MACAO-CRIRES:

curvature adaptive optics for CRIRES

• on VLTI, the Very large telescope interferometer and its dedicated instrumentation; with its MACAO adaptive optics located at the Coudé focus of the 8-m telescopes and with the 1.8 m Auxiliary telescopes, the VLTI will provide both a high sensitivity as well as milli-arcsec angular resolution provided by baselines of up to 200m length,
  
• on SINFONI, an Integral field unit spectrometer, providing 1500- to 4000 resolution spectra of a 32x32 pixels area; each pixel covering 25x25, 100x100 or 250x250 mas on the sky, and
• on CRIRES, a high-resolution (R=100 000) infrared spectrometer, with a 0.2" wide slit on the sky. Its MACAO-based adaptive optics will improve both flux and spatial resolution on the targets. 

 

Commissioning run

	R=10	R=13.5	R=15	R=17.4
Ensquared energy, good seeing	63%	58%	-	-
ensquared energy, bad seeing	56%	50%	43%	30%
best Strehl achieved	61%	47%	25%	(6%)

Performance of the system: good seeing conditions were considered when the seeing was below 0.78”, while the bad seeing conditions were considered when the seeing was above 0.73” (DIMM measurements). The Strehl values and ensquared energy indicated are given for K-band, taking into account the aberrations introduced by the infrared test camera (82%), and are therefore pessimistic. The Strehl value is only indicative for R=17.4 star, as the meaning of the Strehl starts loosing its pertinence at such low values.

side view of the WFS optics, including the fibre bundle head

                               


some views of the fibre bundle. Assembly (top left), cut and scale of the lenslet arrays with a 1 Euro coin (top right), and close-up of the lenslet array as mounted (bottom). On the latter, the entrance of the fibres can be seen as a green circle in the middle of each lenslet. Pictures courtesy S. Tordo, C. Dupuy.

At the connectorised end of each fiber, an individual APD (avalanche photodiode) has been placed, which signal will be transmitted to the RTC. A dedicated board -the counterboard- synchronises the vibration of the membrane with the detection of the photons, in order to measure the curvature at the level of the subaperture; the curvature follows:

The 60-elements curvature vector is then multiplied by a control matrix, to provide the command vector, sent to the deformable mirror's HVA (high voltage amplifier) in the form of 60 voltages. Additionally, a tip-tilt component is extracted from the 60 voltages, and sent to a dedicated tip-tilt mount. This allows reducing the load of the deformable mirror. The HVA is providing a +/-400V range, allowing to compensate with the DM up to 1.2" of seeing at 3 sigma. The controller used in the RTC follows the control law:

y[n]= y[n-1] – K_I/2 (x[n] + x[n-1])- K_P (x[n] – x[n-1])

where y is the position of the DM (in voltage, for example), and x is the voltage offset to the ideal WF, as measured by the WFS. n is the time index, n-1 is the index of the values provided during the last loop cycle. K_I and K_P are close to the integral and proportional gains classically defined in control theory.

Control strategy for CRIRES. The classical sheme for AO applies: fast loop with the deformable mirror, realyed by the TT-mount loop for lower frequencies tip-tilt components, the TT-mount offloading itself on the telescope pointing control. MACAO-CRIRES feeding a spectrograph, a slit-viewer allows refining the pointing of the instrument at the observation wavelength (or close to it), and this loop is described in the second sketch, showing how the slit-viewer controls the field selector of the AO system for correcting the small offsets which appear between the visible AO-guiding and the infrared image on the 0.2" slit.

1. the phase lag calibration aims at synchronising the membrane mirror and the counterboard; the procedure applies a non-flat shape to the DM, and varies the phase lag so as to minimise the curvature signal seen by the WFS. A 90º offset is then applied to the counterboard which produces then the maximum signal.
   Frequency: weekly, during daytime.
   
2. the RoC calibration: the radius of curvature defines the optical amplification of the curvature signal; the smaller the RoC, the bigger the amplification. There is a linear relationship between RoC and voltage amplitude applied to the membrane mirror, which can be calibrated by applying a known defocus to the calibration stage, and measure the curvature seen over the pupil while varying the voltage applied to the membrane mirror.
   Frequency: daily during daytime.
   
3. The flat reference is computed each time another calibration is to be done, and before any observation where the adaptive optics can not be used (deeply embedded object, or extended object without contrasted central feature). With the appropriate procedure, its quality stays below 100 nm rms over a 4 hour period.
   Frequency: daily during daytime, and before any non-AO operation of the instrument.
   
4. Interaction matrices must be computed for a series of interaction matrices, covering the optimal range of RoC used by the system. An interaction matrix records the curvature measured as a function of the voltage applied to the DM. It is therefore a 60x60 matrix which pseudo-inverse provides the control matrix used by the system in closed-loop. The use of the pseudo-inverse comes from the fact that a piston term produced by the DM produces a 0 curvature, not seen by the WFS: it is necessary to filter out one mode of the system to get an optimal correction.
   Frequency: weekly check, daytime.
   
5. The pupil centring is measured  by estimating the curvature signal spread over the neighbours when an electrode is activated. A good centering will provide a pattern symetric around each electrode. Centring, rotation of the pattern and shrinking of the pupil can be measured together by acquiring a single interaction matrix.
   Frequency: daily check, daytime.
   
6. Secondary interaction matrices are computed, by measuring the voltage pattern required on the DM to counteract a tip-tilt introduced by the mean of the tip-tilt mount.
   frequency: weekly check, daytime.

wavefront corresponding to the system modes of CRIRES, and their respective eigenvalues. Left: Secondary interaction matrix plot (the two lines represent the DM voltages required to get a pure tip or tilt; the curves have been shifted up and down to improve their visibility) Right: flat vector reference; the 3-sigma of the vector voltage is below 10% of the available stroke, the maximum used by one electrode is 13%, and two electrodes use more than 10% of the available stroke.

interaction matrix example; the almost diagonal pattern illustrates the advantage of the keystone geometry for the lenslet array, allowing a very good correspondance between actuation and local curvature. The matrix values are given in curvature per 400V (maximum stroke) for the DM. difference IM-(IM)T for a pupil shift (central part). Such difference matrix is used as projection vector to measure pupil decenter.