Magellan AO

The NAS and all the W-unit and VisAO electronics landed in Chile early on Tuesday morning. The first part of MagAO has now arrived in Chile!

Our other big news is that the ASM – meaning the thin shell itself and the ‘unit’, which means the reference body and attached electronics – has left Arcetri and is on the way to the airport in Firenze. This process will take a few days for customs etc. The pictures below, courtesy of Armando Riccardi, show the final stages of packing the ASM and loading it on a truck.

Our mirror is swinging through the air! This is the "small box", containing the 85cm wide by 1.5mm thick thin shell of our adaptive secondary mirror.

The "large box" being loaded on to the truck. The "small box", shown above, is packed inside this box. This system is carefully designed to protect our delicate mirror during the flight.

Here we see the reference body and electronics being readied for packing in their shipping box.

This video demonstrates the MagAO high-level software GUIs used to acquire the star, set up the AO system, and close the loop.  The entire process takes about ~3-4 min. at this time.

(Filmed by Alfio, narrated by Laird, cameo by Katie operating the VisAO camera)

Go to http://www.youtube.com/watch?v=wSiFoG8qgKI to view the video in high-def.

Description:
After the telescope slews to a new target, the guider will acquire the star to within 4” on the Technical Viewer (CCD 47 in AO acquisition mode; otherwise CCD 47 is the VisAO array).  Next, the MagAO Command GUI controls the AO system and is operated as follows:

1. PresetVisAO — Transforms the CCD 47 from its role as the Science Camera (VisAO) to its role as the AO acquisition camera (Technical Viewer).
   • Takes control of the CCD 47, gimbal, and filter-wheel 2 and 3
   • Opens up filter-wheel 3 (coronagraph stops and SDI narrow-band filters)
   • Centers the gimbal
2. PresetAO — Configures the board for acquisition, and uses the estimated NGS magnitude in order to determine the AO system parameters (frame rate, modulation, binning).
   • Start with a guess of the NGS mag (entered by hand, or will be read in from the starlist provided by the astronomer) — required for appropriate speed and binning to get good SNR photometry. These settings are read from a lookup table
   • Opens up filter-wheel 2 (SDSS r’, i’, z’, and 950 long-pass filters)
   • Moves gimbal to point off axis to take CCD 47 darks
   • Sets binning and framerate on both CCD 39 and CCD 47 based on the estimated NGS magnitude.
   • Places the beamsplitter wheel in the dark position for CCD 39 darks
   • Once darks are taken, re-centers the gimbal
   • Wait for all movements to finish (stages, filterwheels, etc)
3. AcquireRef — Aligns the guide star onto the pyramid.  Metric is position of the star on the Technical Viewer (CCD 47).
   • Finds star on Technical Viewer
   • Finds offset from pre-determined home position (green cross)
   • Moves X,Y Bayside stages (entire W-unit) to remove offset
   • The above steps are performed iteratively until the star is on the green cross to within 0.2 mm of stage movement
   • Set filter-wheels for AO. Filter-wheel 1 (beam splitter) will be placed in the correct position based on star magnitude.
4. AutoCenter — Fine-tuning of the alignment onto the tip of the pyramid.  Metric is pupil illumination on WFS camera (CCD 39).
   • Button on the CCD 39 image viewer
   • Moves X,Y Bayside stages to even out the illumination (i.e. zero out tip and tilt)
   • Iterative process; Final precision is ~few microns
5. StartAO — Closes the loop.
   • First closes the loop with only 10 low-order modes to center up the camera lens (the camera lens loop aligns the pupil images to keep the illuminated pixels constant to 1/10th pixel)
   • You can next load artifical turbulence in the lab to simulate on-sky AO
   • Sets the frame rate, modulation, and binning determined in PresetAO
   • Closes the loop with all modes (400 modes in bin 1; 120 modes in bin 2; 50 modes in bin 3; 28 modes in bin 4) and low gain
   • AutoGain starts and iteratively guesses-and-checks the gain for each mode grouping (low, mid, and high) that minimizes the slope RMS
6. Finally, we have an optimized closed loop. The instrument can be turned over to the astronomer for diffraction-limited data acquisition

The MagAO project hosted our Pre-Ship Review (PSR) last week at Arcetri Observatory in Florence.  The purpose of this review was to ensure that lab work on the AO system is complete and that MagAO is ready to move to the telescope for on-sky commissioning.

Six external reviewers (4 in person and 2 by videocon) gave us their time and attention for 2 days in real time, as well as studying up beforehand and writing a report after the fact.  The scope of their expertise covered subjects from large telescopes, commissioning AO systems, instrument development, logistical implementation, real-time software, and adaptive secondaries.

The review lasted 2 days and included presentations, discussions, and a demonstration of the AO system in the tower, controlled from the chem. lab.  Topics included a description of the Magellan telescope, an overview of the MagAO project and instrument, the scientific justification, detailed technical descriptions of all subsystems, logistical considerations, and our plans for commissioning.  The reviewers considered everything in detail and gave us a report in the format of Findings, Comments, and Recommendations.

The review went very well — the reviewers were extremely thorough and helpful.  They included three requirements and four recommendations, as well as findings and comments.  We are working to address the requirements identified by the panel, before we ship to Chile.  Two have been addressed already this week, and the third is waiting for next week.

Here are some photos from the review:

The PSR panel visiting the solar tower lab during the review. The large black ring hanging from the ceiling, which we call 'The Nas', contains our wavefront sensor and the VisAO science camera. Below that are our electronics racks, which will actually mount on the side of The Nas at the telescope.

The MagAO PSR had lunch at Galileo's house. Galileo once walked this very balcony. This is our "album cover" shot.

A happy AO operator is a good AO operator. Katie Morzinski, MagAO's newest team member, operated the AO system for the PSR panel. On her screens you can see the 4 pyramid sensor pupils, the SDSS i' PSF, and a glimpse of our mirror status displays. In the background you can also catch a glimpse of the VisAO camera in action.

Alfio Puglisi explains some of the finer points of AO with pyramid sensors and adaptive secondaries, while Katie Morzinski and Jared Males concentrate on the loop.

At the after party in the MagAO home-away-from-home at Via Romana 89. The entire team celebrated a successful moment in the project after many years of 'blood sweat and tears'.

The MagAO PSR team celebrates the end of an intense 2 weeks. Speakers provided by Enrico "How many Watts RMS?" Pinna and music provided by Katie "Girl Bands" Morzinski.

The PI holds forth - toasting a job well done by the entire team. A beam of light descends down on him to mark the occasion.

The MagAO team, Magellan telescope experts, and Pre-Ship Review panel on the balcony at Galileo's house, inspired by a place where Galileo pointed his telescope to the heavens. Back row: Povilas Palunas, Al Conrad, Doug Miller, Jared Males, Ian Bryson, Marco Xompero, Runa Briguglio. Front row: Alfio Puglisi, Katie Morzinski, Simone Esposito, Carmelo Arcidiacono, Laird Close, Mark Chun

Yesterday we tested closing the loop on a faint R=13 mag star.  Ambient light from computer monitors in the test tower was too bright and so we closed the loop from the “control room” in the chem. lab.  It’s a good thing we got all the computers, desks, and high-speed ethernet links so that our control room is all set up and we could darken the “dome”!

We closed the loop on this R=13 star with the conservative 75th %ile seeing we’ve been using, 0.8” FWHM (r_0~15cm) and 33 mph winds (~15 m/s).  The frame rate was 200 Hz to get enough counts on the WFS with the pixels binned by 2.  Our resulting Strehl was 13% in z’-band which is about 200 nm rms wavefront error.  (The correction for our simulated turbulence not having high-order modes, and the correction for the double-pass on our DM in the test tower have a negligible contribution at this level).  Here is the image in z’-band:

For comparison, typical Keck AO wavefront control on H~7-9 mag NGS is also around 200 nm, or ~50% Strehl in H-band (result from 1000 Keck images from my dissertation).  So we are quite happy with this image quality!

Last week on Tuesday we closed the loop in the test tower, verifying our performance from the November run and validating our excellent wavefront control with MagAO.

We simulated 0.8” seeing (~15 cm r_0) in a 33 mph (~15 m/s) wind and closed the loop on an 8.5-mag guide star running at 1000 Hz with a pyramid modulation amplitude of 0.17 or about 3.5 lambda/D.  We are correcting 400 modes, each one with an optimzed gain.  The resulting image quality (right) is 85% Strehl at 982 nm effective wavelength, or about 65 nm residual wavefront error.  The slight elongation manifested as subtle lobes on the first Airy ring at a position angle around ~60 deg. is most likely caused by time delay along the wind vector.  Note that these images are linear scale, not log scale — and note the similarity of the ideal and the achieved PSFs!

Jared’s note at http://visao.as.arizona.edu/uncategorized/a-better-closed-loop/ explains more about the differences between our simulated turbulence (limited to low and mid spatial frequencies by our 585-actuator DM) and Kolmogorov turbulence in the real atmosphere.  We calculate that true atmospheric turbulence would add a moderate amount of wavefront error, and would bring our measured image quality of 65 nm WFE (85% Strehl) to more like 100 nm (68% Strehl).

Another difference between the test tower and the telescope is that our telescope simulator is double-pass on the ASM, so that our internal static error is doubled to ~75 nm rms as compared to what we should see on sky.  Correcting for this effect improves our estimated on-sky WFE to ~70-80 nm rms (75-80% Strehl)!  For comparison/sanity checking, LBT has thus far achieved 90% Strehl at H-band or ~85 nm rms WFE at best on sky in Arizona.

We (Laird, Jared, Katie, Alfio, Armando, Marco, Runa, Enrico, Simone, Luca, Carmelo, and Alessandro) have been working hard all week here at Arcetri to prepare for the Pre-Ship Review.

For today’s installment we have a video of the VisAO PSF as we test closing the loop and adjusting the gains.  Laird is narrating and Katie is operating MagAO in this clip. 

Descripion of video:  The Magellan AO system loop closing and opening at 1kHz on an 8.5-mag guide star in 0.8” seeing with 33 mph (15 m/s) wind. The image viewer is showing our SDSS i’ PSF, and we have selected a 32×32 subwindow of the CCD 47 (our VisAO science detector) to operate at 42 frames per second. You can see the impact on Strehl ratio (calculated from slope telemetry) and FWHM (fitting the PSF in real time) when the loop gains are low.

First we see the closed-loop PSF.  Then the loop is opened.  When we close the loop, we start with low gain and slowly ramp up first the low-order gain and then the high-order gain.  MagAO is operated with modal gains so that each mode (up to 400 controllable modes) can be operated with its own gain.  The automated gain algorithm searches for the best gain for each mode that minimizes the WFE.  However, gain can also be adjusted by hand, in groupings of low-order (tip/tilt), mid-frequencies, and high-order (above 100 modes for bin 1).

Laird and Katie on the walk to Arcetri, with the only snow we've found in Florence.

Due to the always too short amount of time we get to spend with our instrument here, we work some long hours. Late at night and on weekends the front gate to Arcetri is closed, so we have to go around the back way. It is actually a nice walk, and we discovered the last remaining snow in Florence. In the picture you can see the top of the Solar Tower, where our adaptive secondary mirror is mounted. It is performing very well, and we are putting the last touches on the amazing automation that was developed here at Arcetri.

Laird, Jared, and Katie (new postdoc) arrived yesterday for prep work and then for the Pre-Ship Review that will happen on the 23rd-24th Feb.  Katie is getting up to speed on MagAO while Laird and Jared verify that the system still performs as it did last Nov/Dec.  It’s chilly in Florence and we are running tests such as starting up the system, homing the motors, etc.  We are also setting up the Chem. Lab. as the “Control Room” via ethernet link (and radio for voice) in order to be able to demonstrate remote ops. during the review.  We warmed up the bench and got the system up and running, and will close the loop tomorrow.

The MagAO team (or at least part of it) has returned to Florence for another round of integration and testing. I (Jared) have been here for 2 weeks, and Laird and Derek just arrived. As soon as I arrived we successfully closed the loop again, with only minimal adjustments of the alignment using the X-Y-Z stages. The system came right back up, with less than 5 minutes of work, after 3 months of down time.

A raw CCD47 image in closed loop, with only about 5 minutes of start up work after 3 months of down time. The core is saturated, and since this is a single image with no reduction there is some pattern noise visible.

The last two weeks have been mostly uneventful, consisting of a lot of software development and debugging. In preparation for the next 2 weeks of work on the ASM and the CRO tests we removed the NAS from the tower on Friday. The two videos below show the process of lowering, and then tilting the NAS upright on its handling cart.

While the NAS is off the tower we will be fine tuning the WFS and VisAO camera alignment, as well as testing the AO to Magellan software interface. In about two weeks we’ll reverse the process, and mount everything back in the tower for some more exciting closed loop action.

Last week Laird and I had the pleasure of attending the Arizona Board of Regents meeting in Phoenix, and presenting a poster about the Magellan AO system and the VisAO camera. The session we were invited to was on the impact of scientific research on student’s education at Arizona’s universities. Click the image below to download the poster as a pdf. It contains an introduction to AO in general and visible AO in particular, as well as an overview of the MagAO project. We also took the opportunity to show off some of our exciting results from the test tower in Florence.

Click to download a pdf

Here is a short film of the VisAO camera at i’ (765 nm) in 0.8″ seeing (33mph wind) with the loop open (0.3% SR, FWHM~0.6″) and closed (55% Strehl, FWHM=0.027″) at 800 Hz (400 modes) in the test tower.

This is a >180 increase in peak counts (and >20x gain resolution) obtained by turning on the loop!

Note that the PSF is saturated out to the first Airy ring in this (rather poor) stretch. The red dot is the focus light from the videocamera (ignore it).

On July 21 we closed the loop again (but without wind on the outside of the tower). In these calm conditions we were able to obtain 55% Strelh at i’ (765 nm) which is excellent correction in 0.8″ seeing and 33mph winds. I attach a log10 stretch image below to show the very high contrast PSFs that are obtained with the Magellan AO system. This is slightly better performance then was predicted for an R=8 mag guide star in a 33 mph wind with ro=14cm at 0.55um. Normally at Magellan the seeing should be better (and the wind lower) than what we simulated here, giving us some confidence that VisAO will be an excellent visible AO science camera in >75% of the weather seen at the Magellan telescope.(clockwise) perfect PSF, the MagAO PSF, The AO ON PSF, and the AO OFF PSF (Log Scale)

Note by Jared: due to the way turbulence is simulated in the tower using the mirror itself, there is some missing power in the modes higher than ~585 (the number of actuators on our mirror). This results in an optimistic fitting error during these tests. We can estimate how much lower Strehl at the telescope would be due to this unsimulated turbulence using some AO theory (see Noll 1976). Our most conservative estimate for this correction brings our telescope Strehl down to 37%, from the 55% measured in the tower. Compared to the correction calculated by a more empirical method for the LBT (see Esposito 2010) this correction is probably a little large (resulting in a low estimate for Strehl). Even with this conservative correction, our tower results are exceeding our performance predictions for an 8th magnitude guide star by more than 5% Strehl. This is very exciting!