MAPS 2023A Day 5: Sunset on the ASM run

Manny caught an unsuspecting grad student sunset-watching.

Well, the fun had to end sometime. This post marks the official end of the MAPS 2023A run, after a weather-ful last two nights.

One of the big perks of mountain observing in AZ, especially in June, is replacing the 100ºF highs of the valley with mountain temperatures of 40-60º. The first night, as we braved the bitter winds of civilian twilight for the MAPS team photo next to our colleagues with parkas and patagonias, Joseph and and I had the sinking thought that maybe our office-AC outer layers wouldn’t measure up. For the first few nights, we shivered through the few minutes we had to be outside and remembered the summer heat wistfully.

Lady bugs huddling in the outside cracks of MMT for warmth.

Careful what you wish for. After the chilly 50s of the first few nights, our second to last night we got hit with an uncharacteristically balmy and breeze-less front. Lovely for humans, but very bad for ASMs who need weather-based cooling. So yesterday, we spent a crystal clear night waiting on and off for a hot, crabby ASM with no wind to soothe it.

Tonight, however, the chilly breeze was back, and with it the clouds. We got a sunset so spectacular that the whole control room ran out to see it. Did I miss a green flash? Yes. Yes I did. Apparently I haven’t learned the proper technique even with weeks of observing at LCO.

Even Joseph, perched at the ever-rotating window, had to admit the sunset was worth going out to see.

Though astronomers might love a good sunset, astronomy doesn’t like the clouds they can bring. We ended up being clouded out for a good portion of our final night, an anticlimactic way to end a week of speed-learning a CACAO system. As it became clear that the clouds weren’t going to clear out anytime soon, person by person the crew took off to bed, prepping for mountain-top departure in the morning. What was left by 3am was the skeleton crew, the bare minimum to keep everything running. Brian for the telescope, Jaren for the science camera, Manny for the ASM, Me for CACAO, and Joseph for morale.

Skeleton crew selfie right before closing.

Though Joseph and I were last minute additions to MAPS and the learning curve was steep, it was such a privilege to be able to help with an AO system on such a historical telescope. One that has been on the forefront of segmented mirror alignment and Infrared science. Hope to see you soon MMT!

The last sunrise.

Song of the Day

Sleepy Eyes – Eerie Wanda

MAPS 2023A Day 4: What’s a couple of shmims between friends?

After the previous night’s untimely clouds, we were fortunate to have clear skies and moderate-to-good seeing all night. The MIRAC-5 team continued work on their instrument, which the adaptive optics operator Eden found extremely useful for its fast video feed showing how (and if) our adaptive optics experiments were improving their images.

I, however, mostly spent the night shmim-wrangling.

MAPS (like its cousins MagAO-X and SCExAO), uses shared memory images—shmims—to relay data at high speed from a wavefront sensor, through an adaptive optics loop, to the point where commands are handed off to the adaptive secondary mirror.

There it is! Way high up off the ground!

Of course, a shared memory image is just a hunk of memory with a hint about the type of data it contains (floating-point numbers, integers of various sizes, etc.). One could just as easily use a shmim to store a vector, or a single number, or a data cube.

One such vector was my target yesterday: the vector of “modal gains.” Each entry in the vector scales the system’s correction for that mode by the factor you provide, allowing you to correct low-order modes more strongly while easing off the high-order modes that you may not be able to control as effectively. It’s sort of like the EQ sliders on a stereo.

What’s he listening to?

Still, one may ask, what is a “mode” in an AO system? The answer: something too abstruse, dear reader, to bother this blog with.

That won’t stop me, though! You know how the optometrist fiddles around with their equipment to find the best focus for your eyes, and only then starts in with the next level: correcting for astigmatism? Well, AO systems fiddle around many times per second to get the best focus for the telescope, and the best astigmatism correction, and a few more besides. It turns out there are “levels” beyond astigmatism, accounting for even more subtle changes to the image. We need to correct those for the sharpest image possible.

That’s the short version, anyway. The long one involves math. If you’re looking for a low-stress beach read this summer, I can recommend Adaptive Optics for Astronomical Telescopes by Hardy (1998).

In MagAO-X we divide the gain controls into blocks of modes and use buttons to bump them up or down to improve our correction. MAPS does not use all the same MagAO-X software, and the two of us from the MagAO-X team were missing the fun of clicking a lot of buttons really fast. So, while Andrew develops the real, production-grade, network-enabled button-pushing infrastructure for MAPS, I brought a little bit of MagAO-X from home.

This tool replicates the UI we have for MagAO-X gain tuning, with the additional trick of a parametric gain curve. Rather than tuning the modes by blocks, one can type in some parameters to define that pink curve with a different gain value for every single mode. Clicking “Apply” then fires the numbers off into the appropriate shmim, where they are applied to the AO loop.

Of course, all the software in the world can’t change the laws of physics. Summer seems to have finally made it to Mt. Hopkins, and with it the need for active ASM cooling has become acute. It turns out that bending glass mirrors takes a lot of power, and some of that power escapes as heat. That heat, in turn, makes the whole system (red-faced and panting) call for a time-out while it collects itself.

It’s not often I wish a mountaintop were colder and windier. Tomorrow’s our last night, and we’ll have to see if we can keep our cool when it’s a balmy 54ºF at 1 A.M.

Song of the Day

“Que Sera” by Wax Tailor. It’s a groove, trust me.

MAPS 2023A Day 3: MIRAC-5 has entered the chat

After some berating from @jlong, here is my inaugural post! First, an obligatory observatory picture.

The MMT at sunrise with the full moon on the horizon

The MIRAC-5 team arrived Friday, June 2nd to prepare for installation of the latest iteration of the instrument. Originally conceived in 1988 by Bill Hoffmann at the University of Arizona, MIRAC (Mid-Infrared Array Camera) is an ever-evolving mid-IR camera built for observing between 1 and 18 microns. It has a long and illustrious career of testing out mid-IR technologies, including various detectors and nulling interferometry (test bed for LBTI nulling). As the name suggests, we are on the 5th iteration after almost a 10-year dormancy period!

MIRAC was pulled out of retirement and refurbished to house a new mid-IR detector: the 2k x 2k GeoSnap-18 array manufactured by Teledyne. This GeoSnap is an engineering grade device with a wavelength cut-off of 13 microns. Designed for high-background applications this array is perfect for ground-based mid-IR astronomy! The primary features are wavelength sensitivity at H, K, L, M, and N-Bands (science-grade arrays will extend down to 0.7 microns), well depths of 1.3 million e (twice that for science-grade), and continuous frame rates of >85 Hz without any reset or data acquisition overheads! After commissioning of MAPS and MIRAC, we plan to use these instruments to characterize exoplanet and brown dwarf atmospheres in the mid-IR, complementing the capabilities of JWST MIRI.

The GeoSnap engineering array in its storage contain. Of the full 2048 x 2048 pixels, only one quadrant is bonded with photosensitive material (HgCdTe).

Anyway, off came the PISCES camera and on went MIRAC. Comparatively, MIRAC is a bit of a beast.

U. Michigan graduate student Rory Bowens (right) and U. Arizona PI Jarron Leisenring (front left) and U. Michigan PI Michael Meyer (back left) posing next to MIRAC-5 mounted to the MAPS top box.

Upon hooking up the bazillion cables, we realized that the fiber extender used to transfer the data from the GeoSnap detector to the computer (located in an adjacent rack) wasn’t working as expected. The only solution was to keep the original (short) cables attached and strap the computer to the instrument cart on the telescope. It didn’t seem to complain about the free ride.

After bringing the detector temperature to a cool 40K, we were able to get the acquisition software working on the first try thanks to the diligent efforts of our software engineer Dennis Hart and successful implementation of the @jrmales’s rtimv visualization software.

Early image of a controlled PSF at K-Band running with a frame rate of 5 Hz. Hot pixels have not been masked out.

Throughout the evening, we successfully acquired a number of stars, co-aligned with MAPS, performing continuous PSF display at high frame rates, tested internal pupil chopping operations, measured background levels, and a whole host of other commissioning activities. While a few challenges inevitably arose, which we will address during the summer, it has been a pretty successful and rewarding “pre-commissioning” run so far!

Additional Random Photos

Song of the Day

MAPS 2023A Day 2: “Now we’re starting to look like an AO system!”

Hello. It is your local CACAO expert here, beaming to you directly from the beautiful peaks of the Santa Rita mountains. I am currently being not-paid to convince a wiggly mirror to un-twinkle stars from a four story building that rotates all night long. The whole experience feels like an internship at something between a castle fort and sea-bound vessel. For the record, I do not think that buildings this big should merry-go-round. However, I will admit that this one is impressive.

We continue to be supported from CACAO sages across the globe. Olivier called in last night about to board a Hawaii-bound plane, and Jared logged on from one of the fastest trains in the world. (Thank you wise ones, for being so generous with your travel time.)

Kagayaki 533 for Kanazawa
MAPS at 160 mph

Tonight Olivier split his time between the Subaru team and ours, remotely helping us tie off some CACAO loose ends. Finally, after the trials and tribulations of last night’s clouds and cranky mirrors, we finally have a happily closed loop! Early tonight we had the DoCrimes response matrices loaded up and performing corrections. We’re still working on some other orthogonal control algorithms with CACAO, but for now, enjoy the success of the day, from two different perspectives:

First, mine, as the AO operator:

Here we turn on the Crimes, and see the WFS pupils (top left) flatten, acquisition source (middle) shape up. You can see on the bottom the ASM’s positon, current, and temperature per each actuator.

Second, Jorge’s as the observer (video by Joseph):

AO off, tip tilt on, full AO loop on.

“Now that’s what I call an AO System!” Manny.

The team was still taking data when this photo was taken.

Summer runs are funny in that you seem to always be running out of time. Sun-down at 8pm to sun-up at 4am isn’t as long as you’d think. We’ve been running our observations into sunrise more often than not.

A wave from the light of the full moon!

That said, we still have a few more days to get our loops more stable, but what we’ve accomplished so far has made me proud of my small, button-pushing part of the MAPS team. I have had such a great time working with the crew, have been charmed by MMT, and despite all worries I do believe Joseph and I have proven ourselves helpful.

Song of the Day

TURN THE LIGHT by Danger Mouse

MAPS 2023A Day 1: Clouds, Clouds, Clouds, and Adaptive Secondary Mirrors

The night started off with clouds and the group had a nice and easy time getting to start their systems. Nothing like having time to catch up and get lunch together.

At midnight the weather decided to clear and we had high hopes of starting where we left off last night. It is always good to have high hopes. The ASM decided to change the way it had been functioning for the last night and time was taken to troubleshoot and verify the issue.


What will you hear when you are in the MMTAO control room:

  1. ASM
  2. CACAO
  3. MILK
  4. CHAI
  5. DoCrime
  6. Pisces
  7. TBX

In a sentence:

The TBX cameras sent images to CACAO that is using MILK to send commands to CHAI which caused the ASM to DoCrime which allowed Pisces to see clearly.