One of the joys of astronomy is the travel we get to do. Big observatories are expensive and so are located in the mountainous regions around the world. The best locations are isolated island mountains, such as Hawaii, or ocean-side mountains, such as the Andes in Chile.
UF is a partner in the Gran Telescopio CANARIAS, or GTC (pictured above). GTC is on the top of the volcanic island La Palma, at an altitude of about 23oo meters (7400 ft). La Palma is nearly ideally located; it’s several hundred km from the coast of Morocco and is not heavily populated. GTC is one of several large observatories that make up the Northern European Observatory (a sort-of counterpart to the European Southern Observatory in Chile). Like Hawaii, La Palma is at about 30 degrees north latitude and has an excellent view of the northern sky and an ok view of some of the southern sky.
So far while in grad school (7 years and counting) I’ve spent about 6 weeks at the summit. My first visit was in 2012 and was a day trip that a group of students took; the International School of Astronomical Instrumentation (IScAI) was hosted on the nearby island of Tenerife at the Instituto de Astrofísica de Canarias (IAC), which is also the host institution for GTC. We got to tour several of the telescopes that neighbor GTC. The coolest thing (aside from seeing what €200 million and decades of determination can get you) was that GTC is so finely balanced you can move the whole telescope by hand. It weighs in at around 400 metric tonnes and yet you can move the telescope up and down with ease from the catwalk, and even spin the entire telescope by pulling yourself around on the catwalk. It does has an huge amount of rotational inertia, so you do have to strain a bit to get moving.
My second trip was to unpack CIRCE from its container (above, left), which I also had a hand in building. We arrived in June 2014 and spent two weeks uncrating everything and verifying in the lab that the instrument had survived. While stressful, it was a fantastic experience, made all the better by the fact that when we cooled down CIRCE and tested the detector, everything worked well (I got to travel around the island of La Palma while we waited for the temperatures to stabilize, which takes about 2 days). This is not always a guarantee with infrared detectors — cooling down from room temperature to liquid nitrogen temps has been known to kill them. On top of not killing the detector array we also nailed the clocking of the detector to a few milli-radians, meaning that we didn’t have to futz with the detector position and clocking. That’s important because small changes in tilt of the detector are every difficult to make with CIRCE (imagine trying to hold your camera in place on the lower left corner while moving the upper right corner a fraction of a millimeter, while in a vacuum and at cryogenic temperatures). As it was, we found that we were within a few milli-radian of perfect and so we left it at there.
CIRCE then spent a few months (literally) chilling in the instrument lab while we waited for GTC to install and commission the Folded Cassegrain Rotator. The FC Rotator (as we call it sometimes) moves in such a way as to cancel out the apparent rotation of the sky as seen by GTC (this is a necessary and vital component of modern large observatories, all of which use alt-az telescope mounts). The CIRCE team went over in December 2014 and installed the instrument on the rotator, but the control software wasn’t quite ready and science commissioning was postponed. However, the CIRCE team was able to do many of the engineering-type tasks required, such as verifying that CIRCE could take images and automatically deposit them in the GTC repository, and that communications sent from our control software could make the telescope slew.
During engineering commissioning we discovered that the environmental cover (the shutter attached to the front of the instrument designed to protect the entrance window from dust and frost) was experiencing enough friction to bind up; it was unable to open and close when commanded, so we had to design and build a replacement.
In March 2015 I took part in science commissioning. We had two goals before we started observing in earnest: 1) re-wire the motor control chassis, and 2) replace the environmental cover.
CIRCE’s motors are simple stepper motors that we refurbished for use in vacuum. They’d had given us no troubles in the lab at UF or while testing in the instrument lab at GTC using short (~6 feet), but when we used the ‘flight configuration’ cabling for use on the telescope (which are much longer, ~30 feet), we encountered a swathe of problems.
It turned out that the 30 feet of cables Alan, Nick (our instrument scientist), Scott (our electrical engineer), and I had made in early summer made excellent radio receivers, and the interference that the cabling picked up made it impossible to drive the motors; the signals to move the motors, simple voltage pulses, were swamped out by the radio frequency interference. Poor Alan had to rewire all of the connectors and cables, stripping out and redoing all of the work we’d done in early summer so that the wiring carrying the motor signals were separated from the ground wires and redoing the cable shielding. Once that was done, things seemed to work in the lab and we were able to move each of the mechanisms.
After installation on the telescope (which is terrifyingly exciting to watch, by the way), we noticed that some mechanisms had stuttering problems, meaning that they would not complete movements from position to position, or sometimes not move at all. Scott suggested rewiring the motor controller box (which was a snarl of wiring, designed by the previous electrical engineer) as a way of cutting down on the gremlins, as we referred to the problems. However, time at the summit is precious, and Steve (CIRCE’s Principle Investigator) decided to save that for a later trip. So, in early March, Scott and I went up to the telescope, pulled the motor chassis out of the electronics rack, and rewired it. We also tested the new environmental cover, which worked like a charm.
We replaced the environmental cover, which meant pulling CIRCE off of the telescope, setting it on precision wooden blocks (pictured above), and removing about 2000 tiny screws (actual number of screws: ~30). We also installed a ‘blower’ (pictured at left),
which was designed and 3D printed out of plastic. Its job is to blow compressed air across the entrance window to prevent any fogging or frost from building up on the entrance window . We iterated quite a bit on the design of the blower, which is one reason 3D printing is so great — it costs nearly nothing to produce highly complex shapes with interior hollows that are quite literally un-machinable using traditional subtractive machining techniques.
The next week was full of observing. In the course of 7 nights we observed for 3 first-halves and 3 second-halves of the night, with great weather for most of the run. We sadly didn’t kill all of the gremlins in our motor chassis, as we still had a few difficulties, but the frequency of failure was brought down to acceptable levels. I did a bit of real-time data analysis using superFATBOY (the Florida Analysis Tool Born Of Yearning for high quality astronomical data), which is written in Python by Dr Craig Warner, one of our software engineers.
We had a very successful science commissioning run, marred only by the fact that our half-wave plate rotator mechanism, which was a steel belt driven rotating mechanism, failed to move at all, most likely due to stiction (sticky friction encountered when cooling parts to cryogenic conditions – keep in mind most lubricants make excellent rocks at liquid nitrogen temps!). Even with that failure, we had a spectacular science run and observed many nifty targets, including an extremely red Gamma Ray Burst counterpart search (paper in progress! but not by me), an amazingly deep look at the lensing galaxy cluster SDSS 1004+41, and a few others. Our first target of the run was the star Betelgeuse, which we had hoped would be useful for figuring out the rotational axis location on the detector. When we moved CIRCE off of the Folded Cass rotator and reinstalled it, we used 3 brass locator pins to define the instrument position kinematically. CIRCE is over 3 feet in diameter (<1 meter), and we wanted to test how well we could repeat its position on the rotator. So, by using a bright-ish star (Betelgeuse was WAY too bright!), slightly defocused, we could calculate its center of rotation by rotating CIRCE in 15 deg increments and find the ‘barycenter’ or average position of the star. The repeatability of CIRCE turned out to be extremely good, to within a few millimeters of where it was before — not bad for a 1 Tonne, 1 meter in diameter instrument! As I mentioned earlier, we didn’t get rid of all of the gremlins with the motor chassis rewire, but many of our previous troubles were lessened.
My next journey to GTC was in March 2016 for its first Servicing Mission. We brought a replacement mechanism for the half-wave plate that Yigit (a fellow graduate student) designed. I wasn’t involved with removing CIRCE from the telescope (I followed Steve and Alan a day later), but once I arrived we got to work opening up CIRCE. We replaced the half-wave plate mechanism, added new filters to the filter wheels, and installed the grisms we had liberated from the FLAMINGOS-1 instrument. This necessitated milling out some of the material from the base of the filter box, which we did at the William Herschel Telescope just down the road (they have a full machine shop for parts fabrication).
Alan and I balanced the wheel holding the grisms using steel washers (pictured above to the right) – calculating the weight and position required of the counterweights is not a simple thing because the center of mass of the grisms wasn’t known a priori, nor was their exact clocking.
We then reassembled the instrument, taking care to clean and regrease all of the o-rings. It was an action-packed week, capped when I connected the flex cables together. We use two that are each about 15 inches long; carry the data signals from the detector through the vacuum jacket. We spent a lot of time at UF working on how to hold these two cables together; the flex cables themselves are tricky to work with, as they are enclosed in the Dewar and are totally inaccessible once the air is removed and the instrument cooled down. The manufacturer recommends using tape to hold together the two cables at the connection point; they are springy and have a tendency to spring apart if they’re under tension. The CIRCE team solution was to (of course) 3D print a 3-part clamp
that holds the lower flex cable in place while installing the upper flex cable (which is firmly and permanently attached to an aluminum feed-through plate). It’s a tricky job, one that requires two people working in each other’s space while working with tiny, tiny screws mostly by touch. We did it without a hitch, and I managed not to electrocute the detector, which is directly wired to the flex cables. It’s a bit nerve-wracking to do.
After we closed up and cooled CIRCE down, we ran it thru some lab tests to verify that we hadn’t done anything terribly wrong. We discovered that one of our filter box motors doesn’t work while cold (it worked beautifully while warm). It’s a bummer, but not crippling. We spent quite a bit of time taking spectroscopic standards with the grisms and determining where the spectra were on the detector and what filter combinations worked for us. We also replaced the plastic blower with a stainless steel one (pictured at right) that we ordered from Xometry.com (they are like a broker for 3D printed or CNC machined parts).
After we checked CIRCE out in the lab, we installed it on the telescope. I was given a bright yellow and orange safety harness. We worked alongside GTC engineers to ensure that CIRCE was safely lifted to the Folded Cass station (about 50 feet above the dome floor), and managed to have it installed on the rotator after we broke for lunch. We then ran CIRCE thru her paces, checking that the detector was still functional. However, when we tried to move motors, we were disheartened that none of them moved.
We spent about 3 hours troubleshooting the problem, measuring voltages and amperages and anything else we could think of both inside the electronics rack and at the cable connections. Eventually, after much skull and anxiety sweat, we tried running an extension cable from one of the Nasmyth platforms to the motor chassis directly (bypassing the electronic rack AC switch and Baytech control unit). It turns out that somehow the power delivered to CIRCE via the telescope elevation ring is ‘dirty’, meaning that the pulses that drive the motors are contaminated with noisy, unwanted pulses. By using another power source, we were able to turn all the motors (sans the bad filter wheel motor) and fully check out CIRCE. It took us into the evening and we missed twilight (important for CIRCE because we generate flat fields using the fading IR twilight). However, we were able to observe for about 4 hours and verify that we had both put CIRCE back together properly and that the new mechanism worked as expected.
Here are a few pics that I decided to add because, why not.