having short lifetimes and identifiable parents are ideal for this task. In
particular, NH at 336.5nm and NH2at 570 nm will be measured to trace the hetero-
geneity of NH3(and thus the nitrogen chemistry in the nucleus), CS at 257 nm to
trace the heterogeneity of CS2(and thus the sulphur chemistry), and OH at 309 nm
and OI at 630 nm to trace H2O.
The heterogeneity of other fragments, such as CN (388 nm), will also be mea-
sured, even though we do not know the identity of the parent molecules, because
these species show evidence of an extended source. The recent interest in the dis-
tribution of Na has led us to introduce a sodium filter at 589 nm.
4.3.2. Sublimation Process and Inactive Areas
The results from 1P/Halley showed us that the release of dust is confined to discrete
active areas, comprising only a small fraction of the surface (15%). We have no
information, however, on whether the gas is similarly confined. One of the key
questions to be answered is whether gas is also released from the apparently inactive
areas. The mapping capability of OSIRIS is ideally suited to answer this question
and thereby to assess the effects of an inert layer on the release of gas and dust.
4.4. SERENDIPITOUS OBSERVATIONS
4.4.1. Asteroid Fly-bys
The fly-bys of 2867 Steins and 21 Lutetia will provide interesting secondary targets
on the way to the comet. The main scientific goals of OSIRIS observations of the
asteroids are:
rDetermination of physical parameters (size, volume, shape, pole orientation,
rotation period)

452 H. U. KELLER ET AL.
rDetermination of surface morphology (crater abundance, crater size distribu-
tion, presence of features such as ridges, grooves, faults, boulders, search for
the presence of regolith)
rDetermination of mineralogical composition (heterogeneity of the surface,
identification of local chemical zones, superficial texture)
rSearch for possible gravitationally bound companions (detection of binary
systems).
4.4.2. Mars Fly-by
High-resolution images of Mars (>200 px across the planet) can be taken within
two days of closest approach (cf. recent HST images). This will provide data on the
global meteorological conditions on Mars and allow us to follow weather patterns
over a period of about two days. Images around 12h before closest approach would
be of sufficient resolution to allow us to resolve vertical structures in the atmosphere
at the limb and to estimate the global atmospheric dust content. The solar occultation
during Mars fly-by would allow detection of the putative Martian dust rings.
4.4.3. Earth-Moon System Fly-bys
As with the space missions Galileo and Cassini/Huygens, the Rosetta remote sens-
ing instruments can perform testing and calibration during the fly-bys of the Earth-
Moon system. There are also several interesting possibilities for new science. For
example, the Moon is now known to have a tenuous sodium atmosphere (‘exo-
sphere’). The Na filter on the WAC can be used to acquire maps of Na near the
Moon. Similarly, OI emission from the Earth may be detectable at high altitudes.
Vertical profiles of OI in the atmosphere of the Earth can be derived by stellar
occultations.
5. The NAC Telescope
The Narrow Angle Camera is designed to obtain high-resolution images of the
comet at distances from more than 500,000 km down to 1 km, and of the asteroids
2867 Steins and 21 Lutetia during the interplanetary cruise. The cometary nucleus
is a low-albedo, low-contrast object; hence, good optical transmission and contrast-
transfer characteristics are required. The camera also should be able to detect small
ejected particles close to the comet nucleus (brightness ratio ≥1/1000), placing
strict tolerances upon stray light rejection.
The scientific requirements for the NAC translate into the following optical
requirements. A square field of view (FOV) of width 2.2◦and an instantaneous
field of view (IFOV) of 18.6 μrad (3.8 arcsec) per pixel, a spectral range from
250 nm to 1 μm, and a moderately fast system (f/8) are needed. An unobstructed
pupil is required to minimise stray light. This is particularly important for the study
of gas and dust surrounding the bright nucleus. The requirements are fulfilled with

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 453
an all-reflecting system of 717 mm focal length and an off-axis field, using a 2048
×2048 px, UV-enhanced CCD array. The high resolution over a large flat field
requires a system of three optical surfaces.
5.1. OPTICAL CONCEPT AND DESIGN
A flat-field, three-mirror anastigmat system, TMA, is adopted for the NAC. Anas-
tigmatism (freedom from third-order spherical aberration, coma, and astigmatism)
is attained by appropriate aspheric shaping of the three mirror surfaces, and a flat
field (zero Petzval sum) is achieved by appropriately constraining the system ge-
ometry. Our solution (Dohlen et al., 1996) has an axial pupil physically placed at
the second mirror M2, an off-axis field of view, appropriate baffle performance
and a large back-focal clearance. The optics requires only two aspheric mirrors,
the tertiary remaining spherical. This considerably reduces fabrication cost and
alignment difficulty. The three mirror surfaces are rotationally symmetric about a
common optical axis, but the field of view is sufficiently removed from the axis to
ensure that all rays pass through the system without vignetting. Figure 6 shows a
ray tracing diagram of the optical system. The mirrors are made of Silicon Carbide
(SiC); details of their fabrication, polishing and alignment can be found in Calvel
et al. (1999).
The system is equipped with two filter wheels placed in front of the CCD. In
order to cope with the presence of ghost images (see Section 7.2.3), the filters are
Figure 6. Ray paths through the NAC optical system.

454 H. U. KELLER ET AL.
tilted by 4◦to the optical axis and wedged by 10. In addition to the bandpass
filters, the filter wheels contain anti-reflection-coated focusing plates (Far Focus
Plate FFP and Near Focus Plate NFP, see Table V), which, when used with the
filters of the other wheel, allow two different focusing ranges: far focus (infinity
to 2 km, optimised at 4 km) and near focus (2 km to 1 km, optimised at 1.3 km).
Nominal operation is defined as far focus imaging with an orange filter (centered
at 645 nm with a bandwidth of 94nm). This filter has similar characteristics to that
of the orange filter in the Halley Multicolour Camera.
A plane-parallel, anti-reflection coated plate, referred to as Anti-Radiation Plate
(ARP), was added to the front of the CCD for radiation shielding. Its effect for
monochromatic light is negligible, but the shift of focus is considerable for the two
UV filters (Far-UV and Near-UV), and the Far-UV and focusing plates are affected
by longitudinal chromatic aberration. Table III lists the construction parameters
for the optimised camera design, including filter, focusing plate and ARP. The
system includes an external baffle for stray light rejection and a front door for
protection.
5.2. OPTICAL PERFORMANCE
Figure 7 shows spot diagrams and root-mean-square wave front errors (WFE) at six
points in the FOV located at the centre, the edges and the corners. Since the system
is symmetrical about the y–zplane (see footnote in Table III), the characteristics are
identical for positive and negative xco-ordinates. The wave front error is calculated
for the central wavelength of the orange filter (λ=0.645 μm). As seen in Figure 7,
the WFE is in the order of 0.04 λover the entire FOV. The performance is limited
primarily by a triangular-type (trifle) aberration which is present in varying degrees
over the entire FOV. Astigmatism and coma are close to zero at the centre but
become significant towards the edges.
5.3. STRAY LIGHT REJECTION
The observation of faint cometary physical and chemical phenomena, such as dust
and gas jets from localised vents on the nucleus, require good optical transmission
and high contrast with strict tolerances on stray light. There are two types of stray
light sources. One originates from the cometary nucleus itself (considered as an
extended object), the image of which is in the focal plane. The second source is
the sun, which is allowed to reach an elongation of 45◦from the optical axis of the
instrument.
Rejection of stray light from the nucleus is insured by the TMA design whose
unobstructed pupil minimises diffraction phenomena and scattered light. A low
level of micro roughness of the optical surfaces (<2nm rms) was specified to limit
the stray light contribution. The internal baffle of the instrument was optimised

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 455
TABLE III
Optical specifications for the NAC TMA; Far Focus Plate in wheel 1, Orange filter in wheel 2.
Radius Separation along Conic constant Diameter or Centre yTilt of normal wrt.
Component (mm) zaxisa(mm) or glass type x×y(×z)a(mm) coordinate (mm) M2 axis (deg)
Entrance pupil Flat −690.0 89.4 −6.962
Primary −1103.162 −304.0 CC =−1.65727 136 85.2b0
Secondary −400.598 301.5 CC =−1.000 44.0 0 0
Tertiary −626.583 −354.5 CC =0.000 92 −82.9b0
Filter 1, surface 1 Flat −4.942cFused silica n(645 nm) =1.4567 40 ×40 (×5.000) −86.00 −3.8333
Filter 1, surface 2 Flat −5.00c−4
Filter 2, surface 1 Flat −4.687cOG550 n(645 nm) =1.5368 40 ×40 (×4.746) −4
Filter 2, surface 2 Flat −25.724 −4.1667
ARP, surface 1 Flat −12 Fused silica n(645 nm) =1.4567 51.0 −86.15 0
ARP, surface 2 Flat −8.056 0
CCD Flat 27.7 ×27.7 −86.32 0
azis along the M2 axis, yis in the plane containing zand the central object point, xis perpendicular to zand y.
bGeometrical centre of the mirrors. Impact of ‘central’ ray is slightly offset (∼0.5 mm).
cDistance along an axis tilted −4◦to z-axis, perpendicular to filter wheel plane.

456 H. U. KELLER ET AL.
Figure 7. Spot diagrams for six positions in the NAC FOV. The cross is 5 μm wide. RMS wavefront
errors are given for the centre wavelength of the Orange filter.
by adding vanes and protective black tapes in critical areas. An internal black foil
envelops the whole instrument to prevent light leakage.
Rejection of stray light from the sun requires an external baffle, which can be
closed with the front door whenever the solar elongation is less than 45◦. The
fraction of the power incident onto the detector surface to that entering the aperture
of the telescope baffle was required to be <10−9at angular distances from the centre
of the FOV exceeding 45◦. The external baffle comprises a two stage cylinder with
four vanes, which have a square aperture with rounded corners to fit around the
scientific beam. The baffle is made of aluminium alloy and all internal surfaces are
coated with a black paint. The vanes have a thickness of 0.5mm with sharpened
edges in order to reduce the reflecting area.
5.4. NAC STRUCTURE
The main function of the structure of the NAC is to carry the three mirrors of the
TMA, the dual filter wheel mechanism, shutter mechanism, focal plane assembly,
the external baffle, the front door mechanism (Figure 8), and to maintain them
in proper position during the long interplanetary cruise and the phases of nucleus
observations.
The basic concept is an athermal design, achieved by using the same ma-
terial for the mirrors and the supporting structure. Therefore the optical prop-
erties are maintained during temperature changes, as long as thermal gradients
are limited. Silicon carbide, a very rigid ceramic material with good thermal

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 457
Figure 8. The NAC assembly (see also Figure 9). Focal plane and front door are on the upper left,
primary and tertiary mirrors on the lower right.
properties (low coefficient of thermal expansion, good thermal conductivity), is
used.
The structure is U-shaped with two walls cemented on a connecting tube, see
Figure 9. The main wall carries the external baffle, the secondary mirror and a
magnesium interface plate (I/F plate), which receives the mechanisms and the focal
plane. The second wall carries the primary and the tertiary mirrors, which are bolted
directly on it. The thickness of the two walls is reduced to the minimum feasible in
order to minimise mass. The NAC structure is kinematically mounted on the main
spacecraft structure via three titanium bipods, decoupling it from panel distortion
and reducing thermal flows.
5.5. THERMAL DESIGN
The athermal concept requires that thermal gradients be minimised. A thermal
decoupling is necessary between the SiC parts constituting the telescope and the
subsystems, which have to be maintained within their temperature ranges. The I/F
plate is connected to the main SiC wall via three flexible titanium blades and in-
sulator washers. The entrance baffle and the Front Door Mechanism (FDM) are
connected to the main SiC wall via three insulator washers. In order to min-
imise thermal losses, the instrument is wrapped in a thermal blanket made of
Multi-Layer Insulation (MLI). The strong limitation on available power during
the cruise phase required a careful optimization of the thermal design. With a

458 H. U. KELLER ET AL.
Figure 9. NAC Flight Model during integration and in flight configuration.
non-operational power of 7.5 W, the subsystems are maintained in their allowed
temperature ranges, while the SiC telescope can float to a minimum temperature of
−70 ◦C.
5.6. INTERNAL CALIBRATION
The internal calibration of the NAC is achieved by illuminating the rear side of
the lid of the front door, which acts as a diffusing screen. The illumination system
is composed of two redundant sets of two small lamps placed inside the external
baffle, between two vanes. The four lamps form a rectangle but only two lamps on
one side are lit at a time. The colour temperature of the tungsten lamps is 2,410 K;
they have a quartz envelope and are mechanically mounted with a glass diffuser.
The in-flight calibration system provides a reference illumination to the camera

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 459
in cruise for comparison to the ground calibration flat fields. Deviations allow the
determination of long-term degradation in flight.
6. The WAC Telescope
The prime objective of the Wide Angle Camera is the study of the weak gas and
dust features near the bright nucleus of the comet. For this purpose, the WAC has
to satisfy a number of scientific requirements. The WAC needs a rather large field
of view, 12◦×12◦, to observe both the nucleus and the features of emitted gas and
dust. It has to cover a relatively wide spectral range, from UV to visible, and it has
to provide a high contrast ratio, of the order of 10−4, to be able to observe the bright
nucleus and the weak coma simultaneously.
6.1. OPTICAL CONCEPT AND DESIGN
To obtain the required camera performance, an unobstructed all-reflective, off-axis
optical configuration using two aspherical mirrors was adopted. With this system,
best performance over the entire field of view is obtained, providing a spatial
resolution of about 20 arcsec. The all-reflective solution, unlike a lens design, allows
observation in the ultraviolet spectral range. The unobstructed solution provides
the optimal contrast ratio. With the 20◦off-axis design, the whole field of view
can be covered without significant aberrations. Moreover, a fast f/5.6 ratio was
adopted to allow detection of the cometary nucleus and of the asteroids from a
distance of 106km in 1 s exposure time. The characteristics of the optical solution
are summarised in Table IV.
6.2. TWO-MIRROR OFF-AXIS SYSTEM
The concept of the WAC optical design is shown in Figure 10; it is described in more
detailed in Naletto et al. (2002). The primary mirror (M1) collects the light from
the object at an angle of 20◦with respect to the camera axis and reflects it towards
the secondary mirror (M2), which focuses the light onto the focal plane assembly.
In contrast to the majority of all-reflective cameras, in which the instrument stop is
located on the primary mirror, the WAC stop coincides with M2. Figure 11 shows
the flight mirrors prior to their assembly. To assure a good reflectivity over the
whole spectral range, the mirrors were aluminized and protected with MgF2.
The instrument design includes a set of 14 wedged filters. The filters are mounted
on two wheels. The filter set comprises narrow and wide bandpasses for observations
of emission bands and lines, and of the continuum. A 4 mm thick Suprasil Anti-
Radiation Plate (ARP) was installed directly above the CCD.

460 H. U. KELLER ET AL.
TABLE IV
Characteristics of the WAC design.
All reflective, two-mirror 20◦off-axis design,
Optical concept unobstructed, unvignetted; axis of off-axis: y
FOV 11.35◦(y)×12.11◦(x)
Encircled energy >80% inside a pixel of 13.5 μm sq.
FL at centre of FOV Tangential: 131mm, sagittal: 140 mm
Average image scale at the centre of FOV 7.4 mrad mm−1equiv. 101 μrad px−1
Distortion of the field Barrel type
FL in x128.0–133.4 mm
FL in y139.2–140.0 mm
Residual geometric distortion <2.5% along the tangential direction <1% along the
sagittal direction
Refocusing No refocusing necessary for distance range 500 m to
infinity
Nominal F-number F/5.6
Wavelength range 240–720 nm
Overall reflectivity 52% at λ=245nm to 88% at λ=500 nm
Figure 10. WAC optical concept with the two mirrors, the filter and the anti-radiation plate in front
of the CCD detector.
The sun angle is greatly variable when in orbit around the comet nucleus. To
protect the system from direct sun light, a movable front door is positioned at the in-
strument entrance aperture. Moreover, to reduce the stray light into the instrument, a
rather complex baffle system was realised (Debei et al., 2001; Brunello et al., 2000).

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 461
Figure 11. WAC mirror blanks M1 (left) and M2 (right).
6.3. OPTICAL PERFORMANCE
Figure 12 shows the spot diagrams at the centre, the edges, and the corners of the
FOV. The square boxes correspond to the pixel size. The geometrical performance
of the camera is optimised with residual aberrations essentially lower than the pixel
size. The diffraction effect is negligible, with more than 90% of the energy falling
into a single pixel. The optical performance is maintained essentially unchanged
from infinity down to almost 500 m, so that no refocusing system is required.
Figure 12. WAC spot diagram. The spots refer to the centre, the edges and the corners of the 12◦×
12◦FOV. The boxes correspond to the pixel size of the detector.

462 H. U. KELLER ET AL.
The off-axis design produces slightly different scales in the image plane: the scale
in yis almost constant, 19.9 arcsec px−1, while the scale in xdirection varies from
20.9 arcsec px−1to 21.8 arcsec px−1. The acquired field width seen by the CCD
is subsequently about 11.35◦in yand between 12.09◦and 12.16◦in xdirection
(coordinate system see Figure 10).
The nominal photometric aperture, that is the projection of the M2 stop onto the
primary mirror, is circular with a radius of 12.5mm. However, the distortion causes
the aperture to be slightly elliptical and position dependent. The corresponding
photometric distortion can be removed during calibration of the images.
6.4. STRUCTURE OF THE CAMERA
The lightweight, stiff structure is based on a closed box made of aluminium alloy
machined by electro-erosion. The optical bench ribs are optimised to prevent noise
induced through vibration and to minimise vibration amplification at interfaces
with mechanisms. For thermo-structural stabilization, three kinematic mounting
feet and an external baffle are implemented. A truss structure was designed to
improve the thermal decoupling between the external baffle and the optical bench,
and to minimise the temperature gradient. The telescope is covered by a thermal
blanket, while the inner parts were painted with electrically conductive black paint.
The optic supports are made of the same material as the optical bench to minimise
distortion.
6.5. THERMAL BEHAVIOUR
The WAC thermal control system was designed for the operational temperature
range of the optical bench, 12 ±5◦C. This requirement is derived from the tolerances
in the position of the telescope optical elements.
The total electrical power dissipated into the camera is less than 2 W. The heat
leak of the Focal Plane Assembly (FPA) is in the order of 1 W. With only 1 W to
dissipate, the camera should be thermally insulated from the environment. Thus,
the whole external surface was covered with MLI and the WAC was mounted to the
spacecraft with insulating feet made of a titanium alloy (Figure 13). The strongest
disturbance to the WAC thermal control is due to the large optical aperture, an area
of about 300 cm2, which can point towards many different thermal sources.
The external baffle is the most critical element of the WAC thermal design. A
trade-off analysis was performed. A glass reinforced epoxy structure with absorber
coating, thermally insulated from the camera, was shown to be the best solution.
To extend the operational capability for sun incidence angles below 45◦, a radiator
was added to the upper part of the baffle to reduce the heat flux to the camera.
Below 45◦, operation of the WAC is intermittent, with observational phases until
the allowed upper temperature limit is reached, followed by phases of cooling with
closed door.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 463
Figure 13. WAC Flight Model ready for integration.
The other important requirement for the camera thermal control system is to
provide a temperature greater than −40 ◦C during the non-operative phases. This
is achieved using a heater dissipating 5W.
6.6. BAFFLE SYSTEM AND STRAY LIGHT PERFORMANCE
The primary scientific requirement for the WAC is to be able to image dust and
gas in the proximity of the comet nucleus at heliocentric distances from 3.25 AU to
perihelion. The WAC baffle system has therefore to perform two different functions:
First, to attenuate light from any source, e.g. the sun at angles larger than 45◦,toat
least 4 ×10−9(at the detector). The second function is related to the characteristic
of the WAC optical design with the system stop located at the second mirror M2.
Light entering the system aperture and reaching M1, not being collected by M2,
acts as internal source of stray light. The baffle has to attenuate this stray light
contribution by a factor of at least 10−3.
The baffle system is made of two main parts: the first (external baffle) with
rectangular cross section is localised in front of the M1 mirror and has 17 vanes,
the second (internal baffle) is accommodated between M2 and the detector, and has
4 deep vanes (see Figure 14).
6.7. CALIBRATION LAMPS
The calibration lamps used in the WAC are identical to the NAC lamps. They are
flame-formed bulb lamps with a colour temperature of 2410 K. Four lamps (two

464 H. U. KELLER ET AL.
Figure 14. Concept and view of the WAC baffle.
main and two redundant) are mounted at the fifth vane of the external baffle. The
lamps illuminate the inside of the front door, which diffuses the light into the
optical path. Obviously, this illumination cannot provide a flat field, but a reference
illumination pattern monitoring the system transmission.
7. Interference Filters
Sets of 12 filters for the NAC, and 14 for the WAC, were selected. The NAC filters
will be used to characterise the reflectivity spectrum of the nucleus surface over
as wide a spectral range as possible, and to focus in particular on some possible
or likely absorption bands. With no need for isolating narrow spectral features, the
bandpasses are generally wider than for the WAC, i.e. typically from 24 to 100 nm.
For the WAC, the principal aim is to study the intensity of gas emissions and
dust-scattered sunlight as functions of position and viewing angle in the vicinity of
the nucleus. This is accomplished by centring narrow bandpass filters on a set of
emission lines with slightly broader bandpass filters to measure the continuum.
7.1. SELECTED FILTERS
7.1.1. NAC Bandpass Filters
The 12 selected filters for the NAC are shown in Table V. The NAC filters were
optimised to provide a low resolution spectrum from 250nm to 1 μm. The or-
ange filter at 650 nm will allow a close comparison of the results from comet

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 465
TABLE V
Filters of the narrow angle camera.
Wavelength Bandwidth Peak Trans. Thickness at
Name (nm) (nm) (%) Objective centre (mm) Wheel
FFP-UV 250–850 >99 UV focusing plate for use of filters in wheel 2 4.41 1
FFP-Vis 250–1000 >95 Vis focusing plate for use of filters in wheel 2 5.00 1
FFP-IR 300–1000 >99 IR focusing plate for use of filters in wheel 1 5.15 2
NFP-Vis 300–1000 >98 Vis focusing plate for near-nucleus imaging 4.18 1
Far-UV 269.3 53.6 37.8 Surface spectral reflectance 4.50 2
Near-UV 360.0 51.1 78.2 Surface spectral reflectance 4.68 2
Blue 480.7 74.9 74.6 Surface spectral reflectance 4.67 2
Green 535.7 62.4 75.8 Surface spectral reflectance 4.64 2
Neutral 640.0 520.0 5.0 Neutral density filter 4.64 1
Orange 649.2 84.5 92.4 surface spectral reflectance 4.73 2
HMC orange filter
Hydra 701.2 22.1 87.4 Water of hydration band 4.72 2
Red 743.7 64.1 96.0 Surface spectral reflectance 4.68 2
Ortho 805.3 40.5 69.8 Orthopyroxene 4.69 1
Near-IR 882.1 65.9 78.4 Surface spectral reflectance 4.75 1
Fe2O3931.9 34.9 81.6 Iron-bearing minerals 4.73 1
IR 989.3 38.2 78.1 IR Surface reflectance 4.74 1

466 H. U. KELLER ET AL.
67P/Churyumov-Gerasimenko with those obtained from comet 1P/Halley with
HMC. A cluster of filters was placed in the wavelength range between 800nm
and 1 μm to investigate possible pyroxene and olivine absorptions. A neutral den-
sity filter was added to reduce the photon flux from the comet in the event that
very bright, pure ice structures are revealed by activity near perihelion. The neutral
density filter can also be used if the shutter fails and will be used for resolved
observations of the Earth and Mars.
Clear filter substrates with anti-reflection coating (cf. near and far focusing
plates) were included to modify the focus position by adjustment of the optical
thickness so that the NAC remains in focus down to a distance of just 1 km from the
target. The thickness of each bandpass filter was chosen individually (dependent
upon wavelength and substrate refractive index) to maintain the system in focus.
7.1.2. WAC Bandpass Filters
The 14 selected filters for the WAC are shown in Table VI. Most of the filters are
narrow band filters to study gas and radical emissions. The minimum filter band-
width allowed by the f/5.6 optical design is 4 nm, because narrower bandpass filters
would produce variations in the transmitted wavelength over the field. Continuum
filters were incorporated to allow straightforward subtraction of the dust continuum
from images acquired in gas emission filters. Calculations indicate that high signal-
to-noise ratios in CN, OH, OI, and CS will be easily achieved. Na, NH, and NH2
should be detectable in binned data within 1.2 AU from the Sun. A broad-band
R filter was included for nucleus detection and mapping, in the event of failure
of the NAC. A green filter, identical to that in the NAC, was included for simple
cross-correlation of the data between NAC and WAC. No refocusing capability is
required for the WAC.
7.2. ORIENTATION AND PROPERTIES
7.2.1. Materials and Radiation Tolerance
Radiation tests were performed to ensure that the performance of the filters is not
seriously degraded by cosmic ray damage during the 9 years in cruise. Many of the
substrates are made of Suprasil, which is known to be radiation hard, but some filters
are Schott coloured glasses to achieve a proper out-of-band blocking. Since too little
was known about the radiation hardness of such glasses, and unacceptable damage
levels could not be excluded, laboratory experiments with the Uppsala tandem Van
de Graaff accelerator were performed. A 2 MeV proton beam was shot onto OG590,
KG3, and Suprasil blanks to simulate the solar proton exposure during the Rosetta
cruise. The resulting change in spectral transmission was measured (Possnert et al.,
1999; also Naletto et al., 2003).
Figure 15 illustrates the obtained results by Possnert et al. (1999). The total
proton fluence of 1013 cm−2exceeds the expected fluence for Rosetta by almost
two orders of magnitude. Other experiments using smaller fluences or lower dose

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 467
TABLE VI
Filters of the Wide Angle Camera.
Wavelength Bandwidth Peak Trans. Thickness at
Name (nm) (nm) (%) Objective centre (mm) Wheel
Empty Empty position to allow the use of filter wheel 2 1
Empty Empty position to allow the use of filter wheel 1 2
UV245 246.2 14.1 31.8 Continuum surface spectral reflectance 4.51 1
CS 259.0 5.6 29.8 CS gas emission 4.60 1
UV295 295.9 10.9 30.4 Continuum for OH 4.75 1
OH-WAC 309.7 4.1 26.0 OH emission from the vicinity of the nucleus 4.82 1
UV325 325.8 10.7 31.6 Continuum for OH surface spectral reflectance 4.85 1
NH 335.9 4.1 23.6 NH gas emission 4.86 1
UV375 375.6 9.8 57.3 Continuum for CN surface spectral reflectance 4.60 2
CN 388.4 5.2 37.4 CN gas emission 4.61 2
Green 537.2 63.2 76.8 Dust continuum cross-correlation with NAC 4.71 1
NH2572.1 11.5 60.9 NH2gas emission 4.74 2
Na 590.7 4.7 59.0 Sodium gas emission 4.75 2
VIS610 612.6 9.8 83.4 Continuum for OI surface spectral reflectance 4.65 2
OI 631.6 4.0 52.4 O (1D) gas emission for dissociation of H2O 4.66 2
R 629.8 156.8 95.7 Broadband filter for nucleus and asteroid
detection (NAC redundancy)
4.67 2

468 H. U. KELLER ET AL.
Figure 15. Transmission curves for KG3 glass. The bold curve provides the transmission prior to
irradiation to a proton fluence of 1013 cm−2. The dashed curves show the recovery of transmission
with time at room temperature.
rates yielded much smaller effects, and the general conclusion is that the expected
damage levels are indeed acceptable. Moreover, the figure shows that annealing
at room temperature causes a rapid recovery towards the initial transmission. The
experiments on Suprasil verified that no visible damage occurred in this case.
7.2.2. Physical Parameters
The filters are placed relatively close to the detector in the optical path. This min-
imises the size of the filters but, because of the large CCDs used by OSIRIS, the
required aperture is still fairly large. The required clear aperture is 37.5 ×37.5 mm2.
The physical size of the applied filters is therefore 40.0 ×40.0 mm2with rounded
corners. They are wedged to reduce ghosts and are optimised for operation at
+10◦C.
7.2.3. NAC Ghosts
The NAC suffers from a complex combination of ghost images due to three trans-
mission elements in front of the CCD: the filters, the focusing plate and the Anti-
Radiation Plate (ARP). Two types of ghosts may be distinguished. The ‘narcissic’
ghosts are caused by light reflected from the CCD surface and back reflected from
transmissive elements. Filter ghosts are caused by two successive reflections from
transmissive elements.
The ghost images are out-of-focus replicas of the scientific image, and the
amount of defocus is different for each ghost image according to the extra optical
path travelled. For a point source, the diameter of the ghost image increases with

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 469
increasing optical path, and so the ghost intensity decreases. For extended objects
however, such as the comet nucleus, the integrated ghost intensity is independent
of defocus distance and equals the product of the two reflections encountered.
In order to take advantage of cases where one ghost type is weaker than the
other, the two types are physically separated. This is achieved by introducing a 4◦
tilt of the filter wheel, hence of filters and focusing plates, with respect to the optical
axis, sending filter ghosts to one side of the scientific beam and narcissic ghosts to
the other side. The slight dispersion effect introduced by this tilt is compensated
by the 10wedge of the filters. Also, ghost reflections from the focusing plates are
reduced by using specialised plates for the UV, visible and IR ranges.
The most problematic ghost components are produced by the ARP and the CCD.
While efforts were made to reduce their reflectance over the entire wavelength range,
ghost performance is optimised in the orange/red region, where other performance
criteria (stray light rejection, efficiency, etc.) are also optimal. In this region, ghost
intensity of less than 10−3is required.
7.2.4. WAC Narcissic Ghosts
The peculiar orientation of the WAC filters with respect to the light beam has to
be emphasised. The beam incidence is not normal. The non-wedged surface of the
filter is parallel to the CCD plane which is orthogonal to the camera optical axis.
The angle between the optical axis and the central ray of the light beam depends
on filter thickness and wheel position, e.g. 8.75◦for the green filter in filter wheel
1 and 8.9◦for the red filter in filter wheel 2. The thickest filter side is towards the
filter wheel rotation axis.
The thicknesses of the WAC filters were calculated to have the same focus shift
for all filters taking into account the focus shift introduced by the Anti-Radiation
Plate (ARP).
Ghost minimization with suitable anti-reflection (AR) coatings for the WAC is
even more stringent than for the NAC because of the initial contrast requirements.
Analysis of the ghost images has shown that the secondary narcissic ghost is the
most intense one. This ghost is produced by back-reflection of the beam from the
CCD surface and the outermost filter surface. The ratio of total narcissic ghost over
image intensity depends on the actual filter and is between 0.16 (worst case, for
NH filter) and 10−3(best case, for Green and R filters).
8. Filter Wheel Assembly
The Filter Wheel Mechanism, FWM, positions the optical filters in front of the
CCD detectors with high accuracy. The assembly is composed of
ra support structure
ra common shaft with two parallel filter wheels

470 H. U. KELLER ET AL.
Figure 16. Filter Wheel Mechanism. Two V-shaped flat springs lock the filters by the Vespel cams.
Reed switches are activated by encoder magnets to identify the filter in front of the CCD.
rtwo stepper motors with gears (crown and pinion)
rposition encoders and mechanical locking devices.
Figure 16 shows the fully assembled FWM. The mechanism provides the space
for 16 optical elements (12 filters and 4 focusing plates in the NAC, 14 filters and
an empty position per wheel in the WAC). The selected filters for both cameras are
described in Section 7.1.
Each filter wheel is turned by a stepper motor to position a filter in front of the
CCD in less than 1 s (half wheel turn). All filters are positioned with an accuracy of
±135 μm (10 CCD pixels) relative to the optical axis with ±30 μm of repeatability
(two pixels). The positioning accuracy is achieved by V-shaped Vespel cams, one
on top of each filter, which are locked by stationary V-shaped stainless steel springs
attached to the mechanism support.
8.1. FILTER ACCOMMODATION
Each wheel has eight square openings to accommodate the filters. The filters are
mounted in cover frames of aluminium alloy and further positioned by elastic joints,
which preclude damage to the filter’s surface upon thermal expansion. In order to
minimise light reflection, all mechanism surfaces (except the gear teeth, the pinion
and the motor fixation) are finished in black.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 471
8.2. WHEEL DRIVE MECHANISM
Due to the tight mass, power and timing allocations, titanium alloy was selected
for the central shaft, while aluminium alloy is used for the filter wheels and for the
assembly support. The filter wheel support has three mechanical interface points to
the camera that allow adjustment by shimming to obtain the required alignment to
the optical path.
The wheels are mounted to the central shaft by double ball bearings, coupled
back-to-back with 50 N pre-load. These space-qualified bearings are dry-lubricated
by lead ion sputtering of the stainless steel races. The wheels support the Vespel
crown gear at one side. The pinion on the motor shaft is made of stainless steel. The
gap between the pinion and the crown is adjusted to 50 μm, which is equivalent to
0.07◦backlash in the wheel. SAGEM 11PP92 type stepper motors were fabricated
with redundant windings and were further modified to provide a high holding torque
of 3 Ncm at a power of 10.5 W. Smooth operation is obtained by a ramped step rate
provided by the Mechanism Controller Board (see Section 13).
8.3. POSITIONING ACCURACY AND FILTER ENCODER
Motor movement is achieved by sequential activation of the 4 motor phases, where
two adjacent phases are always simultaneously powered. Each activation step moves
the motors by one rotation step. A change to the next filter position requires 27 motor
steps in either direction. As the motors do not have permanent magnets (variable
reluctance type), they consequently do not have a holding force when not powered.
A mechanical locking device is required to keep the filter wheels in place when a
filter change is completed.
The filter selection is monitored by a binary system where the code is given by 1–
4 SmCo encoder magnets beside each filter and a stationary set of 4 reed switches.
The field distribution of the magnets is focussed towards the reed switches thus
creating a well-defined activation area.
9. Shutter Mechanism
In each camera an electromechanical shutter in front of the CCD controls the
exposure. The shutter is designed to support exposure times between 10ms and
>100 s with a maximum repetition rate of 1 s−1. Typical imaging might use exposure
times of 100 ms and repetition rates of one image every 7s. The shutter is able to
expose the 28 ×28 mm2active area of the detector with uniformity of better than
1/500. A total of 50,000 shutter operations is anticipated throughout the mission.
The shutter comprises two blades travelling across the CCD parallel to the CCD
plane. They are each driven by four-bar mechanisms from brushless dc motors

472 H. U. KELLER ET AL.
Figure 17. Shutter mechanism flight unit. The shutter blades are at the bottom.
(Figure 17). To determine exposure with high accuracy, a customised encoder for
each blade is mounted to the motor shaft.
A position sensor at the final position verifies that the first blade has completed
its travel. A mechanical locking device locks the first blade in open position until it
is released by the second blade at the end of travel, when the exposure is completed.
The back-travel of both blades is provided by springs.
The exposure time is precisely defined by the relative distance (e.g. by the delay)
between the moving blades. The exposure time can be any multiple of 0.5 ms, 10 ms
minimum.
9.1. BLADE MOVEMENT
The blades are moved in the direction of CCD columns with a constant velocity
of 1.3 m s−1. The blades are accelerated and decelerated by a current waveform
controlling the motors in 512 steps each at 8-bit resolution. Figure 18 shows a
typical waveform for the actuation of the first blade, which is completed in 53ms.
The blade movement across the CCD lasts 21.3ms (or 96 px ms−1).
The blade velocity is measured by an optical encoder mounted on the shaft of
the motor. The encoding accuracy leads to a blade position resolution of about
0.08 mm.
9.2. PERFORMANCE VERIFICATION
Uniform exposure across the CCD is achieved by constant blade velocity passing
the detector. In order to satisfy the long-term stability requirements, a calibration

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 473
010 20 30 40 50
Time, ms
-0.5
0.0
0.5
Motor Current, A
Acceleration
Deceleration
CCD transit
Lock
Figure 18. Shutter current waveform to achieve constant velocity across the CCD.
scheme for the shutter blade movement was established. The shutter movement
is optimised by adapting the current waveform for the motors by analysis of the
encoder data. The Data Processing Unit (Section 12) evaluates the encoder data
onboard and generates an optimised waveform in order to achieve uniform exposure
of the CCD. A shutter calibration cycle lasts approx. 15 min per camera and is
executed routinely in flight.
9.3. SHUTTER ELECTRONICS
The shutter electronics controls the operation of the shutter mechanism. As shown
in Figure 19, it is split into a digital and an analogue module. The boards are
accommodated in the NAC and WAC CCD Readout Box (CRB box, see Section
11).
Figure 19. Shutter electronics board.

474 H. U. KELLER ET AL.
The digital module stores the current waveform data for both blades in FIFOs.
These FIFOs are loaded from the Data Processing Unit with the actual waveform
data. Updated waveforms can be calculated onboard or received by telecommand.
The waveforms for both blades can be different. The digital module checks con-
tinuously the status of the memory and the functionality of the mechanism. The
electronics is prepared to identify 11 different types of errors. If an error is detected,
the actual status is immediately reported to the Data Processing Unit.
The analogue module is composed of a capacitor bank with associated current
switches and the circuitry to select the charge mode for the capacitors. The capacitor
bank is needed to feed the motors with a peak power of 20 W during the acceleration
and the deceleration phases (10 ms each). Three different charge modes, e.g. fast,
nominal and slow mode, are implemented according to the desired shutter repetition
time.
9.4. FAIL-SAFE MECHANISM
The fail-safe mechanism configures the shutter into a pseudo frame-transfer CCD
mode in case an unrecoverable mechanism failure occurs. It forces the first blade to
cover one half of the CCD while the second blade is blocked in the starting position.
The open section is then used for imaging. The acquired charge is rapidly shifted
into the covered section for intermediate storage and subsequent readout.
10. Front Door Mechanism
The Front Door Mechanism, FDM, is primarily designed to protect the optical
components inside the NAC and the WAC by reclosable front doors. The inner side
of each door can be used for in-flight calibration in combination with the calibration
lamps. The mechanisms for the NAC and WAC telescopes are identical with the
exception of the shape of the doors, as these are different in order to fit the entrance
baffles of the two cameras. As the front doors cover the field of view of the cameras,
the reliability of the entire subsystem during the mission’s lifetime requires highest
attention.
10.1. REQUIREMENTS AND DESIGN
The main functional and environmental constraints of the mechanism can be iden-
tified and summarised as follows:
rthe door has to prevent contamination of the internal surfaces of the
telescope

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 475
External Cam
Internal Cam
mic
r
os
w
itch
Stepper
motor
Oute
r
cam
I
nne
r
cam
Figure 20. Components of the FDM and cylindrical development of the cams.
rsingle-point failure tolerance requires redundancy and the ability to open the
door permanently in the case if an irreversible system failure occurs (fail-safe
device)
rrequirement to validate open and closed positions
rdynamic load during launch
rnon-operational temperature range (−50 to +70◦C) implies a design for high
differential thermal loads within the mechanisms.
The door mechanism is designed to maintain the moving door always parallel to
its closed position plane, thus avoiding direct exposure of the inner surface to open
space, to the sun, or to cometary dust particles, because collected contaminants
could be re-emitted into the telescope once the door is returned to its closed position.
The parallel motion is achieved with two coupled cams that initially lift the door
followed by a roto-translation which completes the lift and rotates the door. The
shape of the two cams was designed in such a way that both final positions (open and
closed) are self-locking states, so that no electrical power is required to maintain
these positions, even if the system is exposed to vibrations.
Figure 20 shows the main components of the mechanism generating the move-
ment of the door. The internal cam is activated by a stepper motor with a step angle
of 0.3◦and a gearhead with a reduction ratio of 100:1. The combined motion is
transferred to the door by an internal shaft rigidly fixed to the coupling peg and to
the supporting arm.
The actual position of the door can be determined from the number of applied
motor steps. Nevertheless, two micro switches are employed to identify the open and
the closed positions for the housekeeping monitoring. These switches are located
on the external cam and are activated by a disk that is fixed on the peg.

476 H. U. KELLER ET AL.
Figure 21. The integrated Front Door Mechanism.
A preload of the door against the external baffle of the camera improves the
stiffness of the system composed of the sustaining arm, the door and the external
baffle. Potential damage to the baffle due to vibrations of the door, especially during
launch, is avoided by a damping seal. Figure 21 shows the completed FDM.
10.2. RELIABILITY
High reliability of the FDM for the extended lifetime of the instrument is of utmost
importance. The concept comprises not only redundant drivers and motor windings,
but also extensive safety margins in the mechanical design. The latter includes par-
ticularly the mechanical load during launch, the specific implementation of sliding
parts and, finally, decreased sensitivity to the long-term mission environment.
Differential thermal expansion was taken into account by a number of elastic
elements which absorb thermally induced loads. The FDM is covered with a thermal
blanket that efficiently isolates the structural parts of the mechanism from thermal
paths to the environment.
All moving parts must be coupled tightly together to make the arm stiff enough
to sustain the mechanical load during launch. Increased bearing friction by adhesion
or cold-welding phenomena must be avoided. Therefore, an innovative lubricant
coating has been applied, which relies on the low friction properties of MoS2,but
is not affected by sensitivity to humidity. This so-called MoST coating is a vacuum
deposition of MoS2in a matrix of titanium that preserves the lubricant properties
of the coating.
10.3. FAIL-SAFE DEVICE
A fail-safe device is required beyond the general redundancy concept to make the
front door single-point failure tolerant should an irreversible system failure ever
occur. This fail-safe device would open the door once and forever.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 477
Figure 22. Activated fail-safe device of the FDM.
The device is located within the arm holding the door and, to make it fully
independent, is operated on an axis parallel to the cam axis. It provides for a lifting
of the door and a subsequent rotation of 90◦by preloaded springs.
The arm supporting the door has been divided into two parts, which are kept
together by a locking slider. The lock can be released by a Shaped Memory Alloy
actuator. Once the lock is released and the slider is pulled away by a spring, the arm
supporting the door is lifted-up by a coaxial spring. A torsion spring finally rotates
the door and keeps the door in the open position. Figure 22 shows the released state
of the fail-safe mechanism.
A high preload of 70 N at the main spring was applied to overcome adhesion
or cold-welding phenomena, which could appear between the moving parts in the
course of a long-term mission. Friction coefficients for the moving parts were
minimised also by a sputtered MoST coating on the relevant surfaces of the arm
and by a chromium coating deposit on the slider.
11. Image Acquisition System
Both cameras use identical image acquisition systems, consisting of two sepa-
rate subsystems: (1) the Focal Plane Assembly (FPA), accommodating the CCD
detector, the Sensor Head Board (SHB) with the front end electronics, heaters,
temperature sensors and radiation shielding, and (2) the CCD Readout Box (CRB
box), with the CCD Readout Board (CRB), the Housekeeping Board (HKB), the
CRB Power Converter Module (PCM) and the Shutter Electronics (SHE). The
FPA and CRB box are about 50 cm apart and are interconnected by a cable of 62
lines.

478 H. U. KELLER ET AL.
11.1. DETECTOR SELECTION CRITERIA
The detector is a key element of the OSIRIS cameras. Its format and performance
have a major influence on the parameters of the optical system. The pixel size
determines the focal length for a defined angular resolution, and the QE relates to
the F-number. These parameters strongly influence the dimensions of the optical
systems. A constraint on the detector selection was the requirement to select a CCD
device that needed only little further development for space application saving cost,
development time and risk.
The requirement of highest possible QE over the wavelength range from 250
to 1000 nm leads to the choice of backside illuminated CCDs. A minimum pixel
capacity of 105electrons was considered as acceptable. Low readout noise in the
order of a few electrons per pixel was required to achieve sufficient dynamic range
in the image data.
Large CCDs of 2k×2kpixels have some drawbacks compared to smaller
devices. Foremost, they are more sensitive to Charge Transfer Efficiency (CTE)
degradations, which occur under high energy irradiation in space. Therefore, tight
shielding and the capability to anneal defects at elevated temperatures up to +130◦C
were implemented. The storage temperature of the detectors during cruise is kept
near room temperature. A further drawback of the large CCD format is the increased
readout time. Two readout amplifiers cut this interval in half and also provide
required redundancy.
11.2. OSIRIS CCDS
The OSIRIS CCD design is based on the commercially available, backside illumi-
nated non-MPP E2V CCD42-40 devices with 2 output channels. These CCDs fea-
ture the desired pixel size of 13.5 μm2and excellent wide-band QE. High dynamic
range and low power consumption make them well suited for space applications.
The CCD specifications are summarised in Table VII.
The non-MPP clocking register technique yields high full well capacity but
also high dark charge generation. Since the dark charge is almost negligible at
the in-flight operational temperature range of 160–180 K, the OSIRIS CCD takes
advantage primarily of the enhanced charge capacity.
An innovation for the OSIRIS devices was the introduction of lateral (shielded)
anti-blooming overflow protection, so that weak cometary features can be imaged
near bright regions in long duration exposures. The lateral anti-blooming keeps the
entire pixel area light-sensitive so that the QE is not affected. Nevertheless, the full
well charge capacity is reduced by the anti-blooming from 140,000 e−to about
100,000 e−per pixel.
Dark current becomes a significant component at temperatures above 230K.
Therefore, during the device evaluations at room temperature, full pixel-wide clock

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 479
TABLE VII
OSIRIS CCD specification.
Item Specification
Source detector type E2V CCD42-40, non-MPP, backside illuminated, Hafnium oxide
AR coated
Array size Full frame, 2k×2kpixel
Serial register size 50 +2k +50; 50 extra pixel at both ends 48 +2k+48 transmitted
Pixel size 13.5 ×13.5 μm2
No. of outputs 2; either 1 sufficient
Overexposure control Shielded anti-blooming
Operation modes Clock dithering for dark current reduction for operations at
>220 K (optional), windowing, binning
Full well >120 000 e−px−1
System gain ∼3e
−/DU
Readout noise (CCD) ∼15 e−rms
Dark charge generation <0.1 e−s−1px−1@ 180 K ∼400 e−s−1px−1@ 293 K – (with
dithering)
QE 250 nm: 50%, 400 nm: 60%, 600 nm: 88%, 800 nm: 65%,
1000 nm: 6%
Readout rate 1.3 Mpx s−1; 650 kpx s−1per channel
Readout time (full frame) 3.4 s (2 channels)
Vertical clock rate 25 μs per line
Operating temperature 160 K <T<300 K
dithering at a fixed rate of 80 μs/cycle was applied, yielding typical dark charge
reduction rates by a factor of up to 15. With the help of such dithering, useful images
could be obtained up to exposure times of 40 s. Below 230K, clock dithering is
no longer useful, because spurious charge becomes dominant (Kramm and Keller,
2000; Kramm et al., 2004).
11.3. DETECTOR PACKAGING CONCEPT
With high thermal insulation between the CCD substrate and the focal plane hous-
ing, the CCD can be operated either at low temperatures (down to 160K) or can
be heated up to 400 K to anneal radiation defects. Special detector packaging was
required, therefore, to provide sufficient thermal insulation.
The CCD substrate die is glued to a 10 mm Invar carrier plate that is attached to
the housing structure with perfect thermal insulation by two stages of three glass
spheres to mount the device (see Section 11.8). Cooling (by a thermal radiator) and
(electrical) heating is applied from the back side of the Invar plate. The Invar plate
keeps the detector flat to less than 10 μm and provides shielding against irradiation.

480 H. U. KELLER ET AL.
The electrical interface is obtained via a small ceramic interface board mounted
to the Invar plate. On its top side, the board provides 32 gold plated bond pads,
which are aligned with the substrate bond pads on the CCD. A flex circuit connects
the CCD to the SHB through gold-plated contacts on the ceramics.
11.4. READOUT CONCEPT AND IMAGE FORMATS
In stand-by mode, the CCD is continuously clocked at a moderate rate of about
3 ms/line. Prior to an exposure, the CCD is entirely cleared by a fast vertical dump
of 25 μs/line. The exposure is started by the shutter opening and completed by the
movement of second shutter blade.
Full frame or sub-frame (window) images can be read with or without binning
and via either one or both channels. If both channels are used to read a sub-frame
of the CCD, the centre of the sub-frame must be aligned with the centre axis of the
CCD. Binning formats of 2 ×2,4×4and8×8 pixels are supported.
11.5. SENSOR HEAD BOARD
The sensor head provides only limited space for electronics. As a consequence, the
front-end electronics accommodate just the preamplifiers and protection circuitry.
This concept minimises the power dissipation in the FPA section and thus protects
the CCD detector from heating up. The total power loss in the FPA section is made
up of approximately 90 mW within the CCD substrate and about 270 mW on the
SHB.
11.6. CCD READOUT BOARD
Each CRB contains two complete signal chains, including line receiver, the corre-
lated double-sampling (by clamping), the buffer amplifier and analogue-to-digital
converter (ADC). All clock signals are provided by an ACTEL 1280 Field Pro-
grammable Gate Array (FPGA). A second FPGA handles the high speed serial
(LVDS) data link to the DPU.
The implementation of the ADC section was complex because the required data
resolution of better than 14 bit could not be achieved with a single ADC in each
channel. A sub-ranging technique was applied using two simultaneously operated
14-bit ADCs in each channel, one for the range up to 50,000 e−and the second
covering the range up to 200,000 e−. The criterion for the selection of the conversion
result is based on the occurrence of an overflow at the low-range converter. We
selected the LTC1419 type ADC because it was found to be radiation resistant to
at least 50 krad (Tomasch et al., 2000).

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 481
Figure 23. The CCD readout board.
The entire readout electronics could be accommodated on one board with the
help of four different types of thick film hybrid circuits especially designed for
OSIRIS to save both space and power (Figure 23).
About 1.6 W are sufficient for stand-by operation, and 3.2 W are required during
readout.
11.7. HOUSEKEEPING BOARD
The operating conditions of the CCD, the FPA and the CRB are monitored by the
HKB. It measures the voltage and current on all six power lines, and determines the
temperatures of the CCD, the shutter actuators, the ADCs and the related PCBs.
Furthermore, the HKB reads the dosimeter-FET that was implemented to track the
camera total ionizing dose. All housekeeping data are incorporated in the image
header. The HKB requires less than 190mW.
11.8. FPA MECHANICAL DESIGN AND THERMAL CONTROL
As shown in Figure 24, the Invar plate carrying the CCD substrate die is thermally
insulated by two concentric groups of three glass spheres. A titanium ring in-
between holds the glass spheres in place and serves as an additional insulator.

482 H. U. KELLER ET AL.
Figure 24. FPA front view with the CCD in blue colour.
The thermal resistance between the Invar plate and FPA housing is in the order of
500 K/W.
A cold finger connects the Invar plate to a radiator for passive cooling. The
thermal conductivity of the cold finger is a trade-off between two contradictory
requirements. Operating the CCD requires low gradients between the radiator and
the CCD. On the other hand, annealing the CCD with limited heater power requires
a considerably higher gradient on this path. Developing a ‘thermal switch’ between
the radiator and the CCD would have been an elegant solution but was judged to
be a technological risk and thus was rejected. The compromise is a cold finger
consisting of 45 single aluminium sheets of 0.1 mm thickness. The temperature
gradient between the radiator and CCD during operational conditions is in the
order of only 1 K. The lightweight radiator is mechanically mounted to the FPA
housing by a supporting structure of 7 GFRP (glass fibre reinforced plastic) tubes
(Figure 25).
The total conductive and radiative heat input onto the cold parts from the warm
FPA housing, from the telescope in front of the CCD and from the SHB electronics
sums up to nearly 1 W. To reach the nominal operating temperature of 160 K, a
radiator surface of 430 cm2is needed. The radiator has a view factor of nearly
2πsr to deep space. Its outer surface is painted with a conductive white paint
having a low solar absorption coefficient αand a high infrared emissivity ε. The
low αhelps to reduce the heat input and temperature changes of the CCD during
asteroid flyby phases, when sun incidence on the radiator cannot be avoided.
The CCD temperature is passively controlled by the efficiency of the radiator,
but it can be adjusted electrically by 4 heaters mounted to the cold inner parts. A
0.5 W heater is used to adjust the operating temperature to levels higher than defined

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 483
Figure 25. NAC and WAC focal plane assembly.
by the passive cooling. Two 3 W heaters are powered by the spacecraft during non-
operational phases to keep the CCD at elevated temperatures and minimise radiation
damage effects. A 25 W heater is installed to decontaminate and anneal the CCD.
11.9. RADIATION PROTECTION
Solar protons dominate the radiation effects to the CCD in the inner heliosphere.
They generate lattice defects at a sensitivity threshold of a proton fluence of
108cm−2or a dose equivalent of 100rad (Holland et al., 1990). Passive shield-
ing of the CCD was implemented in both cameras to ensure high performance at
maximum solar activity after 9 years cruise.
Dense absorbers and a dedicated quartz window plus the optical filters in front
of the CCD reduce significantly the number of incident protons. However, radia-
tion shielding to 20 g cm−2or 150 MeV kinetic energy is required for appropriate
reduction of the proton fluence for the Rosetta mission.
Mass and volume restrictions do not allow full implementation of shielding.
Additional measures were taken to ensure a CTI (Charge Transfer Inefficiency) at
levels better than 2 ×10−5. Raising the CCD temperature to near room temperature
during cruise reduces significantly the damages by long-term annealing (Hopkin-
son, 1989). Additional annealing at elevated temperature up to +130◦C will remove
accumulated degeneration up to 85% (Abbey et al., 1991) but is limited due to high
mechanical risks and stress at the CCD.
Thus, the OSIRIS CCD radiation protection is designed to an accumulated proton
fluence of 109cm−2with passive shielding to a minimum of 5g cm−2or 70 MeV

484 H. U. KELLER ET AL.
equivalent. Heaters ensure long-term and high-temperature annealing during the
mission and stacking of low and high-zabsorbers along the particle track reduces
x-ray and neutron generation inside the shielding (Dale, 1993). The nominal CCD
operational temperature (160 K) will freeze out traps caused by residual lattice
defects and enlarges the emission time constant. CTI verification in flight will
allow real-time determination of defects.
Radiation effects were studied on two OSIRIS CCD samples that were exposed
to 10 and 60 MeV proton irradiation up to a fluence of 2 ×1010 cm−2. We found
that long-term room temperature annealing significantly reduces the increased dark
charge leakage, while high temperature annealing is particularly suitable to cure
degraded CTE (Kramm et al., 2003).
12. Data Processing Unit
12.1. ARCHITECTURAL DESIGN
The main driver for the design of the OSIRIS Data Processing Unit (DPU) is the
need to control camera operations as well as to acquire and process image data
from the two CCD arrays. This made a Digital Signal Processor (DSP) desirable.
The DPU is based on a development of the ESA Technical Directorate using the
radiation-hard version of the Analog Devices ADSP21020 processor.
Because of power limitations, the speed of the processor had to be substantially
reduced. A large local memory was implemented so that extensive processing tasks
could be executed by batch processing rather than in real-time. This scheme does
not impose a significant limitation to the operations, because the data transmission
rate is the most restricting requirement.
The DPU architecture is driven by the following requirements:
rhigh data rate from both cameras operated simultaneously (up to 40 Mbits−1)
routput data rate (to spacecraft mass memory, SSMM) limited to 10Mbit s−1
rsupport of ‘movie’ operation (up to 64 images with ∼1 s image repetition
time)
rsingle-point failure tolerance
rlimited resources of mass and power
rdecision to base the architecture on an existing TSC21020 processor board.
The discrepancy between the input and output data rate necessitated a DPU
internal storage of minimum one WAC and one NAC image. The data transfer
rate to ground is limited to 100 Mbit/day. Image transfer rates can be increased
by image compression at the expense of image fidelity. State-of-the-art (lossy)
wavelet compression with a compression ratio between 4 and 8 provides the best
balance between image degradation and a substantially increased image count. For

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 485
the observation of dynamic events, the optimum balance is expected to require even
higher compression factors.
Compression can be performed either in real-time (in step with the incoming
camera data) or by intermediate storage and subsequent processing of the stored raw
data. The latter was implemented because the limited bandwidth of the telemetry
restricts the time for image acquisition but leaves time for software compression.
12.2. IMPLEMENTATION
The DPU block diagram is shown in Figure 26. The DPU consists of four elements
each mounted in one mechanical frame of the Electronics Box (E-Box, see Figure
5): rMain Processing Element (PE), consisting of the processor board with the
DSP, local memory, spacecraft interface, and internal IEEE 1355 interfaces,
and a memory extension board
rRedundant Processing Element in cold redundancy
rMass Memory Board (MMB), containing 4 Gbit of image memory, control
logic, and IEEE 1355 interfaces
Spacecraft
Interfaces
E-Box
internal
Interface
NAC IFP S/C-Therm.
NAC CRB Img Data
NAC CRB Control
WAC CRB ImgData
WAC CRB Control
Red. MCB Control
WAC Struct S/C-Thm
DPU Interface Board DIB A
Main Processing Element
Mass Memory Board
Red. Processing Element
DIB B
DIB C
ED
DPU Power
Main PCM Ctrl
Red. PCM Ctrl
Main MCB Control
NAC IFP S/C Thm
Main DSP Thm
Red. DSP Thm
NAC FDM S/C-Thm
NAC Struct S/C-Th m
NAC CCD S/C-Thm
WAC CCD S/C-Thm
WAC StructS/C-
Therm.
NAC FDM S /C-Therm.
Main TC/TM
J1
Main SSMM
a
b
c
d
e
Redundant TC/TM
J2
Redundant SSMM
a
b
c
d
NAC Struct S/C-Th erm.
e
NAC CCD S/C-Therm.
WAC CCD S/C-
Therm.
Pwr
Pwr
IEEE
1355
DPRAM
Controll er
I/F
IEEE
1355
DPRAM
Controll er
I/F
IEEE
1355
IEEE
1355
FIFO
FIFO
FIFO
FIFO
LU
Detect Partition 1
Address
generator
Timing
Ctrl
Symbol ECED
Data formatter
Cmd
interpreter
.
.
.
.
Partition 0
Ctrl
port
Pwr
Power
distribution
Pwr
Power-up
logic
Pwr
S/c
I/f
RESET_A
RESET_B
MPE_ON
RPE_ON
•
Pwr ••
RPE_ON
(*)
(*): to DIB C, power-up logic
A_Clk_off
B_Clk_off
IEEE
1355
ϑ
DPRAM
8k x 32 Img
Mem
Data
Mem
Realtime
Clock
DSP
•••
••
•
Prg
Mem
BBC +
Boot
Prom
IEEE
1355
ϑ
DPRAM
8k x 32 Img
Mem
Data
Mem
Realtime
Clock
DSP
•••
••
•
•
MPE_ON
(*)
Prg
Mem
BBC +
Boot
Prom
S/c
I/f
MMBon_off
MMBon_off
•
•
Figure 26. Block diagram of the OSIRIS DPU architecture.

486 H. U. KELLER ET AL.
TABLE VIII
DPU characteristics.
Item Specification
Central microprocessor Rad-hard Temic processor based on Analog Devices ADSP21020,
20 MHz, 60 MFLOPS, data transfer from local memory to mass
memory 38 Mbit s−1
Mass memory 4 Gbit net
Program memory 8kbyte (8 bit) PROM (bootstrap kernel)
256 kword (48 bit) E2PROM (OS, task-specific program modules)
256 kword (48 bit) program SRAM All program memory latch-up
and SEU immune
Local data memory 4 Mword (32 bit) fast SRAM for local image storage
128 kbyte non-volatile E2PROM
128 kword (32 bit) fast SRAM for stack/variables storage (zero
wait state, latch-up and SEU immune)
Digital interfaces Serial command interface to NAC CRB and WAC CRB
Serial data interface to NAC CRB and WAC CRB
Serial command and data interface to MCB
Serial command and data interface to PCM
Spacecraft interface IEEE 1355 interface to SSMM
Redundancy Processing Element, spacecraft interface: dual cold redundant
MMB: graceful degradation
Protection against single
event errors
SEL: latch-up detectors (MMB only)
SEU: memory error correction, SSCDSD (MMB only)
DSP watchdog
Power management Nominal and low power mode, implemented by: (a) program
SRAM deselect, (b) clock disable of DIB A or DIB B, c) MMB
switch-off
Operating system Real-time operating system (Virtuoso)
rDPU Interface Board (DIB), with the power-up logic, the interfaces to the
NAC CRB, WAC CRB, MMB, Mechanism Controller Board (MCB), Power
Converter Module (PCM), and IEEE 1355 interfaces.
An overview of the DPU performance characteristics is given in Table VIII. The
unfolded DPU flight unit in a test configuration with a prototype of the MMB is
shown in Figure 27.
12.3. PROCESSING ELEMENT
The two PEs of the DPU are identical and cold redundant. The activation of the
selected PE is done during power-up of the DPU by the telemetry sample signal.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 487
Figure 27. DPU FM under test with a prototype of the Mass Memory Board.
Each PE is implemented on two printed circuit boards and provides:
rTSC21020F 32-bit floating point digital signal processor clocked at 20 MHz
r1.5 Mbyte of zero wait state program memory
r512 kbyte of zero wait state data memory
r16 Mbyte image memory on the extension board
r1.5 Mbyte E2PROM for non-volatile program and data storage
r128 kbyte E2PROM for non-volatile parameter storage
r32 kbyte dual port communication memory
r3 IEEE 1355 high-speed communication links
rSMCS332 communication controller
rTM/TC (telemetry/telecommand) interface controller.
The processing performance of the DPU allows wavelet image compression of a
2k×2kimage in less than 50 s at a compression ratio of c=12 and in less than 30
s of a 1024 ×1024 image at c=4 by using an optimised assembler code (Christen
et al., 2000).
12.4. MASS MEMORY BOARD
The MMB provides a user capacity of 4 Gbit, which is implemented in 4M4
DRAMs, 4-high stacks of 64 Mbit capacity each. The memory is protected by
an extended (80, 64) Reed-Solomon code against single 4-bit wide symbol errors.
The array is split into 2 partitions. A partition is composed of 8 word groups, each

488 H. U. KELLER ET AL.
of them containing 4 Mword of 80 bits, which are accommodated in twenty DRAM
chips.
The word length of 80 bits is structured into 16 data symbols and 4 parity
symbols. The code is capable of correcting errors in one symbol of 4 adjacent bits
and of detecting errors in 2 of such symbols (Fichna et al., 1998). The address
management provides logical addressing of the configured memory space in terms
of active word groups, start address and block length.
Data integrity is provided by background scrubbing for Single Event Upset
(SEU) error removal. Overcurrent sensors and circuit breakers in the supply lines
protect each partition individually from Single Event Latch-up (SEL) events.
The MMB has dual redundant communication interfaces and two independent
partitions. In case of a failure in one partition, this partition can be switched off
(graceful degradation). The processor workspace can store and process a complete
raw image in the absence of both partitions of the MMB. The MMB communicates
via two differential IEEE 1355 serial interface links, each providing simultaneous
data transfer of 38 Mbits−1in both directions. All peripheral functions of the mod-
ule are accommodated in Actel 1280 FPGAs. Each FPGA includes functions for
enhanced SEU tolerance.
12.5. REDUNDANCY CONCEPT
The DIB is split in two identical parts interfacing the cameras: DIB A for the NAC
and DIB B for the WAC. In consequence, the two camera chains (CCD detector –
CRB – DIB) are redundant. The PEs are cold redundant. The two PEs and the two
camera chains are cross-strapped using the capability of the IEEE 1355 ASICs to
accept three interfaces.
13. Mechanism Controller Board
The Mechanism Controller Board, MCB, drives the motors of the front doors and
of the two filter wheels in both cameras (see Section 8). The four-phase dual-
step, variable reluctance stepper motors were built with redundant windings, which
can be powered alternatively or in parallel. The MCB also acquires housekeeping
data from the position encoders on the front doors and filter wheels and from the
temperature sensors of the two cameras.
13.1. MCB DESCRIPTION
The MCB consists of two boards, the Control Board and the Drivers Board, both
mounted into the E-Box MCB frame (Figure 28). They are interconnected via two
flexible, low profile boards (Sferflex technology). Connections to the cameras are
provided via 62-pin connectors. Communication with the DPU is established via
RS-422 type line drivers and receivers.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 489
Figure 28. The mechanism controller board.
The Control Board hosts the digital circuitry for communication and mechanism
control and the analogue housekeeping acquisition. The digital functions are con-
centrated in two FPGAs performing the command decoding, the data collection,
the packaging and the data transmission. They also translate the DPU mechanism
commands into motor phase pulses that are transferred to the Drivers Board. Full
dual redundancy was established for the line drivers and for the digital and analogue
conversion circuitry. The main and the redundant analogue data acquisition mod-
ules can read the temperature sensors regardless of which stage is in use. Similarly,
both main and redundant modules can read the position encoders of all mechanisms.
The Drivers Board accommodates the motor drivers. As each camera unit con-
tains three motors, 48 line drivers are required to feed the main and the redundant
phases of all motors. Driving the motors directly from the +28 V spacecraft primary
power rail requires electrical isolation of the motor switches from the remaining
electronics by optocouplers.
13.2. CONTROLLER FPGA
The Controller FPGA establishes the communication, the command decoding and
execution, the housekeeping data acquisition and the reset functions.
The communication link to the DPU uses standard RS-422 interfaces, one in
each direction. Each transferred byte is 8 bits wide with 1 start bit, parity even and 1
stop bit. A communication packet consists of one command or packet identification

490 H. U. KELLER ET AL.
byte and of an appropriate number of parameter bytes (0–5 bytes on receiving, 3–
87 bytes when transmitting). Each packet is finally terminated by a checksum byte.
The command decoder interprets the received commands. Commands are related
either to the data acquisition, to the stepper controllers or to the parameters of the 6
motors, i.e. phases and step pattern. Each command is approved for coherence and
possible transmission errors in parity, frame or checksum. Motor control commands
are forwarded to the Controller FPGA. Invalid commands are rejected, and an error
flag is returned in the status word.
The MCB recognises two types of reset signals, the (internal) power-on reset
and the system reset command from the DPU. Both resets re-initiate the MCB to
the default parameter set.
13.3. STEPPER FPGA
The Stepper FPGA consists of three blocks:
rcontroller block containing two independent controllers to allow simultaneous
operation of two motors with independent parameters
rmotors block providing the generation of phases to the motors, with the option
to define the initial phase of movement
rcommunication block for the interface to the controller FPGA.
The program repertory includes parameters for the minimum (first and last step)
velocity, for the maximum velocity, for the acceleration and the deceleration ramps
and for the total number of required steps. An example for a nominal ramping
profile for a single filter change is provided in Figure 29.
010 20 30
Step number
0
40
80
120
160
Steps / s
010 20 30
0
40
80
120
160
noitareleceDnoitareleccA
V_min 0x66
V_max 0x16
Acceler. 0x08
Deceler. 0x08
Figure 29. Nominal ramping of the motor step frequency.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 491
The stepper motor controller keeps the final phase powered for a holding time
of 463 ms to obtain high position accuracy in deceleration. Another important
functionality of the MCB is the warm-up of the motors and its mechanisms by
powering individual or dual windings. The power command in heating mode must
be repeated once a second until the envisaged heating effect is achieved.
14. Power Converter Module
The Power Converter Module, PCM, provides power for the OSIRIS instrument.
High conversion efficiency is achieved with a regulated switching DC/DC converter
technique. Unused instrument subsystems can be switched-off by solid-state relays.
Distribution of noise is substantially reduced by filtering and isolation.
The initial idea was to use a central power converter unit for the entire in-
strument. During the development, the power conversion tasks were split into a
main PCM and small dedicated converters close to the CCD Readout Boards to
reduce the risk of noise pick-up. Thus, the PCM is comprised of a module in
the lower compartment of the E-Box, and the NAC CRB PCM and the WAC
CRB PCM, which are located in the respective CRB boxes. E-Box PCM-to-DPU
communication is provided via standard bi-directional RS-422 serial interface
links.
14.1. E-BOX PCM
The E-Box PCM is accommodated on two boards of 190 ×190 mm2, namely the
Power Control Board and the Power Distribution Board (Figure 30).
The Power Control Board hosts the digital circuitry that performs primarily the
following tasks:
rexecution of commands from the DPU, e.g. the distribution scheme for primary
and secondary power or requesting housekeeping data, supported by redundant
microcontrollers
rgeneration of control signals to the switches and latching relays on the Power
Distribution Board
rcollection and transmission of housekeeping data from the Power Distribution
Board
rdetermination of the primary current limitation threshold. The threshold can
be modified according to actual operational modes.
The Power Distribution Board contains the power units and analogue circuitry:
rmain and redundant DC/DC power converters to feed the DPU, the MCB, the
PCM, and other consumers such as lamps, heaters etc.

492 H. U. KELLER ET AL.
Figure 30. Bread-board model of the E-Box Power Converter Module.
rpower distribution as requested from the Power Control Board
racquisition of associated housekeeping data.
The Power Distribution Board receives the main and redundant primary power
lines through separate connectors. The entire power conversion stage, consisting of
the EMI filter, the inrush control circuitry and the DC/DC converters, is established
with full redundancy. The following power distribution stage (including the inrush
current control as well as the slope and delay control of the secondary voltages)
distributes the secondary power to the shutter electronics, to the MCB and other
consumers. Solid-state relays and optocouplers are used where necessary.
14.2. PCM SOFTWARE
The onboard PCM software is stored in an 8-kbyte-wide PROM. The PCM software
has the following autonomous functions:
rsupervision of the state of the non-active (redundant) microcontroller
rdetection of a primary current limit violation
rdetection of a housekeeping limit violation
rdetection of Single-Event Upset.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 493
Figure 31. Power converter module for the CCD readout board.
Limit overruns are reported to the DPU and, if necessary, endangered subsystems
are disconnected autonomously by the PCM software.
14.3. CRB PCMS
A CRB Power Converter Module (CRB PCM, see Figure 31) is located in each of
the two CRB boxes. It provides six supply voltages for the detector electronics, the
housekeeping board and operational heater at the CCD.
The CRB PCM provides electrical isolation from the spacecraft primary power
bus. It is based on a regulated switching DC/DC converter with high efficiency.
Minimisation of conducted EMC, emission and susceptibility is achieved by EMI
filters and snubbers at the switching stage of the converter.
Due to the sensitivity of the analogue signal chain to external noise, con-
verter switching was synchronized with the pixel readout. The CCD Readout
Board provides the synchronization signal to the converter. The phase of this
so-called ‘Sync Pos’ signal can be shifted by one of 32 possible incremen-
tal steps relative to the pixel readout period to achieve operation with lowest
noise pick-up. If, however, the synchronization signal is not available, the CRB
PCM operates in free-running mode. In nominal operation, this condition occurs
only during power-on when the circuitry for synchronization signal is not yet
settled.

494 H. U. KELLER ET AL.
15. Onboard Software
The OSIRIS flight software is composed of three sections:
rkernel software
rOSIRIS UDP library
rOSIRIS science library.
15.1. KERNEL SOFTWARE
The kernel software provides the low-level functionality of OSIRIS, e.g. the TM/TC
interface and the hardware drivers. A novel concept of onboard operational pro-
cedures was implemented for instrument control, image acquisition and process
sequencing. The OSIRIS Command Language (OCL) is a systematic approach to
generate or to adapt onboard application software for instruments with varying op-
erational profiles during mission. It continues and extends previous approaches, as
e.g. applied in SOHO/SUMER by Kayser-Threde (Birk, 1992). OCL comprises a
middleware-system for (1) application layer functions with operational sequences
in a high-level language (so-called User Defined Programs, UDPs, and Persistent
Operational Programs, POPs), (2) upload of UDPs, and (3) onboard script ex-
ecution, supported by virtual machines which interpret the precompiled scripts.
Software integrity is supported by onboard checks as well as by language-inherent
features.
The architecture of the OSIRIS OCL system is shown in Figure 32. It consists
of a space segment with the UDP manager and virtual machines, and a ground
segment with a compiler to generate UDP token code including a translator to
convert the code to a series of telecommands. The UDP manager handles the UDP
code onboard. The token code can be stored in memory or transferred as POPs to
non-volatile RAM. UDPs are invoked either by the UDP manager directly or via a
timeline.
The structure of the OSIRIS flight software is outlined in Figure 33. The UDPs
represent the upper layers (application layers). OCL is located at the mid-layer
(level 4) in the software architecture.
The major functionalities of the five software layers are:
rlevel 4: the token interpreter executes POPs and UDPs. UDPs can call func-
tions in the low-level kernel software (the run time library) or call other UDPs.
The UDPs are executed from ground using an execute-by-name scheme (with
parameters)
rlevel 3 contains library functions such as image acquisition and basic
image processing. It contains a library of image evaluation functions, like

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 495
Ground Segment
Space Segment
Compiler
UDP Manager
TC Data
Virtual Machine
NVRAM Token Memory
Timeline Lower Level
Functions
Source
Code
Local
Memory
Uplink
Ground Segment
Space Segment
CompilerCompiler
UDP ManagerUDP Manager
TC DataTC Data
Virtual MachineVirtual Machine
NVRAM Token Memory
Timeline Lower Level
Functions
Lower Level
Functions
Source
Code
Source
Code
Local
Memory
Local
Memory
Uplink
Uplink
Figure 32. Architecture of the OSIRIS command language system.
Parameter Table
User interface library
Memory
management HK Calibration Resource Access
Control
Health
Monitoring
Downlink/Processing Manager
Priority Queue Manager Logical H/W
Layer
TMGeneration
Science Library
OSIRIS UDP library
DSP H/
W
Dedicated H/
W
Virtuoso Nano Kerne
l
Micro Kernel
Driver Software
I/O Ctr
l
Intermediate Level
Level 0:
Level 1:
Level 2:
Level 3:
Level 4: UDP Manage
r
H/W Services
Kernel Software
Figure 33. Structure of the OSIRIS flight software.

496 H. U. KELLER ET AL.
histogram, sub-framing, binning, bright point determination, and the SPIHT
(Set Partitioning In Hierarchical Trees) wavelet-based image compressor sup-
porting lossy and lossless compression
rlevel 2 interfaces to the Run Time Operation System, RTOS, and the driver
software. It consists of service functions to the serial devices of NAC, WAC,
MCB and PCM and to the MMB. Telemetry, telecommand and the IEEE 1355
interfaces to the SSMM are served. All software interfaces above this level
are hardware independent
rlevels 0 and 1 interface the DPU hardware with the next higher software
level. Level 0 consists of hardware descriptions, like address, port, and data
definitions. Level 1 is shared by the RTOS Virtuoso (Eonic Systems) and
driver software. RTOS interacts with processor devices. The drivers serve
dedicated hardware. The boot loader program can load program data from the
internal E2PROM or from the spacecraft mass memory via the IEEE 1355
interface.
The kernel modules and the data flow are shown in Figure 34. Interconnections
between the modules are made in different ways, depending on data and command
flow. Data driven program parts use Virtuoso mailboxes, Virtuoso FIFOs (on a
single word base), as well as Virtuoso events (one-bit information used by driver
software) to communicate with other modules.
DriverS/W
DedicatedH/W
DSP H/W
ServiceFunctions
InstrumentContr ol
ScientificSoftwar e
ScientificSoftwar e
Level0:
Hardwaredescription
RTOS
H/KMonitoring
PCM
Services
MCB
Services
NAC
Services
WAC
Services
Telecmd.
Services
Telemetry
Services
SSMM
Services/
FileOps.
Image
Acquisition
OMMB
Services
Image
Evaluation
Image
Compression
Command
Execution
Token
Interpreter
Image
Sequencing
VirtuosoReal-timeOperatingSystem
HardwareDriver&BootLoader
Spacecraft
Sensors
Power-on
Power-save
TestModes
POPs
UDPs
toall
modules
ControlFlow
DataFlow
DataFlowforEmer gencyMode
2
4
5
4
4
3
33 2
3
2
2
2
2
2
2
1
1
1
S/W Level
DPUH/W
Services
2
toall
modules
(RTL)
H/KCollection
4
Figure 34. Interaction between software modules within OSIRIS.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 497
Maintainability of the OSIRIS DPU software is guaranteed by a typical set of
programming conventions (naming conventions, header format and commenting
rules, etc.), revision control and back-up strategies.
15.2. OSIRIS UDP LIBRARY
The OSIRIS UDP library is a collection of routines written in OCL language.
The library provides an interface between the scientific users of OSIRIS and the
hardware specific details required by the kernel software. The following modules
exist (see Figure 33):
rmanagement of the parameter table. This module handles the modification and
dump of the parameter table via the TM/TC interface and persistent storage
of the parameters in the OSIRIS non-volatile memory
rthe resource access control module protects resources to be accessed by par-
allel running UDPs
rOSIRIS image and data memory is accessed via the memory management
module
ra telemetry generation module implements the high-level protocol for message
events, generic data dumps and image data transfer
rdownlink is managed by a module providing downlink prioritization via a
number of queues
rthe hardware abstraction layer (logical H/W layer) allows the user to command
the hardware modules using logical parameters, e.g. move filter wheel to a
specified position instead of turn filter wheel a number of steps in a given
direction, and provides calibration of all HK channels
rhealth monitoring is provided by real-time monitoring of OSIRIS and will
safeguard the instrument in case of anomalies. All currents, voltages, and
temperatures are monitored as well as the radiation environment via data
from the SREM radiation monitor onboard Rosetta
ra module providing self test and performance tuning functions.
15.3. OSIRIS SCIENCE LIBRARY
The OSIRIS science library is a collection of UDPs that implement complex sci-
entific observations as single commands. Various multi-spectral sequences are im-
plemented, where full spectral cubes can be acquired. The point of the scientific
UDP library is that the library can easily be expanded to serve future needs of both
simple serial activities and highly complex activities requiring onboard intelligence
and data processing. An example of a proposed UDP is an onboard cometary out-
burst detector that periodically acquires images and only downlinks the data if
brightening of the comet is detected.

498 H. U. KELLER ET AL.
OSIRIS
DPU
Camera
Simulator
Simulator Ctrl.
Simulator Ctrl.
GSEOS V
Software
EGSE PC
Spacecraft
Simulator
Commands
Telemetry
Powe
r
Commands
Telemetry
OSIRIS
NAC/WAC Sensor Cmds.
Sensor Data
Sensor Cmds.
Sensor Dat
a
Figure 35. Commanding concept for the OSIRIS EGSE.
16. EGSE and Telemetry
16.1. EGSE AND ASSOCIATED SOFTWARE
The OSIRIS Experiment Ground Support Equipment (EGSE) consists of a stan-
dard PC equipped with subsystem simulators providing all interfaces for ground
testing as well as for flight operations (Figure 35). Hence, the EGSE supports
the entire instrument development and maintenance and also provides quick-look
presentations and health monitoring.
The operating software is based on the software package GSEOS V running on
Windows XP. GSEOS V supports the tests of the instrument under near-real-time
conditions. A data-driven concept is used instead of less efficient polling. GSEOS
V is configured for OSIRIS using the built-in G-compiler, which is based on the C
language and is enhanced in some properties to support the data-driven concept.
GSEOS V provides numerous functional modules:
rcommands can be sent to the instrument directly or via network; manual,
time-tagged or event-driven commands are accepted
rinstrument data are grouped into blocks (e.g. science or housekeeping data).
Blocks are user-defined structures on the bit level. Data processing is data
driven; if a data block is received, the decoder module calls the related, user
defined function for processing
rinstrument data can be displayed in various formats (hex, decimal, text, bitmap,
histogram, plot)
rdata can be checked and evaluated; if a limit is exceeded, a user-defined
reaction can be activated (e.g. power off, print of an error log)
rcommand, status and data protocol logging
rincoming and outgoing data can be saved on disk and can be replayed
rthe EGSE can be connected to the network on an IP level. All program func-
tions are accessible via the network. Special communication protocols (e.g.
CCSDS, SFDU) are implemented.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 499
16.2. TELEMETRY CONVERSION
The DPU acquires the image and associated housekeeping data, performs pre-
processing (compression) if required and provides the packaging for the telemetry.
The information about the processing steps is attached to the telemetry so that the
data can be reconstructed on ground.
When the telemetry data are received on ground, the images and their headers
are extracted and recovered from the data stream. The binary telemetry header is
converted to ASCII format and finally stored with the image data in individual files.
The data archiving is organised by the OSIRIS software written in IDL. A
software library retains routines for reading, writing, and processing the OSIRIS
image data. The files are stored in eXternal Data Representation (XDR) format.
PDS (Planetary Data System) formatted data are generated as well.
17. Calibration
17.1. GROUND CALIBRATION
The OSIRIS system was calibrated in several stages. Prior to final system inte-
gration, the NAC underwent a series of unit tests at LAM in Marseille to verify
focus, thermal stability, stray light performance, geometric and absolute calibra-
tion. At UPD, the WAC underwent a series of focus tests. After mating with the
flight electronics at MPS, a further series of calibration tests was performed at sys-
tem level (Figure 36): focus, flat-fielding, spectral response (including temperature
Figure 36. OSIRIS in ground calibration.

500 H. U. KELLER ET AL.
dependence), geometric distortion, as well as electrical offset and system gain of
the CCD readout chain were successfully tested. More than 300 Gbyte of data were
acquired with the OSIRIS system at MPS.
17.2. IN-FLIGHT CALIBRATION CONCEPT
The OSIRIS instrument is required to operate over a period of 11 years in a harsh
radiation environment. The geometric distortion of the cameras needs to be re-
measured in flight because it might have changed during launch. The relative align-
ment of the cameras can be affected by the change from the 1 g ground environment
to zero gravity in space. It is certain that the detectors will be affected by energetic
particles during cruise leading to changes in the detector dark current and charge
transfer efficiency. The optics might be sputtered by energetic particles, and the
primary mirrors might eventually further deteriorate because of dust particle im-
pacts. Pinholes might be created in the filters. Accurate calibration of the image
data can therefore only be obtained if additional in-flight calibration is performed.
The approach to in-flight calibration is summarised by Table IX.
Dark current images are potentially costly in terms of data volume, but must
be acquired to reveal detector radiation damage. Ground-based distortion maps are
adopted as a baseline for optical correction, but require verification by observation
of star fields. This is fairly straightforward using, e.g., fields studied by Landolt
(1992). The co-alignment of the cameras can be determined by observation of star
TABLE IX
In-flight calibration methods.
Item Method
Bias and dark current Shutter closed, opaque filter combination
Dark current dependence
on temperature
Shutter closed, varying instrument heater parameters
Geometric distortion Observation of star fields
Co-alignment Observation of specific stars
Flat-fielding Door closed, calibration lamps on, comparison with ground-based
data, or smeared exposure on extended sources
Scattered light Observation at different solar elongations, observation of extended
objects (Mars, Earth, Moon)
Linearity and shutter
performance
Multiple exposures at different exposure times
Point spread function Star observations
Relative spectral response Solar analogue stars and solar system objects
Absolute calibration Standard star observations

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 501
TABLE X
Photometric standards used for absolute response calibration.
Right Ascen- Declination VSpectral
sion (2000) (2000) Name HR HD (mag) type Remarks
02 28 09.4 +08 27 36 73 Cet 718 15318 4.28 B9III Hamuy et al.
04 50 36.8 +08 54 02 2 Ori 1544 30739 4.37 A1Vn Hamuy et al.
08 10 49.3 +74 57 58 BD +75D325 – – 9.55 O5p HST
08 43 13.6 +03 23 55 Eta Hya 3454 74280 4.3 B3V Hamuy et al.
10 48 23.5 +37 34 13 BD +38D2179 – 93521 6.99 O9Vp HST
11 36 41.1 −09 48 08 Theta Crt 4468 100889 4.69 B9.5Vn Hamuy et al.
13 09 57.1 −05 32 18 Theta Vir 4963 114330 4.38 A1IVs+Am Hamuy et al.
14 45 30.4 +00 43 03 108 Vir 5501 129956 5.7 B9.5V Hamuy et al.
18 36 56.3 +38 47 01 Vega 7001 172167 0.03 A0V HST
19 54 44.7 +00 16 26 58 Aql 7596 188350 5.61 A0III Hamuy et al.
20 47 40.4 −09 29 43 Eps Aqr 7950 198001 3.77 A1V Hamuy et al.
21 51 11.1 +28 51 52 BD +28D4211 – – 10.51 Op HST
22 41 27.5 +10 49 53 42 Peg 8634 214923 3.4 B8V Hamuy et al.
00 01 49.4 −03 01 39 29 Psc 9087 224926 5.1 B7III-IV Hamuy et al.
fields and of specific stars. Those are the same photometric standards that provide
point spread function (PSF) and spectral response calibration. A list of photometric
standards for OSIRIS is given in Table X. Relevant data can be found in Hamuy
et al. (1992, 1994), Bessell (1999), and the HST CalSpec data base (HST CalSpec,
2005).
The most complicated task is flat-fielding. To facilitate this, both cameras carry
main and redundant calibration lamps that can be turned on to illuminate the front
doors when they are closed. The disadvantage of this approach is that the illu-
mination of the telescope aperture is not uniform, neither in the NAC nor in the
WAC. Therefore, only changes in the flat field between ground testing and in-
flight can be determined. A complicating factor is that the calibration lamps have
a limited lifetime. Hence, the main lamps will be cross-calibrated against the re-
dundant lamps on an occasional basis (with the redundant lamps otherwise left off)
to compensate for deterioration over time. Attempts will also be made to check
the flat-fields by taking long exposures of the cometary nucleus during spacecraft
slews.
The first images taken with the OSIRIS cameras are shown in Figures 37
and 38. These are ‘random’ star fields (i.e. without requesting specific space-
craft pointing) that showed that the performance of OSIRIS lives up to expec-
tations. A more detailed discussion of calibration issues is the topic of a dedicated
paper.

502 H. U. KELLER ET AL.
Figure 37. First light with the NAC, 5s exposure with clear filter (FFP-VIS +FFP-IR).
Figure 38. First light with the WAC, 5 s exposure with the red broad band filter. The WAC image
contains the NAC view at the centre 1/5 FOV (horizontally mirrored).

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 503
18. Operations
OSIRIS is operated via the Rosetta spacecraft instrument timeline (ITL). The time-
line allows commands to be executed at a specified time or relative to timeline
events (for example: time of closest approach to Earth). ITL command sequences
are transferred to the instrument using Orbiter Operational Requests.
OSIRIS supports the use of high-level UDPs (see Section 15). UDP commands
can initiate a single image acquisition as well as complex observational scenar-
ios. Hence, UDPs represent a toolbox that can be used for scientific observations.
Approved UDPs have a higher reliability than dispatching a bulk of single com-
mands from the mission timeline. It has been suggested therefore to implement
substantial observation tasks, for example for the asteroid fly-bys, by single UDPs.
Commanding via UDPs definitely will be the preferred way to operate the OSIRIS
cameras.
It is envisioned that the science team will develop and implement UDPs for its
specific scientific investigations. These new UDPs can be easily merged into the
onboard software after validation on the ground reference model of OSIRIS.
19. Conclusions
There were considerable difficulties during the selection phase of the scientific
imaging system of the Rosetta mission. However, the OSIRIS design concept that
Figure 39. Colour composite of the Orion nebula M42, obtained with the OSIRIS NAC during
commissioning.

504 H. U. KELLER ET AL.
Figure 40. OSIRIS catching a glimpse of Earth and Moon from 73 million km distance.
was finally approved promised an instrument that would provide an outstanding
scientific return. Many strict requirements were placed on the system during the
design phase (e.g. operational lifetime, stray light, shutter accuracy, filter wheel
speed, CCD readout rate). Most of these were achieved with only modest reduction
of requirements in one or two areas where technical constraints demanded. The
data to be returned by OSIRIS will provide a comprehensive survey of the nucleus
of comet 67P/Churyumov-Gerasimenko and the surrounding dust and gas coma.
OSIRIS will also make a major contribution to asteroid science through its multi-
spectral capability. Its design is such that even in 2014 a superior system would be
difficult to build.
Two examples of the quality of OSIRIS images are presented in Figure 39, the
Orion nebula M42 obtained with the NAC in commissioning, and Figure 40, a
glimpse back on Earth and Moon, acquired during the Rosetta pointing campaign
from a distance of nearly 0.5 AU.
Acknowledgements
The support of the national funding agencies of Germany (DLR), France (CNES),
Italy (ASI), Sweden (SNSB), and Spain (MEC) is gratefully acknowledged. Sub-
stantial support for the development of the Data Processing Unit was provided by
the ESA Technical Directorate through the Technical Research Programme.
In addition to the formal co-authors of this paper (comprising lead scientists,
Co-Is, project managers, and lead engineers), the project was supported by an enor-
mous number of scientists, engineers, and technicians involved in the day-to-day
development of the hardware. These include J. C. Blanc, D. Pouliquen, M. Saisse
(France), A. ´
Alvarez, A. L. Arteaga, A. Carretero, M. Fern´andez, H. Guerrero, P.
Guti´errez, J. L. Lizondo, V. Luengo, J. A. Mart´ın, M. A. Mart´ın, J. Meseguer, J.
M. Mi, L. Moreno, J. Navarro, A. N´u˜nez, E. Ragel, D. Rodr´ıguez, G. Rosa, A.
S·nchez, J. C. Sanmartin, G. Tonellotto (Spain), W. Boogaerts, W. Engelhardt, K.
Eulig, B. Fiethe, A. Fischer, M. G¨artner, K. Gr¨abig, K. Kellner, A. K¨uhn, W. K¨uhn,
J. Knollenberg, W. Neumann, J. Nitsch, P. R¨uffer, H. Sch¨uddekopf, U. Sch¨uhle, I.

OSIRIS – THE SCIENTIFIC CAMERA SYSTEM ONBOARD ROSETTA 505
Sebastian, S. Stelzer, U. Strohmeyer, T. Tzscheetzsch, M. Wassermeyer (Germany),
B. Johlander (ESTEC), M. Baessato, P. F. Brunello, S. Casotto, F. Don·, M. Laz-
zarin, E. Marchetti, F. Marzari, P. G. Nicolosi, F. Peron, F. Rampazzi, B. Saggin, G.
Tondello, S. Verani, P. Zambolin†(Italy), J. Lagerros, and B. Davidsson (Sweden).
References
Abbey, A., Holland, A., Lumb, D., and McCarthy, K.: 1991, Proceedings ESA Electronic Comp.
Conference, ESA SEE-N91, p. 307.
Adams, J. B.: 1974, J. Geophys. Res. 79, 4829.
A’Hearn, M. F., Belton, M. J. S., Delamere, W. A., Kissel, J., Klaasen, K. P., McFadden, L. A., et al.:
2005, Science 310, 258.
Bessell, M. S.: 1999, Publ. Astron. Soc. Pacific 111, 1426.
Birk, M.: 1992, SUMER DPU Command Language (SCL), DPU-IF-E-0200-01-KT, Kayser-Threde
GmbH, Munich.
Brunello, P., Peron, F., Barbieri, C., and Fornasier, S.: 2000, in Fischer, R. E., Johnson, R. B., Smith
W. J., and Swatner W. H. (eds.), Current Development in Lens Design and Optical Systems
Engineering, SPIE 4093, p. 79–88.
Calvel, B., Castel, D., Standarovski, E., Rousset, G., and Bougon, M.: 1999, 3785, 56.
Campbell, D. B., Harmon, J. K., and Shapiro, I. I.: 1989, Astrophys. J. 338, 1094.
Christen, A., Wiest, R., Rastetter, P., Weih, E., Fichna, T., Br¨uggemann, M., et al.: 2000, Proceedings
DASIA 2000, ESA SP-457, p. 199–210.
Colwell, J. E.: 1997, Icarus 125(2), 406.
Dale, C.: 1993, IEEE Trans. Nucl. Sci. 40(6), 1628.
Debei, S., Fornasier, S., Ramous, P., Barbieri, C., Da Deppo, V., Brunello, P., et al.: 2001, UV/EUV
and Visible Space Instrumentation for Astronomy and Solar Physics, SPIE 4498, p. 324–334.
Divine, N., Fechtig, H., Gombosi, T. I., Hanner, M. S., Keller, H. U., Larson, S. M., et al.: 1986, Space
Sci. Rev. 43,1.
Dohlen, K., Saisse, M., Claeysen, G., Lamy, P., and Boit, J.-L.: 1996, Opt. Eng. 35, 1150.
Donn, B.: 1991, in Comets in the Post-Halley Era I, Newburn, R. L., and Rahe, J. (eds.), Kluwer
Academic Publishers, p. 335–359.
Fichna, T., G¨artner, F., Gliem, F., and Rombeck, F.: 1998, Proceedings FTCS-28, Munich, p. 408–413.
Guti´errez, P. J., Ortiz, J. L, Rodrigo, R., and L´opez-Moreno, J. J.: 2000, Astron. Astrophys. 355,
809.
Guti´errez, P. J., Jorda, L., Samarasinha, N. H., and Lamy, P.: 2005, Planet Space Sci. 53, 1135.
Hamuy, M., Walker, A. R., Suntzeff, B., Gigoux, P., Heathcote, S. R., and Phillips, M. M.: 1992,
PASP,104, 533.
Hamuy, M., Walker, A. R., Suntzeff, N. B., Gigoux, P., Heathcote, S. R., and Phillips, M. M.: 1994,
PASP,106, 566.
Hapke, B.: 1993, Theory of Reflectance and Emittance Spectroscopy, Cambridge University Press,
Cambridge, New York, Melbourne.
Holland, A., Abbey, D., Lumb, D., and McCarthy, K.: 1990, SPIE 1344, 378.
Hopkinson, G. R.: 1989, IEEE Trans. Nucl. Sci. 36(6), 1865.
HST CalSpec: 2005, Web page http://www.stsci.edu/hst/observatory/cdbs/calspec.html.
Huebner, W. F., Boice, D. C., Reitsema, H. J., Delamere, W. A., and Whipple, F. L.: 1988, Icarus 76,
78.
Keller, H. U., Kramm, R., and Thomas, N.: 1988, Nature 331, 227.
Keller, H. U.: 1989, ESA Physics and Mechanics of Cometary Materials, p. 39–45.

506 H. U. KELLER ET AL.
Keller, H. U., Knollenberg, J., and Markiewicz, W. J.: 1994, Planet. Space Sci. 42(5), 367.
Keller, H. U., Curdt, W., Kramm, J. R., and Thomas, N.: 1995, in Reinhard, R., and Battrick, B. (eds.),
Images of the nucleus of comet Halley, ESA SP-1127.
Keller, H. U., Britt, D., Buratti, B. J., and Thomas, N., 2004, in M. C. Festou, H. U. Keller, and H. A.
Waever (eds.), Comets II, University of Arizona Press, Tucson, p. 211–222.
Keller, H. U., Jorda, L., K¨uppers, M., Guti´errez, P. J., Hviid, S. F., Knollenberg, J., et al.: 2005, Science
310, 281.
Kramm, J. R., and Keller, H. U.: 2000, in Optical Detectors for Astronomy II, Amico, P., and Beletic,
J. W. (eds.), Kluwer Academic Publishers, p. 55–62.
Kramm, J. R, Sierks, H., Barthol, P., M ¨uller, R., Tomasch, G., and Germerott, D.: 2003, Benefit from
annealing proton irradiation defects on the OSIRIS CCDs, MPAE Report MPAE-W-472-03-02,
Max-Planck-Institut f¨ur Aeronomie, Katlenburg-Lindau, Germany.
Kramm, J. R., Keller, H. U., M¨uller, R., Germerott, D., and Tomasch, G.: 2004, in Scientific Detectors
for Astronomy – The Beginning of a New Era, Amico, P., Beletic, J. W., and Beletic, J. E. (eds.),
Kluwer Academic Publishers, p. 131–135.
K¨uhrt, E. and Keller, H. U.: 1994, Icarus 109, 121.
K¨uhrt, E.: 2002, Earth, Moon and Planets 90(1), 61.
K¨uppers, M., Bertini, I., Fornasier, S., Guti´errez, P. J., Hviid, S. F., Jorda, L., et al.: 2005, Nature 437,
987.
Landolt, A. U.: 1992, Astron. J. 104, 340.
McDonnell, J. A. M., Lamy, P. L., and Pankiewicz, G. S.: 1991, in Newburn, R. L., Neugebauer, M.,
and Rahe, J. H. (eds.), Comets in the Post-Halley Era, Kluwer Academic Publishers, p. 1043–1073.
Naletto, G., Da Deppo, V., Pelizzo, M. G., Ragazzoni, R., and Marchetti, E.: 2002, Appl. Opt. 41(7),
1446.
Naletto, G., Boscolo, A., Wyss, J., and Quaranta, A.: 2003, Appl. Opt. 42(19), 3970.
Possnert, G., Lagerros, J., and Rickman, H.: 1999, in Advances in Optical Interference Coatings,
SPIE 3738, p. 428–435.
Richter, K., and Keller, H. U.: 1995, Icarus 114, 355.
Rodionov, A. V., Jorda, L., Jones, G. H., Crifo, J. F., Colas, F., and Lecacheux, J.: 1998, Icarus 136,
232–267.
Sagdeez, R. Z., Smith, B., Szego, K., Larson, S., Toth, I., Merenyi, E., et al.: 1987, Astron. Astrophys.
187, 835.
Schleicher, D. G., Millis, R. L., and Birch, P. V.: 1998, Icarus 132(2), 397.
Simonelli, D. P., Thomas, P. C., Carcich, B. T., and Veverka, J.: 1993, Icarus 103, 49.
Smith, B., Szego, K., Larson, S., Merenyi, E., Toth, I., Sagdeev, R. Z., et al.: 1986, in Proceedings of
the 20th ESLAB Symposium on the Exploration of Halley’s Comet, Battrick, B., Rolfe, E. J., and
Richard, R. (eds.), ESA SP-250, p. 327–331.
Thomas, N., and Keller, H. U.: 1987, Astron. Astrophys. 187, 843.
Thomas, N., Boice, D. C., Huebner, W. F., and Keller, H. U.: 1988, Nature 332, 51.
Thomas, P. C., Veverka, J., Simonelli, D., Helfenstein, P., Carcich, B., Belton, M. J. S., et al.: 1994,
Icarus 107, 23.
Thomas, P. C., Black, G. J., and Nicholson, P. D.: 1995, Icarus 117, 128.
Thomas, N., Keller, H. U., Arijs, E., Barbieri, C., Grande, M., Lamy, P., et al.: 1998, Adv. Space Res.
21(11), 1505.
Tomasch, G., Harboe-Sørensen, R., M¨uller, R., and Tzscheetzsch, T.: 2000, in IEEE Nuclear and
Space Radiation Effects Conference, IEEE 00TH8527, p. 26–31.
Tsou, P., Brownlee, D. E., Anderson, J. D., Bhaskaran, S., Cheuvront, A. R., Clark, B. C., et al.: 2004,
J. Geophys. Res. 109, E12S01.
Vilas, F.: 1994, Icarus 111, 456.
Weissman, P. R.: 1986, Nature 320, 242.