An absolute ton of the cost of this was down to the extremely precise manufacturing required to make it. Some of that would be reduced now that they've made it once, but we're still talking about one-off components here. It won't be an order of magnitude.
One thing I am curious about is how many spare parts were produced - in small scale high precision manufacturing like this there's often multiples of components produced, with only the highest spec components shipped out. What could we cobble together with the rejects and leftovers? And what would that give us, results wise?
My understanding from that was they were measuring it to make and test a computer model to simulate the shield in 0 g, not to get exact measurements to "it to the right size". Now its in space, they know how it performs and how accurate the model is. I definitely could be wrong.
The other huge cost drivers are testing and calibration. Even if you have all the parts to make another one, you'll spend a lot on expensive labor to put it together, tune, calibrate, and verify it.
This is actually where NASA gets a significant amount of hardware: manufactured parts from intelligence programs that were the rejects from those production lines. Hubble was a prime example of this, but there are many others.
I'm not sure he's precisely correct, but Hubble's primary mirror was changed from 3 meters to 2.4 meters, which happened to be the same size as the primary mirrors for the KH-11 KENNEN series of spy satellites. The book Power To Explore - History of Marshall Space Flight Center 1960-1990 says of the decision to change the mirror:
> "In addition, changing to a 2.4-meter mirror would lessen fabrication costs by using manufacturing technologies developed for military spy satellites."
There's quite a difference between "used infrastructure built by the military for their needs" and "manufactured parts from intelligence programs that were the rejects from those production lines".
Especially the strong implication that Hubble is made out of lower tolerance parts, since they are rejects, is a claim that need significantly more evidence than what you provided here.
Pointing a telescope at something bright and nearby is easier than something very distant and dim, so the precision requirements are obviously in the opposite direction.
> Pointing a telescope at something bright and nearby is easier than something very distant and dim, so the precision requirements are obviously in the opposite direction.
You're wrong. Between 1979 and 1998 the MMT Observatory used six repurposed spy satellite mirrors that had been donated by the NRO, manufactured for the cancelled KH-10. Furthermore in 2012 the NRO donated two unflown spy satellites to NASA, the mirrors from which may be used for the upcoming Roman Space Telescope. Spy satellite mirrors make great telescope mirrors.
> "manufactured parts from intelligence programs [...]"
As I point out above, the NRO has donated at least eight primary mirrors that were specifically manufactured for cancelled or unused spy satellites. The Hubble mirror may be another example of such, or it may have merely been fabricated by the same people.
> that were the rejects from those production lines".
That's the only dodgy part of his claim, but it's not that strange of a claim when you recall that Hubble's primary mirror infamously had a spherical aberration. I presume NASA didn't know about the aberration before they launched HST, otherwise they should have corrected for it on the ground instead of having astronauts correct it in space. But it doesn't seem completely outlandish that the NRO gave NASA mirrors they had previously deemed unsuitable for their own use.
That people accept mirrors gifted for free is not strong evidence of their superior quality, when you get something for free you can make do with something that's not really what you want.
As an example, the MMT observatory was so happy with the "cobble six telescopes together" design foisted upon them by the donation of the mirrors that within 8 years of completion they were officially[1] reporting they hoped to replace them with a single mirror by 1993. Honestly I think the MMT is the only telescope I've heard of where the main mirror was so bad they opted to replace it with a completely different design within 20 years.
Anyway, the Hubble mirror has a different focal length than the KH-11 so can't have been used as is, it would have to be reground and polished for the new curvature and so it would have been essentially just a mirror blank.
> Especially the strong implication that Hubble is made out of lower tolerance parts, since they are rejects, is a claim that need significantly more evidence than what you provided here.
Well, it did ship with a faulty mirror after all :)
Just joking, I know this was a manufacturing error specific to Hubble which kinda proves it wasn't a reused part. I think the mirror would have a slightly different focal length anyway? But maybe 400k is far enough to be "infinity" even at that scale.
Spy satellites don't have a fixed distance to their target; they have elliptical orbits and take take pictures of earth at different angles, different distances to the target. So both spy satellites and telescopes use mirrors focused to infinity.
> I know this was a manufacturing error specific to Hubble which kinda proves it wasn't a reused part.
It was ground very precisely into the wrong shape during the final stage of manufacturing. But I don't think that proves anything one way or the other. Even if the mirrors were finished in unique ways, they may have started identical earlier in manufacturing.
> It was ground very precisely into the wrong shape
It was said at the time that the aberration made the Hubble near-sighted and, indeed, was corrected with lenses. I had suspected, apparently incorrectly according to yours and others comments indicating that spy satellite mirrors are interchangeable with and used for astronomy without alteration, that the mistake was one of habit, because I figured they were ordinarily making near-sighted mirrors to specification to focus up to a few hundred miles rather than stellar or galactic distances. I am honestly am still having trouble accepting that a mirror designed for a telescope to look no further than a few hundred miles is identically focused to infinity precisely like similar telescopes that are designed to peer with a lower bound of at least millions of miles. But if you say so.
> I am honestly am still having trouble accepting that a mirror designed for a telescope to look no further than a few hundred miles is identically focused to infinity precisely like similar telescopes that are designed to peer with a lower bound of at least millions of miles.
Consider the 2.4m lens to be the base of an isosceles triangle with a height of 100km, the edge of space and much lower than the satellite's orbit. This triangle has two angles of 89.9993 degrees. If the lens was 10cm thick and you wanted to taper the edges to match this angle, you'd need to bring in the near edge one micron, or about 1/50 of a human hair.
100km or 100,000 lightyears, they're still effectively straight away (focused to infinity).
> So both spy satellites and telescopes use mirrors focused to infinity.
That's just not what focal length means in the context of mirrors, the focussing happens by moving the secondary not by re-grinding the entire primary mirror or whatever it is you are imagining.
structural didn't suggest that NASA repurposing military tech is "bad and evil". Why would it be? That's just your own bias, and one I doubt you can back up with much rational argument. Was the Gemini program "evil and bad" because the Titan II boosters used were originally developed as ICBMs? That wasn't evil, that was smart.
Anyway, he's substantially if not completely correct and it's not hard to find sources backing up the connection between the HST and KH-11 spy satellites if you bother to look it up. In the sibling comment to yours, I provide a link to a book about NASA, hosted on NASA's website a few years ago, that says Hubble's primary mirror was fabricated using technology developed for the military, that a 2.4 meter mirror was specifically chosen to make that reuse possible. The mirrors were almost certainly manufactured by the same people on the same tools. Read the wiki page for the KH-11 and you'll see that the apparently similarities run deeper than that. Dimensions and general structure of the satellites are also believed to be the the same. None of this is fringe.
I would assume at least one duplicate part, as when a friend made a component for a sattelite he had to build a second so they had an identical part on earth to study/test in case of later issues.
Barely worth to launch another one. It would be more cost effective to invest some R&D into building a telescope with a different target wavelength to get more science than if you had two of the same telescope. The engineers will also absolutely want to fix any issues they found in Webb already.
The degree to which science has advanced since the Webb project started can't be understated either. We have a fundamentally better understanding of the technology available and what we even want to look at. Much better to simply move onto the next project, of which there are currently very many.
Was going to reply along these lines. I was fortunate to live with someone who was working on the James Webb and telling me excitedly about it — back in 2006! Surely even with the various upgrades/spec changes/delays, things have moved sufficiently that whatever is started even today will be a marked upgrade.
In any event, many many areas to aim at, and relatively limited funding unfortunately.
Paul Sutter has a great astronomy and physics podcast called "Ask a Spaceman". His "Five Exciting Missions After James Webb" episode (20 min) got me really excited for the future: https://www.youtube.com/watch?v=DiYVsoxbxAI
I'd like to see a deep space version of the Gaia astrometry space telescope.
It measures the parallax shift of stars, and is basically the one reliable way of directly measuring how far away a star is from us. Unfortunately, it's at L2, and therefore has a baseline of 1 AU. Another Gaia way out at 20AU would have capacities no Earth-based telescope could ever have.
This seems like the kind of thing where two would actually be useful. Is there any benefit to making both observations at the same time? Or are the scales so great that it doesn't really matter?
You'd almost certainly want to launch several. You get one data point per half-orbit, when you're at opposite sides of the Sun. This is tolerable for the Earth, where an orbit is one year. But a full orbit out at 20AU takes eighty four years! Collecting a useful number of samples with one spacecraft would take centuries, while two spacecraft in opposition on the same orbit can measure parallax instantly.
There has been a lot of work on earth based telescopes (eg the 30 meter, giant Magellan, and some array based telescopes) that are going online in the next decade.
For the most part, yes, with adaptive optics and corrective measures being taken to deal with more satellites in orbit, ground telescopes are superior or at least comparable to space telescopes given their ability to be much larger.
Space telescopes these days are primarily being designed for observations that simply can't be done while in the atmosphere (eg the wavelengths JWST and NGR look at). The value of a space telescope in the same wavelength range as what ground based telescopes usually use would mainly benefit in terms of being able to have much longer exposures.
Yes they are very much possible, and cheaper than space too.
What starlink does is ruin part of the images, and if the thing you were interested in observing happens to be blocked by a starlink trail you're hosed: a thing literally blocked what you tried to see and you lost the nigh (because usually you get just a bit of the a night for your observation). Other things that ruins your night is clouds, so starlink effectively makes the weather at a site worse, only you find out after the night that it was all a waste.
To some extent you can plan around it, but as the mega constellations grow they'll have to avoid each other more frequently and there's no rules for how that shits coordinated, so you maybe you can know in advance that the night is wasted.
But the risk that a satellite is in an undocumented orbit by the time you try to observe will likely be very high in the future.
Have you taken a photo while someone else used a flash? The flash is also only on for a fraction of the camera exposure but you sure as hell notice when it happened and it went off close to what you wanted to depict you will just have to take a new photo.
The length of the occlusion isn't very relevant when the thing going in front is orders of magnitudes brighter than what you are trying to observe.
a leo telescope doesn't look anything like a star (mainly because it's moving too fast). the way you deal with this is by not taking hour long exposures, and instead take thousands of second long exposures. then you can composite them all together, cropping out the bits that look like satellites from each frame. it's a little annoying, but pretty easy to automate.
My understanding is ground based telescopes imaging in the same wavelengths also have to deal with distortion in the atmosphere, star link interference would be easier to filter out compared to the other stuff (which is why locations for these mega ground based telescopes are chosen with utmost care )
Disclaimer. Not a physicals or astronomer, just a enthusiastic backyard amateur astronomer who reads a lot about telescopes .
How do you figure? A fleet of space telescopes research teams could interact with through an API without much cost and zero approval, would for sure advance science by a lot. I find it weird to see a statement like this, so maybe you have something else in mind. It's a stretch to go from "building a different one would give new classes of insights" to "having more people being able to use this thing we only have 1 of a kind is barely worth it".
> A fleet of space telescopes research teams could interact with through an API without much cost and zero approval, would for sure advance science by a lot.
100 % someone would point at the sun by accident and burn all the sensors within six months if access is unrestricted and the api is powerful.
In any case easy access to data will mostly result in data lying idle on disks somewhere because people are busy doing something more interesting. Analysing data is many months of work, and you don't get more months just because there is more data.
I wonder how feasible it would be to count on Starship to succeed, so that each flight could deliver one hexagonal mirror segment, eventually culminating in a giant composite mirror
Based on how JWST folded up, the width of the hexagonal segments is already a sizable chunk of its overall launch diameter. Doing what you suggest could absolutely result in a bigger overall telescope, but the complexity would be increased vastly more than the overall telescope diameter.
You would not need to fold it at all. Instead, launch dozens of Webb-scale scopes for a fraction of the price, able to point in that many directions at once.
I don't think you understood what I meant. If you launch the segments by themselves, the width of a single segment would still be limited to the internal diameter of the launch vessel, we're not going to just bolt a naked mirror to the front of a rocket.
If you compare how the JWST was folded, the width of the individual segments was already close to the maximum allowable diameter of the launch vessel. Leaving the rest of that launch vessel empty won't get you a much bigger final mirror.
JWST folded is about 3 tiles wide, in a housing 4.7m in diameter, unfolding to 6.5m. So a JWST style fold in a Starship would be double. A one-mirror-segment-per-flight would be something like 36m, in the same arrangement. But since you would be constructing from separate sections, the diameter is theoretically limitless.
Just stack the mirrors inside Starship, and assemble them in space.
Below, an example of a 300m diameter telescope using 8-meter mirror segments assembled in-orbit
I guess in-orbit assembly would then be a whole different can of worms. But, a very useful can of worms - once it's figured out, it's like horizontal scaling - you can just keep sending up more parts.
There aren’t plans to launch an identical telescope. Perhaps the costs of manufacturing and testing are high enough that we might as well launch a different telescope with different capabilities?
The Roman Space Telescope is a wide field instrument that is now under development and slated to launch in 2026 [1]. The Astro2020 decadal survey from the National Academies also recommended “a large (~6m diameter) Infrared/Optical/Ultraviolet space telescope” to observe exoplanets [2].
I'd be keener to see some of the larger diameter rockets coming online soon be used as a housing. 9m to play with there and perhaps with NASA's amazing origami skills then that could really open the door to some huge space telescopes.
Yeah I think they'll be building LUVOIR with the tooling and software designed for the Webb, but much larger so it fits into Starship/SLS and for a wider range of wavelengths.
There's no way the cost would be minimal. I would even wager that it would be just as or more expensive to build a duplicate than to build a new design based upon what was learned. Given the decades over which the James Webb Telescope was developed, it has parts and designs in it that are, well, decades old.
And the bill of materials is unlikely to have been the primary cost factor. Extensive research, development, and testing was performed.
Honestly this is probably true sigh. the temptation is to say most of the cost of manufacturing an item is in r & d, setup, confirmation testing, etc these are mostly one time costs, the actual price of manufacturing the item is fairly trivial in comparison, this is what makes economy of scale work.
So one james web space telescope costs 10 billion and take 20 years to build. However two space telescopes cost 10 billion 100 million and take the same twenty years to build. three would be 10B 200M + 20Y etc.
But this would of only worked if you built the second one at the same time you had the setup and test facilitys prepared for building the first one, I will be generous and assume the r & d is not so space and time critical.
Personally I think the main argument for making a second is that you are putting this thing in a location where it cannot be serviced. The parable of having all your eggs in one basket comes to mind.
They have different focal lengths, so even if the glass was originally for a KH-11 it couldn't have been used in Hubble "as is", which makes it nothing more than a rough polished mirror blank.
Why would it need to be used in Hubble? We don't even know if these satellites are KH-11s. They might be some other design that is capable of fulfilling a NASA mission. The point is that Hubble is suspected to be based on a spy satellite so there is potential technology transfer. Maybe these are just useful as parts.
LUVOIR is a foldable mirror telescope, larger than JWST, UV to near IR, like Hubble. Congress has a bad test in mouth from JWST delays and overruns. Pehaps dozens of majornew discoveries will help.
Roman is next in queue with many of its parts already built and operational around 2028.
Wait for starship, make a new bigger one that doesn't have to fold like crazy fragile origami, manufacture and send 10 of them for the price of making and sending this one.
