ADONIS Adaptive Optics System
Numerical Simulation Performances
PSF calibrator Star - Deconvolution methods
Note: The ADONIS adaptive optics system is controlled by the telescope operator.
A recent paper presented in the San Diego SPIE meeting, summarizes
the Adonis performances. Get the gzip file (284Kb).
The Strehl Ratio, the Full Width at Half Maximum (FWHM) and the Encircled
Energy are the performance parameters we consider.
A Strehl Ratio of 1 corresponds to the full diffraction limit performance
of the telescope at the wavelength of observation. Any adaptive optics
system achieves partial correction only, with demonstrated Strehls up to
Note that the diffraction limit spatial resolution is achieved very
closely already at Strehls of 0.15 - 0.2, which allows for post-processing
deconvolution techniques to restore the full diffraction limit in the images.
Higher Strehls do not add to the spatial resolution, but add to the photon
concentration and SNR of the observed features.
The plot of experimental Strehl Ratios vs FWHM in diffraction limit
units is shown below.
The Point Spread Function [PSF] is variable in time if the atmospheric
seeing varies with time.
We have observed on average slow [tens of minutes] drifts of the average
seeing conditions, however on short time scales [seconds] we have rapid
variations of the seeing and of the AO PSF. Those who need short exposures
may seriousely consider post-facto frame selection before starting the
data reduction and analysis.
Due to the strong dependence of an adaptive optics system on atmospheric
conditions (seeing, averaged wind speed of the turbulent layers), the following
characteristics are given for average atmospheric conditions (seeing=0.8
arcsec, averaged wind speed=10 m/s). We distinuguish the case when the
WFS reference star is on-axis and the case when it is off-axis:
1. The object observed is used for wavefront sensing, i.e.
on axis reference star.
The Strehl vs FWHM recorded data represent the performance of the system.
We have measured the performances which can be expected observing at
2.12 microns, with the Reticon Wavefront Sensor. These have been
published in the ESO technical report (330 Kb, Gzip Postscript). We also show from the same report some
Strehl Ratio vs the magnitude of the reference star (NGS), Figure
Strehl Ratio vs FWHM in arcsec, Figure
50% Encircled Energy diameter vs reference star magnitude, Figure
Strehl Ratio variation vs time, Figure 6
If the spectral type of the object observed is higher than M0, one magnitude
should be added at fig.2 x-scale.
Using the EBCCD, due to its limited pixel dynamic range, Strehl
Ratio higher than 35% have not been achieved. In particular, it has to be
noticed that for a reference star brighter than a magnitude of about 12
(for a typical [B-V]=0.6),
the same AO correction quality will be reached, whatever the magnitude is
(at a frequency of 200 Hz), the final limitation being the seeing and airmass.
For an object fainter than magnitude 12, the AO correction quality will
decrease, down to a few percent for the faintest objects.
The number of photons in the wavefront sensor is related to the color-index
of the reference star. Redder stars provide more photons at a given magnitude,
as shown for the ebccd in Figure 7
2. The reference star has an angular offset from the science
The larger the separation between program source and reference star
is, the less efficient the compensation will be.
Figure 3 is a plot of theoretical Strehl ratio behavior versus angular
separation. This Strehl value has to be multiplied with the Strehl achievable
by the system, with the star on axis. It has been established with a theoretical
Cn2 profile and checked for some separation angles. The maximum
field angle between the object and the reference star we recommend is 30
Numerical simulation performances: Reticon
and EBCCD A numerical simulation of the Adonis Adaptive Optics
system has been done to provide the observers with simulated PSFs
and expected performances. The performances have been checked
with our average results during Adonis runs. The results allow users
to forecast or simulate their observations.
The results have been published and the reference will soon be made available here (18\08\00).
The D/ro value (ratio of the telescope diameter to the Fried parameter)
taken is 18. This choice corresponds to a seeing of 0.62" at 0.6 µm
which is the weigthed WFS wavelength.
We report results for Adonis using the Reticon (200 Hz) and the EBCCD
(100 Hz) wavefront sensors. The PSF images as well as the encircled energy
are given, for each natural guide star (NGS) V-magnitude. It has been assumed
a K0 spectral type NGS, the magnitude may be rescaled via the measured
color dependance of the WFS response. For the ebccd this has been measured
experimentally and reported.
The PSF represent an IR long exposure of 10 seconds.
One of the most important limitations of an adaptive optics system is the
lack of sufficiently bright reference stars close to the observed object.
It is recommended to use the object itself whenever possible to sense the
wavefront in the visible.
The degree of correction depends on the visible magnitude on the reference
star, on the angular distance from the object and on the seeing conditions.
The angular distance from reference to science target lowers the PSF quality
due to field anisoplanatism. This effect is shown in Fig. 3, however
one should bear in mind that it can be weaker or stronger, depending on
the altitude of the turbulent layers. We have observed 30% changes with
respect to the theoretical curves shown in Fig.3
Recommended max. zenith angle: 50° (above this zenith angle the correction
efficiency decreases rapidly); Max. zenith angle: 60°.
Maximum reference object diameter for wavefront sensing: 2 to 3 arcsecs
depending on the seeing conditions.
PSF calibrator star - Deconvolution methods
If desired, a point spread function (PSF) measured with an unresolved calibrator
star specified by the astronomer might be recorded in order to carry out
an a posteriori deconvolution. We strongly recommend this. We also recommend
to use the data reduction routines we distribute,
at least if the user has not experience with AO data. The purpose of observing
a calibrator star is to obtain a good estimated PSF in the IR for adaptive
optics correction identical to astrophysical program source. It
will not be however exactly identical to the PSF used in your science acquisition,
since the seeing changes continuosely. Therefore, linear or direct deconvolution
methods should be used with extreme care to avoid artifacts, especially
if faint structures are searched for.
The PSF calibrator star should be measured as closely in time as possible
(before and after each exposure) to the program source observation in order
to limit the effects of seeing evolution. This condition can be achieved
if the flux (number of photon per m2) at the input of the telescope
in the bandpass of the wavefront sensor ([450-850]nm) is the same for the
reference star and for the calibrator star (with a tolerance of 20%). This
may lead to the following guidelines to choose the
Additional conditions should be fulfilled:
For reference star fainter than mv=10 the optical magnitude
of the calibrator star should be equal to that of the reference star to
ensure identical correction efficiency.
Have the same spectrum type than the reference star used for wavefront
A set of neutral density filters are available in front of each wavefront
sensor to attenuate the flux of the reference star. Those density filters
can help to adjust the flux of the PSF calibrator star equal to that of
the reference star (+/- 1 magnitude).
The equivalent integration time for the PSF should be long enough to integrate
the correction variation (t > 3 mn).
The calibrator star should be bright enough in the IR to obtain a good
quality PSF in a relatively short time. In case the observed object is
very faint, requiring for instance exposures longer than 30 min, an image
of the calibrator should be recorded often enough, say every 10 min. The
PSF chosen for deconvolution will then be the more representative of the
seeing conditions during the exposition.
Be located within 10 deg. maximum around the object observed.
If it is not possible to find a psf calibrator close to the object
with the right flux in the IR and visual, at least both object and psf
must be suitable for the same wavefront sensor camera. If the astronomer
chooses the psf according to the visual magnitude only (because there is
no IR magnitude available) he/she should come here with 2 or 3 psf candidates,
to be sure than one of them will be okay in the IR.
Hubin N. et al., New adaptive optics prototype system for the ESO 3.6m
telescope: Come-On-Plus, SPIE 1780-87 (ESO pre-print no 48)
Rousset G. et al., First diffraction limited astronomical images with
adaptive optics, Astron. and Astrophys., 230, L29-L32, (1990)
Rigaut F. et al., Adaptive optics on the 3.6m telescope: results and
performance, Astron. and Astrophys., 250, 280- 290, (1991)
Hofmann R. et al., SHARP and FAST: NIR Speckle and spectroscopy at MPE,
ESO conference on Progress in Telescope and Instrumentation Technologies