The CFHT Adaptive Optics Bonnette :
|
Optomechanics | Total number of mirrors in science train | 5 + 1 beamsplitter (transmission) |
Total number of mirrors in WFS train | 9 + 1 beamsplitters (reflection) |
Transmission Science train (Visible) | 70% (V) excluding beamsplitter |
Transmission Science train (IR) | 75% (H), 70% (K) including dichroic |
Input/Output F ratios | 8 / 19.6 |
Overall Bonnette dimension | Diameter 120 cm, Thickness 28 cm |
Flexures | Approximately 15 microns/hour |
Optical quality | lambda / 20 rms at 0.5 microns with DM flat |
Instrument clear field of view | 90 arcsec diameter |
Wavefront sensor | |
Type | Curvature |
Number of subapertures | 19 |
Detectors | APDs (45% peak QE, approximately 20e-/s dark current) |
Field of View | 1-2 arcsec depending on optical gain |
Deformable mirror | |
Type | Curvature (+ dedicated Tip-Tilt) |
DM Number of electrodes | 19 |
DM Stroke | approximately +/- 10 microns |
TTM stroke | +/- 4 arcsec |
DM first mechanical resonance | > 2kHz |
Overall DM dimension | 80 mm |
Pupil size on DM | 42 mm |
Conjugation | Telescope pupil |
Control | |
Sampling/command frequency | Selectable (1000Hz, 500Hz, 250Hz,...) |
Max bandwidth 0dB rejection | 105 Hz |
Max bandwidth -3dB close-loop | 275 Hz |
Control scheme | Modal, 18 mirror modes controlled |
Close-loop mode gains optimization | Update every 30 seconds |
Instrumentation | |
IR imager | HAWAII array. 0.035 arcsec/pixel |
Visible imager | 2K x 2K pixels. 0.03 or 0.06 arcsec/pixel |
Visible integral field spectrograph |
The optical design (Richardson, 1994) includes an F/8 off-axis parabola that collimates the beam and image the telescope pupil on the 19 electrode curvature mirror. A F/19.6 off-axis parabola, mounted on a fast tip-tilt platform, direct the beam to the science instrument. Prior to the science focus, a beamsplitter reflects part of the light to the visible wavefront curvature sensor. Optionally, an atmospheric dispertion compensator can be inserted in the collimated beam for observation at visible wavelength.
The optical bench during its integration at the DAO. Kind of crowded in there... The geometry (19 electrodes/subapertures divided up into two rings plus a central electrode) is well suited to circular pupils; the inner ring and the central electrode allow to solve Poisson's equation over the pupil while the outer ring allows to measure the boundary conditions (Roddier, 1988; Rousset, 1994). Such a system, with few degrees of freedom but a high bandwidth, is particularly well suited to Mauna Kea seeing conditions where turbulence is weak yet fast. Figure 1 : Optical path of the instrument. The central folding mirrors are on a movable slide, so that the direct and the corrected focus are co-incident. The wavefront sensor is remotely controlled along three axis and allows to select a reference star different from the science object. Modal control and mode gain optimization (Gendron & Léna, 1994; Rigaut et al, 1994) maximize the instrument performance according to the state of turbulence and the guide star magnitude. We have modified the modal control as presented in Gendron&Léna (1994) to adapt it to close--loop operation. Using the mode coefficient power spectra over the last 30 seconds approximately and a model of the close--loop transfer functions (calibrated in laboratory), new mode gains are computed and updated on a time scale between 30 seconds and 2 minutes, allowing the system to track seeing variations. The system has been tested in laboratory at 0 and 20 degree C for flexures, optical quality, and bandwidth (Lai, 1996). Figures are reported in the table above. The quoted lambda / 20 rms at 500 nm refers to the optical quality of the science path from the input focus to the output focus, where the mirror shape has been adjusted using an interferometer as a wavefront sensor located at the science focus. When the curvature WFS is used to cancel the bench optical aberrations, the optical quality of the science path is lambda / 8. This degradation (lambda / 20 to lambda / 8) comes not only from the fact that the curvature WFS is unable to detect high spatial frequency aberrations (which the DM is not able to compensate anyway), but more that this high spatial frequency aberrations induce, through spatial aliasing an error in the estimation of the true low spatial frequency terms. To this have to be added non-common path aberations between the science and the WFS optical pathes. During the engineering runs, we used focal enlargers (both in the visible and in the near IR to adapt the CCD/IR array sampling) which degraded the image quality down to approximately lambda / 4 rms at 500nm. This is still acceptable - although marginaly - compared to the residual lambda / 2 rms at 500 nm typical of compensated images.