The AO Bonnette project has entered the final stages of fabrication with the successful acceptance of several subsystems at Laserdot in July. Extensive tests at DAO of optical plate hardware in June proved to be extremely useful in identifying specific problem areas. Solutions to these problems together with the recent delivery to DAO of several critical optical components has opened the door to acceptance tests for the principal optics, the mechanical hardware and the opto- mechanical bench control system at DAO in the coming months.
The Tip/Tilt mirror mount assembly and the two Bimorph Adaptive Mirrors successfully passed extensive subsystem acceptance tests at LASERDOT in July. The tests themselves were conducted over a 3 month period by staff from Laserdot, OPM, and finally CFHT.
The Tip/Tilt Mirror Mount assembly, although formally contracted to Laserdot, is the result of design and fabrication efforts of the OPM through a commercial development contract between Laserdot and OPM. The Tip/Tilt mirror mount is controlled by a dedicated local servo controller. The system exceeded the requirements of all technical specifications at both 0 deg. C and 25 deg. C. A short summary of performance characteristics is given below :
Small amplitude bandwidth (-3db) 947 Hz - x axis 902 Hz - y axis Mirror stroke: at the mirror +/- 200 arcseconds on the sky +/- 4.0 arcseconds RMS jitter at the mirror 0.4 arcsec on the sky 0.008 arcsec.
The mounting of the tip tilt mirror itself turned up a delicate but solvable problem. During the acceptance tests a flat mirror was used in place of the real off-axis paraboloid. The figure of this optical 'flat' was monitored during installation. Because of the stiff mounting constraints needed to ensure the high mechanical bandwidth of the system it is relatively easy to distort this surface. However with care both stiff mechanical support and good optical quality were demonstrated.
The pair of bimorph mirrors were tested with equally satisfying results. Before these mirrors were subjected to optical quality tests, both were subjected to fairly severe environmental tests to ensure that they were both mechanically and electrically robust. To start with each mirror was driven for 250 hours at 50 Hz with alternate electrodes simultaneously driven with opposing polarity between -400 V and + 400 V - i.e. with inter- electrode voltages of 800 V. These tests were carried out atroom temperature, at 0ø C, and at a reduced atmospheric pressure typical of Mauna Kea. Both mirrors passed with flying colors. Under these severe test conditions the effects of hysteresis in the piezo wafers which make up the mirror substrate become clearly evident.
Figure 2: Interferograms of the Bimorph mirror. The top picture shows the mirror at room temperature, without voltages applied to the electrodes. In the bottom interferogram, the mirror has been flattened by applyinf a set of appropriate voltages to the electrodes.
Optical tests of the mirrors at zero input voltage showed the optical surfaces of both to be dominated by several fringes of residual focus error and astigmatism, as might be expected for substrates 82 mm in diameter and 2.5 mm thick. The influence functions for each of the 19 electrodes were measured interferometrically, and then used to create a control matrix to drive the mirrors flat. Interferograms of Mirror-2 in the zero voltage and best 'flat' condition are shown in Figure 2. When driven flat, the mrs wavefront error measured on the area to be illuminated when in use on the AOB was 0.054 waves at 632.8 nm - well within specification. Note also the lack of print- through of electrode connections since these - with the exception of the central electrode - are located outside the exposed mirror area. The influence of the central electrode does not extend into the illuminated mirror surface.
The optical surface of the mirrors, by the way, are amazingly robust. The replicated silver coating is SiO overcoated and withstood - on coating witness samples - 25 strokes of a pencil eraser with no apparent damage.
At DAO real progress has been made on the control of x,y,z stages, the integration of both science field and WFS optical trains, and the lenslet array which feeds light to the wavefront sensor's 19 avalanche photo-diodes.
The lenslet array story has turned into a true saga. The unit was returned to the manufacturer fully 3 times before final delivery - only to have lens elements start to separate within hours of receipt. Apparently the optical adhesive was old and did not cure correctly. Bold action by DAO staff enabled the glue line in question to be fully separated, cleaned and re- assembled ( after a glass block thickness was reduce 300 micron - and briefly lost in the mail !) to finally (we hope!) complete this chapter of the project. The lenslet block will be integrated into the assembled wavefront sensor in the coming weeks.
With the receipt of the f/20 tip/tilt off-axis paraboloid at DAO earlier in the summer, DAO was able to completely assemble the "science-path" optics. This includes the servo mount for the tip/tilt mirror shipped from France after its acceptance tests. Under visual examination and from data taken with a phase- shifting interferometer this optical train is fully diffraction limited. Similarly, as of this writing, all the optics of the Wavefront Sensor - save for the lenslet array - have been mounted and initially aligned. Current indications are that these optics are similarly of high quality. One problem with the alignment mechanism for a tiny toroidal secondary mirror located immediately in front of the lenslet array resulted in a redesign of its mounting hardware, and several of the mounts for the remaining optics have been reworked as well.
At this point all the optics except for the coatings on the 50:50 beamsplitter have been received and meet specifications. Final assembly of the wavefront sensor is underway. Once connected to the fibers of the avalanche photodiodes, tests of the wavefront measuring capabilities of this assembly will start almost immediately.
Most problems brought to light before and during the operational tests of the mechanical systems in June have been solved. In particular, a few software 'bugs' and a design error in a commercial motor driver card have been identified and repaired. The result is that lost step and homing problems previously experienced on several linear slide assemblies have been solved. One remaining potential problem - that of differential flexure or lost motion between the wavefront sensor and the science focal plane - remains something of an open question. Considerable effort at DAO has been devoted to checking and improving this stability over the past year. Further stability tests of these units as a function of sky position are scheduled to take place once the wavefront sensor has been assembled and tested. If important problems turn up provisions have been made for integrating a counter-balancing system on the wavefront sensor x,y,z, linear slides.
At OPM, after completion of the tip/tilt mirror mount, progress continues on the fabrication of various test fixtures to be used during the final integration phases of the project. The instrument tip/tilt test stand is nearing completion. Initial tests of a seeing generator indicate that pseudo-Kolmogorov turbulence can be generated, although at the moment the equivalent values of 'r0' are somewhat lower than desired. Current plans call for delivery of the opto-mechanical assemblies to OPM in March, 1995 and delivery to CFHT in the fall, roughly one year from now.
The crosstalk effect between the 4 quadrants of the Redeye Nicmos chip is known to all users. This is a detrimental effect that can slightly affect the photometry in crowded fields, or when a bright object is in the detector's field of view. This problem is likely to be caused by the pre-amplifiers of the read-out electronics, which have four channels that may not be perfectly isolated. A re-design of these pre-amplifiers is being completed at CFHT. We propose here a method to correct a-posteriori for this defect during the data reduction process.
The model we have developed assumes that each quadrant first induces a ghost image in each other quadrant (0.15% crosstalk). This ghost image then undergoes an exponential decay (1/e at x = 7 pixels). A short IDL code has been written to remove these ghost images. This code is listed below. It should be understandable even for non-IDL users, and is simple enough to be easily translatable in other languages.
Figure 3: Raw images (left> and crosstalk corrected images (right) for a Dark (top) and an image of a Galaxy (bottom). The dark areas in the lower images are due to the coronograph mask used during this run.
function quadcor, image ; Function to correct for the crosstalk in CFHT Redeye images ; Syntax : ; result = quadcor(image) ;where Image is the raw image and Result is the crosstalk corrected image. ; Initializations frac = 0.0015 expo = 7.0 ghost = image * 0. ; Create the ghost image, i.e. basically the actual image is convolved ; by an exponential decay along the detector lines (towards positive ; X coordinates). "shift" is an IDL function that translates the whole ; image by an integer number of pixels (here, i pixels in X, 0 in Y) : for i=1,30 do ghost = ghost + exp(-i/expo) * shift(image,i,0) ; Duplicate each quadrant in each ;other quadrant : host = ghost + shift(ghost,0,128) + shift(ghost,128,0) + shift(ghost,128,128) ; Subtract the ghost image from the actual image : corrected = image - frac * ghost return,corrected end
Figure 3 shows some examples of uncorrected and corrected frames. Since the correction operation is purely linear, there are no detrimental effects on the photometry.
The OSIS project is getting close to the fabrication phase. This project is to replace the SIS optics by a new optical train covering 0.37 to 1.8 micron. This will allow to extend SIS properties as follows: (i) use of the Redeye detectors for stabilized imaging in the 1-1.8 micron domain, (ii) multi-slit spectroscopy in the 1-1.8 micron domain with the Redeye cameras (2'x2' field), up to 4'x4' fields when new arrays will be available, a unique capability worldwide; one expects S/N=10 in 1 hour for H ~19 for spectroscopy at R=200; (iii) wider field of 3.7'x3.7' in the 0.4 to 1 micron domain with a 2Kx2K CCD, other properties remaining the same as the current SIS. This upgrade of SIS is a collaboration between DAO, OPM and CFHT. The optical design has been completed at DAO and shows excellent performances. The selection of a manufacturer for the optics is in the final stages, fabrication is expected to start shortly. Mechanical designs for wider adaptor to accommodate both CCDs and the Redeye cameras is being designed in DAO. Exact specifications for grisms for the near-IR are being prepared. We expect integration by the end of 1995I with acceptance tests on the sky early in 1995II. SIS will not be available for several weeks in 1995II while the new optics and mechanical parts are integrated. Users interested in this new mode should check for the status of this upgrade prior to sending proposals for 1995II.
O. LeFevre (Observatoire de Paris-Meudon), D. Crampton (Dominion Astrophysical Observatory), J.-G. Cuby (ESO and Observatoire de Paris-Meudon)
In an attempt to simplify the interaction with MOS and SIS instrument, a graphical user interface and status display was developed. Information about the calibration lamps and cassegrain bonnette status is also included. The status window displays the light path inside the instrument, this visual display will allow the observer to quickly check and modify the configuration of the spectrograph and the calibration subassembly (see figure 4).
Since the release to the observers of the new SIS guiding system (see previous bulletin), it has been used for several multi-spectroscopy and high resolution imaging programs. Some minor changes have also been made in the command software.
It is now possible to guide safely on stars as faint as R ~ 18.5, i.e. close to the limiting magnitude of the POSS in red. Figure 5 shows the field that is available for finding a guide star, as compared to the science field (represented for a CCD with 2048x2048 pixels of 15 micron) and to the guide probe size. With such a field of view (3.5' x 4.5') and sensitivity, it is very unlikely that a suitable guide star could not be found, even on "empty" fields.
Figure 5: SIS field of view. North is at the top and East to the left. The Cassegrain Bonnette allows rotation of the field by -90 deg. and +90 deg.
For most programs in multi-object spectroscopy, it is acceptable to have the guide probe in the science field. On the other hand, it is better to avoid any occultation for direct imaging programs. This can be achieved with the X coordinate of the guide probe close to the maximum value of 4000, i.e. virtual Y coordinate for the CCD close to 2500 (see fig. 5). We recommend to observers to choose in advance the guide star for each of their fields. Sometimes, a slight decentering and/or bonnette rotation could be needed in order to avoid occultation.
Figure 6 shows the window displayed when activating the "guider" box during a Pegasus session. When entering the CCD coordinates of a star, as measured on a previous exposure, as X and Y, the target coordinates of the guide probe are calculated. They are displayed when the probe is in place. For instance, entering CCD coordinates (848, 1304) will yield probe coordinates (2063, 3872). Then, the star is found in the field of view (3") of the guide probe, close to its center. When clicking on the "update" push-button, the total counts in the APD (for 1 sec. integration) and the percentages of flux in each quadrant are displayed, as well as the position of the star's centroid. For a precise centering, use the arrow control boxes. Use the slide bars Wavelength to select step sizes (no more than 2 or 3 units) for each axis. Note that the best centering of the star does not correspond to the maximum ofthe total flux detected by the APD, because of the small gap between the 4 quadrants. This is mostly noticeable with excellent seeing. One should also look at the oscilloscope which gives a real time display of the star's centroid motion. The gain in resolution (active guiding vs Cassegrain bonnette guiding) is typically 0.1" for a mean seeing of 0.7". It is slightly better for fast guiding on a bright star and vanishes for the faintest usable stars.
Typical values for the integration time of the tip-tilt mirror command:
This supposes a dark sky (level around 3000 counts/s). During grey time, adjust the integration time in order to have at least S/N = 2 for each quadrant.
Bear is the FTS+Redeye bi-dimensional Fourier Transform Spectrograph. The raw data are the images of the primary interferograms. Reducing this data cube leads to a cube of monochromatic images ( defined by the actual resolution) over a given spectral range.
We have developed at CFHT a software tool that allows easy visualization of the reduced data cube, and provide a handy way to extract both the spectral and spatial information. This "cube viewer'' is designed to work in the Interactive Data Language (IDL) environment.
An example of how the graphic interface looks like is shown in the figure 7. Modes and actions can be selected by on-off or action buttons. There are two separate displays : for the image (left) and for the spectrum (right).
There are basically two modes of display, One-dimensional (1D) and two-dimensional (2D) :
This tool has already been used by several observers inside and outside CFHT. More information and a complete (though short) user manual can be obtained from the author at email@example.com.
Several important technical problems have delayed official release of the high resolution coude f/4 spectrograph, now called GECKO. Several observers have requested and used GECKO with good success, but the status has been considered developmental until certain goals for performance and reliability are reached. The reader is reminded that GECKO is a single order spectrograph using a mosaic of four, 316 g/mm echelle gratings to cover a 30 cm square beam from the collimator. Order separation is by either narrow-band interference filters or cross dispersing grisms. It was designed for optimum use with Richardson-type image slicers and has an f/4 (monochromatic beam) camera employing a parabolic mirror and a three-element refractive corrector. The support system for the detector is remotely controllable with six degrees of freedom.
Although work still remains to be done, we wish to announce that as of 1 August 1995, the start of the 95II semester, GECKO will be supported to provide the following capabilities:
The current system gives the following count rates of a 0.0 magnitude star with a thick, UV-coated 15 micron pixel CCD at several common wavelengths (that are not necessarily at the peak of the order response):
Wavelength Order electrons/sec/column 3130 A 18 360 4047 A 4 6,000 5890 A 9 34,400 6707 A 8 25,600
The accompanying figure gives an illustration of the excellent resolving power of Gecko versus the f/8.2 spectrograph. At top is a spectrum of the Na I D lines of the lambda Bootis star HD 192640 obtained with the f/8.2 spectrograph, the 600 l/mm grating in first order, and the blue image slicer. The resolution is approximately 20,000 in this configuration.
Figure 8This resolution was sufficient for the discovery of a shell component to the lines in this star but is likely not high enough to permit a conclusion regarding the circumstellar or interstellar origin of the shell lines. The bottom spectrum was obtained with Gecko in 9th order and the red image slicer and has a resolution of approximately 120,000. Note that some fringing caused by an order sorting filter is apparent in the Gecko spectrum but a recent modification to the filter wheel mount should remove this problem in the future.
J. Glaspey, D. Bohlender