Introduction to MIOSOTYS: a multiple-object, high-speed
I Chun Shih
, Alain Doressoundiram
, Yannick Boissel
, Fran¸coise Roques
, Frederic Dauny
, Andree Fernandez
, Jean Guerin
, Hsiang Kuang Chang
, Chih-Yuan Liu
Institute o f Astronomy, National Tsing Hua University, Hsinchu, Taiwan;
LESIA, Observatoire de Paris, Meudon, France
MIOSOTYS is a multiple-object, high-speed photometer. It is currently operating on the 1.93m telescope at
Observatoire de Haute-Provence (OHP), France. The instrument consists of a multi-ﬁbre positioner which can
access maximum 29 targets simultaneously, and an EMCCD camera which is capable of recording low-level light
at high frame rate. This pape r will describes the instrument’s speciﬁcations as well as the performance, i.e.,
signal-to-noise ratio, under the current conﬁguration (ProEM CCD + 1.9 3m telescope).
Keywords: Instrumentation, MIOSOTYS, Time resolution, Imaging, Photometry
MIOSOTYS is ba sed on MEFOS (Meudon ESO Fibre Optical System) which is a multiple-ﬁbre positioner and
was ﬁrst designed for multi-objects spectro scopy, and was mounted on the 3.6-m ESO telescope at La Silla, Chile
in the 90’. It rema ins in excellent shape and, recently, has been re-commisioned by LESIA, Paris Observatory
to conduct high time resolution photometry. The instrument now consists of a multi-objects ﬁbre system and
a high speed EMCCD camera. It has be implemented at the cassegrain focus of the 193 cm telescope at the
Observatoire de Haute-Provence (OHP), France
. The ﬁbre positioner moves 29 arms to the targets within a
ﬁeld of view of 25? arc-minute. Each arm is equipped with an individual viewing system for accura te setting
and carries one individual ﬁbre that inter c ept 12?” arc-sec on the sky. All the 29 ﬁbre images are projected
onto a EMCCD camer a for fast photometry acquisition. Thus, the instrument will provide the observational
abilities for various astronomical researches, such as sur veying the small objects in Kuiper Belt, fast variability
in compact binaries, young stellar objects in star formation, etc.
Three technical observations have been carried in November 2 008, March and April 2009 at OHP. We have
test the newly designed mechanical interface, calibrated the instrument in this conﬁguration with the existing
guiding system and bonette, and investigated the observation ability for future science operations. The science
observations begin in February 2010 with a dedicated high speed E MCCD camera which replac es the moder-
ate one used for technical evaluation. In this paper, we discuss the instrument’s properties and performance
measured in laboratory as well as in actual observation. This is to provide potential observers with a practical
guideline to plan their observation strategies.
The instrument consists of three parts: 30 ﬁbre positioning arms ﬁxed on a platform, an Acquisition a nd Guiding
Image System (AGIS) above the arm platform, and a CCD camera.
2.1 Arms and ﬁbres
There are 30 positioning arms arranged in a c ircle of 200? mm diameter of at the edge of the ﬁeld. One of the
arm (#1) is used solely for guiding sy stem, and r e st of 29 arms are for observing targets. Each arm s weeps a
triangular zone by 2 motions: translation by 130mm and rotation by ± 7 deg ree. The image ﬁbre is ﬁxed on
the arm tip which is electrically insulated. The image ﬁbre is a glass ﬁbre bundle of 900mm lo ng, the diameter
of the elementary ﬁbre is 18µm. Its tra nsmission eﬃciency is about 40-45%. The viewing surface is 1.9×1.9mm.
The output end of the 29 image ﬁbres are projected on to a CCD camera through the optics of AGIS with a
2.2 EMCCD camera
We chose the ProEM
camera manufactured by Princeton Instr uments
as the imaging sensor for the instru-
ment. The sensor of e2v CCD201B is a back-illuminated, frame-transfer EMCCD with 1024×1024 image pixels
(or active area 13.3×1 3.3 mm). The large ima ge area c an cover all of 29 ﬁbre images. The peak Q.E. at 530 nm
is 95%. The air cooling system maintains the operating temperature at -55
C or lower. At the temperature the
typical dark current is . 0 .008 electron/pixel/second. The re adout noise then depends on the readout modes:
electron-multiplying (EM) or low-noise (LN).
The EMCCD has dual readout ampliﬁers (or ports, see ﬁgure 1 ), one is a traditional series register for LN
mode. The other, for EM mode, is an extended multiplication register which provide 1 to 1000 times multipli-
cation which can be controlled in linear, absolute step. The dual ports design means that the camera can be
optimised to perform diﬀerent type of observations. For example, EM mode is suitable for low-light, high speed
conditions, and LN mode is for more conventional observation (i.e., long exposure). The readout noise in EM
mode is signiﬁcant greater (50 electro n rms at readout rate 10 MHz), but is eﬀectively reduced to . 1 electron
rms when multiplicatio n gain is suﬃciently applied (see section 3.2).
3. INSTRUMENT CHARACT ERISTICS
The CCD camera’s ability largely decides the performance of MIOSOTYS. This s e c tion describes some impor-
tant characteristics of the ProEM CC D camera, such as timing properties, and signa l-to-noise ratio. Other
components of the instrument and the telescope (O.H.P. 1.93 m) also play roles here.
3.1 Eﬀective exposure time
Eﬀective exposure time is restricted by CCD’s transferring and readout processes from imaging sens or to ADC
register output. By default, the ProEM camera operates at the “Frame Transfer” mode. Initially, the exposed
sensor receives incoming photons within a pre-programmed time (T
). Once the exp osure is ﬁnished, all the
electrons are shifted to an identica l, but masked sens or (see ﬁgure 1). It takes certain amount of time (readout
) for electrons in the masked sensor to be transferred through a readout register. During the transferring,
the emptied exposed sensor can immediately receive new photons. Thus it is very useful in applications which
require continuous imaging (100% duty cycle).
, however, one should be aware of that the exposed sensor has to wait for the electrons in the
masked se nsor to completely be read out. While in waiting, the exposed sensor still re mains open to the source.
Consequently, the actual e xposure time (T
) is eﬀectively equal to T
. For example, the T
of a full 1024×1024
pixels image at readout rate of 10 MHz is ≈100 ms, thus one c an have time resolution which is at least equal to
or longer than T
Figure 1. Comparison of traditional CCD and ProEM EMCCD array structures. Credit: Princeton Instruments
Fortunately, ProEM camera provides observers several methods to reduce T
. O ne is to deﬁne sensor’s region
of interest (R.O.I.). The pixels outside the se le c ted regions will be skipped during the readout process, thus
. In MIOSOTYS’ general operating mode, its 29 ﬁbres are arranged into a 6 by 6 s quare matrix,
and the size of each ﬁbre image projected on the CCD is ∼80 × 80 pixels (see ﬁgure 2 ). As a result, only an
area of 480 × 480 pixels are read out instead of 102 4 × 1024 pixels. Consequently, T
at the same readout rate
is reduced to . 60 ms. Second is to bin the pixels, i.e. 2×2, during the readout process, thus T
can further be
reduced. To achieve ﬁner time resolution, one can combine fewer R.O.I. and/or binning conﬁgurations, as long
as the data quality is acceptable.
3.2 Signal-to-noise ratio
The signal-to -noise ratio (S/N) of a CCD is given by the well-known ”CCD equation” which ha s the form
is the total (sky subtracted) number of photons from the source; n
is the number of pixels contained
within the software aperture; N
is the number of sky photons per pixel; N
is the dark current in elec trons
per pixel per second; and N
is the readout noise in electrons per pixel. In case the background and instrument
noises are low enough compared with N
, the S/N will approximately equal to
However, in the situatio n o f high time re solution photometry, the level of incoming photons from source may
be comparable to that of the instrument noises, so that the S/N deteriorates. The dark current, with suﬃcient
cooling and very short exposure time, is low enough to be ignored, therefore the readout noise becomes the
dominant noise factor term. Ever worse, the readout noise increa ses dramatically when readout rate goes faster.
Figure 2. (Left) Image of all 29 ﬁbres p rojected on the CCD sensor (1,024×1,024 pixels). (Right) Output of R.O.I., in
which the active size reduces to 480×480 pixels. The vertical lines are artiﬁcially added by data acquisition software, and
can be removed after bias correction.
3.2.1 Electron-multiplying technique
To amplify the signal (photoelectrons) from instrument background, the camera uses electron-multiplying ga in
technology. T he multiplication takes place in the ex tended multiplication register through a process called im-
pact ionisation. The process is to amplify the electrons before they reach the output ampliﬁer and subsequent
electronics. This will eﬀectively boost the signal above the readout noise of the system. The process can take as
many as stages which corresponds to one pixel in the register. The probability of multiplication per stage is p,
and the total eﬀective gain, G, is related to the number of stages, N. This gain factor is then given by
G = (1 + p)
The probability of multiplication in each stage is actually small, in the range of 1% to 1.5%; however, by passing
through large number of stage s, the total multiplication gain can be quite high. Reading image out through the
multiplication register introduces an additional noise term called excess noise factor (F ), thereby the S/N of an
EMCCD is given by
) + n
) + (N
) + (N
where F is ∼1.4.
Applying a ppropriate amount of EM gain (G) eﬀectively reduces the readout noise to below ∼1 electron.
However, the excess noise factor due to multiplicatio n proce ss also consequently reduces the S/N by the factor.
As a result, the advantage of EMCCD becomes appa rently only whenever it is operated in high-speed, low-light
situations. Because the number of incoming photons per frame is low, the penalty of F is not so signiﬁcant (see
3.3 Transmission eﬃciency
Fibre images are projected onto the CCD sensor through a lens component (within AGIS), and it is realised
that the transmission eﬃciency varies depending on the light passing thro ugh diﬀerent part of the lens: images
at centre have better transmissio n than that at the edge. To quantify the transmissio n eﬃciency, we measured
all 29 ﬁbr e s individually using the same, stable light source. Table 1 lists the rela tive transmission eﬃciency
Figure 3. Simulation of CCD S/N relating to conventional and EM gain modes. Conventional (black solid line): traditional
readout register. I n ProEM camera, its fastest readout rate (5 MHz) is lower than that in EM gain register (10 MHz),
so that the S/N is higher. EM: 1 stage (red dashed line): the readout goes through EM register, but no gain is applied,
thereby readout noise is not suppressed, and its S/N is even worse. EM: 500 stages (red solid line): 500 stage gain is
applied in the register. The S/N begins to improve as source becomes fainter. The magnitude scale is derived from the
observation took place in March 2009.
compared to the ﬁbre 28, and ﬁgure 4 shows the geometric distribution of the transmission eﬃciency. In cases if it
is possible observers are suggested to assign central ﬁbres to fainter targets to achieve reasonable signal-to -noise
Table 1. Relative transmission eﬃciency by optical system.
Fibre eﬀ. Fibre eﬀ. Fibre eﬀ. Fibre eﬀ. Fibre eﬀ.
28 1.00 22 0.89 24 0.79 1 8 0.63 25 0.53
30 0.98 1 5 0.86 10 0.76 11 0.62 3 0.48
16 0.96 2 9 0.83 12 0.76 21 0.61 13 0.45
8 0.94 7 0.81 27 0.70 5 0.58 20 0.45
9 0.92 17 0.81 6 0.67 4 0.55 26 0.40
23 0.91 2 0.80 14 0.66 19 0.53
3.4 O.H.P. 1.93m telescope
The telescope is the host for MIOSOTYS mission. Beca use the instrument typically operates at time scale of .
1s, it is clear that the signals from targets with magnitude & 12 will be overwhelmed by the readout noise if EM
gain is not applied. Based on recent observation, ﬁgure 5 shows the improvement on S/N using EM gain of ∼500
at time resolution of .0.1s. O ne can obtain acceptable S/N of ∼ several hundreds from targ e ts with magnitudes
Figure 4. The geometric distribution of transmission eﬃciency.
between ∼12.0 and ∼14.5. Furthermo re, it is worth to notice that increa sing EM gain would not improve S/N
linearly, the experience suggests that ∼ 500 is enough for our mission.
As discussed earlier, EM gain is only useful when the rec eived photon is lower than certain level. One has
to justify the use of E M gain by considering important parameters, such as time resolution, and target brightness.
We introduced MIOSOTYS, a multi-object, fast photometry instrument based on EMCCD camera. The instru-
ment is mounted on the 1.93m telescope at O.H.P. and maximum 29 ta rgets can be observed simultaneously
at time resolution of &50 ms . The electron-multiplying technology on the camera allows observers to access
fainter targets with higher time resolution. It opens a new door for temporal-related astronomical researches,
such as searching small Kuiper Belt objects, fast oscillation in compact objects and young stellar objects. We
also welcome and invite other observers to use this instrument.
This work is supported by the National Science Council (NSC), Taiwan with grants “NSC 96-2628-M-007-012-
MY3”, “NSC 98-2923-M-00 7-002-MY3”, and “NSC 98-2811-M-007-001”, as well as by the Agence Nationale
pour la Recherche (ANR), France with grant “BLAN08- 2
328288” under the project of “Beyond Neptune”.
Figure 5. Comparison of S/N to diﬀerent conﬁgurations of EM gain and exposure time. The results were derived from the
same group of stars and the same ﬁbres were used. The magnitudes are quoted from the NOMAD catalogue. Observation
date: (circle) 01 March 2010; (square) 28 February 2010
1. L. Mortara and A. Fowler, “Evaluations of charge-coupled device / ccd / performance for astronomical use,”
SOLID STATE IMAGERS FOR ASTRONOMY: SPIE#290 1981 P. 28 290, p. 28, Jan 1981.
2. D. Tody, “The iraf data reduction and analysis s ystem,” IN: Instrumentation in astronomy VI; Proceedings
of the Meeting 627, p. 733, Jan 19 86.