The Giant Magellan Telescope doesn’t seem so giant in the renderings, until you see how the mirrors are made.
The telescope will require seven total mirrors each 27 feet (8.4 meters) in diameter for a total combined diameter of 24.5 meters. Half of an Olympic size pool’s length. A little over four times the diameter of the James Webb Space Telescope.
According to the website, the mirrors are cast at the University of Arizona mirror lab and take four years each to make. They’re made from blocks of Japanese glass laid out in a giant oven. The whole process of casting the glass takes a year, from laying it out to the months of cooling, it’s a painstaking process.
Once the cooling is done there’s another three years of polishing to get the mirror just right. If you’ve ever had to set up a metal block for precision machining on a mill, you might have an idea of why this takes so long. Especially if you make that block a few tons of glass and the surface has to be ground to micron tolerances. A lot of clever engineering went into this, including, no joke, a custom grinding tool full of silly putty. Though, at its core it’s not much different from smaller lens making processes.
The telescope is expected to be finished in 2024, for more information on the mirror process there’s a nice article here.
micron? nope, it’s nanometer class precision required here!
Well, to be fair, that article does state the polishing *tool* needs to follow the mirror surface with an accuracy of a micron. It would be easy enough for a rushed and underpaid writer to interpret that to mean that’s the accuracy of the mirror surface. Elsewhere in TFA it does say the mirror surface is to get with 25 nm of the target.
four years? that must be a funny workplace to work there: are you finished yet??? nope, i need two more years :)
Well, and then there is 6 more mirrors after that, so make it a few more years.
They seem to be well on the way to completion. I believe 2 segments are already complete. From the pics on their website, segments 3 & 4 are polishing, and segment 5 is in the oven. They started with 140 tons of glass, and each finished segment is about 16 tons!
>Hey! Go back to work!
>Polishing
Heck of a coffee break!
https://imgs.xkcd.com/comics/compiling.png
So how much bad luck do you get for breaking a mirror that takes 4 years to make?
Ask the folks that screwed up Hubble telescope to get a baseline.
And having a backup hubble mirror fabricated that did not get send into space, but was actually correct.
Wasn’t the first Hubble mirror just a standard KH-11 spy sat mirror for earth abservation vs one focused to infinity, I understand that they had a few more unlaunched/obsolete NRO Keyhole-11 sats around stored for future Hubble launches if needed.
I wonder why with modern image processing a big mirror is still better than dozens or even thousands of say 1 meter mirrors. This is a serious question and not me trying to sound all smartass. From what I have seen the eyepiece is only for VIPs, otherwise any academic in service telescope is on the net on time share with a great CCD imager anyways.
One Large Mirror:
Pros: Diffraction limited to the one perimeter, Only one surface to align, can be thin to allow adaptive optics.
Cons: really big and hard to make, polish, move, install, clean, re-silver etc.
Multiple mirrors:
Pros: lightweight, faster individual fabrication, moving, installation, replacement etc.
Cons: Diffraction at every mirror edge, off-axis shape for segments (unless spherical), each mirror has to be MADE to within 25 nm surface accuracy and smoothness, but also ALIGNED to that tolerance. Just imagine trying to align two independent pieces of glass 30m apart so that their faces were parallel to an ideal paraboloid within 25nm! That’s better than 1ppb accuracy.
Yes, the Hubble is based on a spy satellite. But I have been told, that the interferometric measuring tool “acquired” some splinter of paint or dust or a tiny chip of material on a critical place of it’s optics during storage. It was indeed out of use for several years at that time, much longer than before between the regular production of the satellites, as this type of spy satellite was not used or produced any more.
This chip or splinter disturbed the laser, bent or diffracted the light while measuring and checking the mirror during it’s production, resulting in the error.
Those folks must be a lot more patient than I am. I think after about 4 days I would have said “Ehh, good enough.”
Isn’t it cute that most people still think glass flows at room temperature?
We wouldn’t spend three years polishing one of these mirrors if it was going to sag into a puddle over the next ten years.
I find that particular persistent myth highly irritating especially when people try to hang onto it for dear life. “But but but.. cathedral glass…”, “But but but… no crystaline structure..” or “But but but, no fixed melting point….”.
The one good thing about it, is how it reminds me on how many things I can be mistaken and oblivious about it.
“No fixed melting temperature”? Do these people think steel flows at room temperature because it gradually gets softer as you heat it up?
How come you begin thinking about steel when hearing about glass? Not the same kinds of material AFAIK.
To be fair, it’s been many decades since the glass making process started using molten tin to make smooth and wrinkle-free glass it’s easy to forget old glass did come from factory looking like a 2 year old cantaloupe melon.
Half of the windows in my house are old wrinkled glass (that’s how old my house is!) and I get asked often about it “melting” over decades.
A slogan from one of the best mechanical engineers I know: “everything moves”. It is all a question of how much and the time involved.
I seem to recall an article where the authors had calculated approximately 10^15 years for room-temperature glass to flow measurably. Of course glass on Earth will flow much faster than that; In only ~10^10 years the sun will swell into a red giant and engulf the earth, at which time all the glass on the planet (as well as all rocks, and even iron beams in large buildings in NYC) will flow… ;-)
In the 80ies we still earned this in school.
“learned”
“The telescope will require seven total mirrors”…
Eh?
“The telescope will require a total of seven mirrors”
Much better. Thank you Aziz.
Seven mirrors total. :-P
But he’s right though, you want them to be totally mirrors, if they were half mirror, half brontosaurus they wouldn’t work right.
Surely it needs a total of at least eight mirrors – there’s the secondary at the focus-ish point to redirect light down through the dish, and maybe a few more kicking about.
You definitely want each one to be totally mirror though.
Seven primary mirrors. A number 8 maybe as a “spare”. We aren’t counting secondary or tertiary mirrors, just the big ones.
So after light reflects off the 7 mirrors, it all goes to camera, no further reflection?
Should be 7 primary mirror, the secondary will reflect the light back to the sensor array through a hole in the center primary mirror.
You would probably be surprised to hear that an 8051 (actually an 8031) controls the mirror oven rotation.
at least it’s not an arduino
.must. .resist. .urge. .to. .mention. .555….
MPC555?
No, I would have totally expected some new and trendy part that has yet undiscovered quirks, potentially weird bugs, and unknown durability.
Makes a lot of sense. You don’t want to use something that’s been around forever and you know all the quirks. As long as it gets the job done.
And then a few years later: “shit this part is toasted and nobody make it anymore”
I remember a few years back in a major semiconductor company, someone was looking for an PC motherboard with ISA support, in order to replace one in a special tester…
We actually still use 8051 derivatives at work.
Actually a TIL moment for me that they use borosilicate glass for the mirror. Most other telescope used some sort of ceramic-glass for their lower thermal expansion.
I want one of those glass chunks!
borosilicate is supposed to be low thermal expansion no?
They were doing some work the other year with nanoparticles in relation to borosilicate so not sure if it got an upgrade if ceramic-glass was better before.
low expansion yes, but zerodur is almost 0. So hard to polish and costly that it drives price very high.
I remember bumping into the website for the first mirror of the Large Binocular Telescope (LBT). That was a long process. And even after the mirrors were complete, they still had to finish the observatory platform and the enclosure. Building an optical telescope on the ground is as close as we’ll come to building Gothic cathedrals in our lifetimes. And the Magellan is for now the biggest one in process. LBT only had to move the mirrors between 2 points in Arizona. Magellan has to ship ALL those mirrors to South America. It’s going to be a long, long time for first light on the fully finished, fully assembled Giant Magellan Telescope.
Think of the slip rings needed to transfer power for the heaters to the spinning oven. If using a spinning oven and a honeycomb structure on the back had been though of in the early 1900’s, they’d have been making larger mirrors back then. Even spinning the mold for the size of solid glass mirrors made then would have drastically reduced grinding and polishing time.
The heaters probably don’t need to spin.
Radiative heat transfer? But I could imagine, that during the heat up no rotation is necessary. Heat it up to the right consistency and then start spinning until it has cooled sufficiently. During the cool down not so much power is needed, if heating has to be continued at all.
Take a look at Mr. Preston’s book, “First Light”. In it he describes the process behind how the telescope mirror used by Palomar Mountain’s observatory was made, and by Corning glass works. As it happens they made two. The first blank suffered from a mold flaw. The second one turned out perfect and is still in use. It is made of the same stuff, Pyrex.
I watched #2 or 3 being polished. It is amazing what this huge machine can do when coupled with laser interferometry. This to to create a mathematically defined curve on an optical surface. The smallest mirror polishing pad is only a few inches wide vs a 30+ foot wide machine.
Then think about just how big the telescope will be when there will be 7 of them.
I found this chart on wikipedia to be very informative/impressive: https://en.wikipedia.org/wiki/Giant_Magellan_Telescope#/media/File:Comparison_optical_telescope_primary_mirrors.svg
Especially when you realize there were actually plans to BUILD the Overwhelmingly Large Telescope and have now decided on the “more manageable” Extremely Large Telescope with a 39 meter diameter segmented primary mirror. I’d love to at some point visit one of these giant observatories, just to marvel at the stupendous amount of engineering that goes into one of these giant machines.