Canada-France-Hawaii Telescope Canada-France-Hawaii Telescope
1 Instrument Description A "Bonnette" in CFHT's jargon is a guiding head or an intermediate stage between the telescope and the instrument (camera or spectrograph). PUEO is a bonnette. It will be mounted under the telescope, and the instrument attached under it. In itself, it has no data taking capabilities, a camera or spectrograph must be attached to it. From the point of view of the user, this is a "conversion stage" which is fed with a f/8 beam (telescope) 0.8" image quality (typical) and provides as output a f/20 beam with image quality of 0.2" (median). The following picture is a photograph of PUEO.
A set of folding mirrors feed the bonnette optics which allows wavefront correction. This pair of mirrors can be slid out of the telecope beam which allows a direct f/8 configuration (backup mode). The position of the f/8 (uncorrected) and f/20 (corrected) focii is coincident (however requires a re-focussing from the user).
For the following description, see the AO bonnette diagrams provided below. The top, as seen in this diagram, is mounted onto the telescope. 
The cover (not shown), onto which a shutter is mounted, encloses the optics and add some rigidity to the very sturdy aluminum cast housing. The optical plate (bottom of AOB mechanical housing) is 3 cm thick. It accomodates the standard CFHT C3 and C4 bolt holes for mounting instruments. The next drawing shows a clearer picture of the optics in the bonnette. The first element in PUEO is the folding mirror which sends the beam onto the f/8 collimator, through the Atmospheric Dispersor Compensator (ADC), then onto the deformable mirror, on the f/20 camera mirror, through the beamsplitter and finally onto the last folding to the corrected f/20 focal plane. The beamsplitter transmits a fraction of the light to the f/20 output focal plane, and reflects the rest to the wavefront sensor. The f/8 focal plane is naturally located between the first folding flat and the f/8 collimator. A X, Y, Z stage located there allows the insertion and accurate positioning of a target, and 2 artificial stars: 8 micron and 100 micron in size (0.06" and 0.71").
The light reflected by the beamsplitter is fed into the wavefront sensor. It is focussed on the membrane mirror (f/100) where the star image can be viewed by the insertion of a diagonal mirror (the diagonal mirror is inserted in front of the membrane). Further out in the beam another diagonal mirror can be inserted to view a pupil image (useful diagnostic tool). Viewing the membrane, or image at the f/100 focus (3 arcsec), is the best way to make sure that the reference source is well centered for wavefront sensing. After 5 reflections on folding optics a pupil is imaged on the lenslet array. The images formed by the array is focussed on the optical fibers coupled to the 19 APDs.
The ADC can be inserted in and out of the beam as the user wish. It might be preferable not to use it for observations at wavelength longer than 2 um because it might increase the background to unacceptable high levels (this particular item is going to be tested during the scientific comissionning in May and June '96).
The deformable mirror is a bimorph type with 19 electrodes and has been fabricated by the company CILAS (previously known as Laserdot, France).
The black dots represent the electrode connections.
. The interferograms show the bimorph mirror without (top) and with voltages applied to flatten it (bottom). As can be seen on the previous pictures, the electrode connections are located outside the pupil area. Therefore, print-through of these, if present, won't degrade the imaging quality of the mirror (the pupil edge falls between the inner and outer aperture rings, within the gap). The optical quality when flattened (appropriate voltages applied to the electrodes: bottom panel) is excellent: /10 in double pass or
/20 in single pass.
The camera f/20 mirror is mounted on a gimball to carry out the tip-tilt correction. The full bandwidth of the tip-tilt mount/mirror combination is 900 Hz at -3 dB. The maximum throw on the sky is +/- 4.0 arcsec and the RMS jitter is 0.008 arcsec.
A set of beamsplitters will be available to the users. They have the following characteristics:
| Beamsplitter Number | Transmitted (to detector) | Reflected (to sensor) |
| 1 | ~10% | ~90% |
| 2 | ~90% | ~10% |
| 3 | ~50% | ~50% |
| 4 (dichroic) | ~100% (> 1 um) | ~100% (0.4>L>1.0 um) |
(click on a beamsplitter number for a scan of the transmission and reflectivity curves)
The sensor is a curvature sensor based on F. Roddier work's. The pupil image is analyzed by an array of micro-lens distributed in a pattern similar to the electrode geometry of the bimorph mirror. The in and out of focus modulation is done by a vibrating membrane mirror. The amplitude of vibration of this mirror determines the distance in and out of focus of the pupil image on the micro-lens array. This is an important variable in the close loop control of the AO system. This parameter must be adjusted by the observer. There is a simple correlation between the membrane amplitude needed for given seeing conditions (the setting will probably look like a slide bar with a seeing scale). The whole wavefront sensor is mounted on a series of 3 perpendicular slides allowing adjustment of focus (along optical axis) and two axis positioning of the WFS within the field. The WFS can thus acquire a reference source (guide star) anywhere across the 90 arcsec field of view.
The first generation instrument will be an imager. A simple focal enlarger has been designed and fabricated to convert the f/20 into a f/40 beam. This provides an adequate sampling of 0.02" per 15 micron pixel. FOCAM can be mounted on this enlarger. MOCAM (Mosaic of CCDs of 4Kx4K) is not recommended for the simple reason that if one uses the AO bonnette with a focal enlarger offering a finer sampling, and 19 modes correction, the emphasis is small field high resolution. A 30 mm pixel CCD provides a field of view of 43 arcsec, which is substantially larger than the isoplanetic patch (as of this writing, the emphasis is being put on the 8Kx10K mosaic, and the future of MOCAM is uncertain). Indeed the actual corrected field of view (referred to as the isoplanetic patch, or let's say the field of view where the PSF is corrected and homogeneous) will be smaller than this value, depending on the number of modes corrected (tests will be carried out during the May and June '96 commissioning runs to characterize typical sizes of the isoplanetic patches).
As for Infrared Imagery, many solutions are envisioned at this stage. The RedEye Narrow camera is being retro-fitted with a 1Kx1K chip 18 micron pixel NICMOS array. This will increase the sampling from 0.08" to 0.04" per pixel. It is envisioned as a temporary solution, since only part of the chip would be adequate for imaging, approximately 700x700 pixels (limited by the optics of RedEye). However, this solution is acceptable for commissioning, characterizing the instrument, and the first observing runs.
The second solution has been offered to CFHT by the MONICA group (Montreal University), who will adapt MONICA, and make it available to the community for AO observations. Few details are known at this stage. The NICMOS chip used will be the same as we know (256x256). MONICA's optics has been modified to allow a sampling a 0.035 arcsec/pixel. The full field of view is 9 arcsec.(If this is your wish to use MONICA, contact D. Nadeau or R. Doyon at the University of Montreal).
CFHT has received bids for the fabrication of a 1Kx1K infrared camera(s) that would offer a proper sampling for OSIS observations and AO bonnette observations. At the time of this writing (April 1996) the University of Montreal has offered the most advantageous bid and has been chosen to fabricate the camera. The Montreal group collaborates with the Observatoire de Midi-Pyrennees, and the fabrication should take 9 months (starting March 1996).
An integral field spectrograph is being designed and built at the Observatoire de Lyon for the adaptive optics bonnette. OASIS offers various spectrographic modes, and imaging modes. The commissioning for this instrument is foreseen at the beginning of 1997.
The planned modes for OASIS are:
Imaging (0.2"/pixel)
TIGER spectroscopy (spatial sampling with lenslet array)
ARGUS spectroscopy (spatial sampling with optical fibers bundle)
Long-Slit spectroscopy
Fabry-Perot Scanning Interferometry
PYTHEAS (FP Scanning + TIGER)
The constructors (Observatoire de Lyon, R. Bacon PI) have for a long time been involved in TIGER type spectroscopy, and this mode is their main driver to develop OASIS. However, this mode imposes most of the complexity of the instrument, and implementing the others represent a minor difficulty. For more details on these modes and instrument, refer to the OASIS Web Page 
or the TIGER Web Page in Lyon.
Integral Field spectroscopy is envisioned as the appropriate spectroscopic tool for adaptive optics observations. The basic argument being that before correction, the observer does not know where to point. For instance long-slit spectroscopy, with a slit of 0.4" width would require a knowledge of the position of features of interests (emission knots, galaxy nuclei etc) to an accuracy better than 0.1". This is not possible before AO observation. Integral field spectroscopy does not impose this requirement. A two dimensional field of view allows a +/- 1 arcsec uncertainty on positioning on the area of interest.
The above figures show a schematic of the mechanics and optics. On the schematic, the telescope would be on the left, and the CCD is at the right. The second figure is a simulated 3-D rendering of the instrument.
The control loop bandwidth is 90 Hz at 0 dB. The wavefront is decomposed into control modes on which gains can be applied. An automatic algorithm analyzes the wavefront at regular interval (~5 min) and determines the modes which are the most efficient for the correction. The gains are adjusted to decrease the bandwidth on the modes which contribute less to the correction. These will also constitute the quasi-static aberration correction (for example the slowly varying optical aberration of the combination telescope instrument). The mode gain optimization has thoroughly been tested in last March and proved highly efficient in choosing the most appropriate gain setting. The wavefront rms deviations decreases by 35% when the automatic gains are selected, compared to an observation taken with all gains equal to 1.
The performance of the AO system is better when the input is better. The Primary Mirror (PM) of the CFH telescope has been retro-fitted with motorized defining pads. These pads allow the optical axis of the PM to be aligned according to the telescope pointing to cancel out any flexure induced COMA. In the near future we plan to have the alignment updated each time the telescope is pointed to a new object, or by a manual command activated by the user. This function is not automated so that it does not happen during an exposure. For the time being we know that flexure induced coma is far less at the Cassegrain focus than at prime focus.
Return to the Table of Contents
