For most of us, mirrors are something we buy instead of build. However, [Unnecessary Automation] wanted to craft mirrors of his own for a custom telescope build. As it turns out, producing optically-useful mirrors is not exactly easy.
For the telescope build in question, [Unnecessary Automation] needed a concave mirror. Trying to get that sort of shape with glass can be difficult. However, there’s such a thing as a “liquid mirror” where spinning fluid forms into a parabolic-like shape. Thus came the idea to spin liquid resin during curing to try and create a mirror with the right shape.
That didn’t quite work, but it inspired a more advanced setup where a spinning bowl and dense glycerine fluid was used to craft a silicone mold with a convex shape. This could then be used to produce a resin-based mirror in a relatively stationary fashion. From there, it was just necessary to plate a shiny metal layer on to the final part to create the mirror effect. Unfortunately, the end result was too messy to use as a viable telescope mirror, but we learn a lot about what didn’t work along the way.
The video is a great journey of trial and error. Sometimes, figuring out how to do something is the fun part of a project, even if you don’t always succeed. If you’ve got ideas on how to successfully spin cast a quality mirror, drop them in the comments below. We’ve seen others explore mirror making techniques before, too.
This is one of the alchemical projects for home telescope makers. I’ve seen and heard from people who have tried this over the years. From internet strangers to professors and a friend. Even had a scheme to do it myself once.
I’m glad people are still trying. I hope more people try because if someone figured out a way it would be so completely worth it.
I wonder if hydroforming a round metal disk might be a better way to go for a large mirror.
I was thinking of directly making the mirror 100% metal. But I am sure the reason it is not done is the thermal coefficient of expansion of most metals is huge. And then you start to think about heating/cooling the whole mirror to a constant temperature with a closed control feedback loop (thermocouple, a peltier device and a basic PID controller).
The problem boils down to surface roughness/precision. You cant take some sheet metal and bend it and voila mirror. This may give you a rough shape, accurate even to the millimeter or something in the 100’s of microns. But the bends will form buckles on the already very rough surface.
So the next logical step is polish. Polishing a sheet of metal down to the 10s of nanometers (or less some measures go to the Angstrom) of smoothness on a gently curved surface is mostly not feasible. Maybe with some very serious machine tools, like 100’s of thousands of dollars, but for the average person this pursuit is about hopeless.
It’s one of those things where you can see a shiny piece of metal, maybe polish a candle stick and go “wow that’s reflective”. But when you compare it too a commercial grade front surface mirror the cloudiness/diffusivity of the image is unremarkable.
I hope this isn’t discouraging. It is the devil in the details. Glass has been used for a long time for a good reason. It’s really really hard to find a material that can be shaped to this degree of smoothness.
Indeed, what’s regularly called a “mirror finish” usually is just a “shiny surface”. while telescope mirrors require absolute accuracy. I once read about telescope mirrors from a spinning disc of mercury and it was quite difficult. First you start with a bowl that is already quite accurate, then you pour in a lot of mercury and spin it to cover the whole surface, and then you suck out as much mercury as possible. This is a necessary step to reduce the effect of vibrations. There are always vibrations, and a thinner layer of mercury dampens these better.
you could take that idea of using a thin layer to avoid vibrations and apply it to the resin idea: 3D print the correct shape first (so instead of a printed dish without shape, pre-shape it to the required form. Then apply the thin layer of resin (it’s cheaper too!) apply the ubiquitous flame to pop bubbles and let it harden.
https://www.cloudynights.com/topic/403093-what-does-110-wave-mean/?p=5166379
they need to be really ‘flat’ as such, it would probably be better to mould a blank of about the right shape and then use a jig to ‘figure’ the mirror into the correct shape, it’s a lot of work but it’s the traditional method for making your own glass mirrors/lens
While I agree I don’t think you’d actually have any real problem applying the usual glass mirror grinding type methods to that metal blank to get the surface finish and shape you want if you really wanted to use metal for some reason (assuming your formed blank is remotely stiff enough to actually be a useful mirror that doesn’t warp under its own weight etc).
I can’t think why you would try to use metal like that right now when any old glass is likely much more dimensionally stable under external forces and temperature changes, likely much more resistant to the environment, and likely quicker to work with too. But no doubt there are some good reasons out there, weight perhaps?
What about using a magnetic field to manipulate a ferrofluid? Or use ferrous particles in a resin to hold the form as it dries?
For high density fluids, I thought of: motor oil, and tar.
Both are quite viscous, which may reduce the impact of variations in spinning speed.
Pitch drop experiment, but observed in Zernike aberrations
With a thermal expansion coefficient of a ballpark 150 ppm/K, I wouldn’t expect the result to be usable. Unless you manage to iron out the distortions again with adaptive optics :D
Why not just buy some PMMA stock and machine it on a lathe? Looking on ebay I see relatively cheap precise metalworking lathes available at $400.
Cheap != precise. Besides, how in the world do you intend machining a large radius-of-curvature part on a manual lathe?!?
with a printed shape to grind/polish the blank into shape? Or have a profile lasercut and let that do the primary cutting.
A lathe will not get to the level of accuracy required. It would unfortunately be many orders of magnitude away from being formidable.
The allure behind spin casting or twirling buckets of liquid metals is, maybe just maybe some material can get surface roughness down to an acceptable level. Mercury bucket mirrors work, but getting something to stay solid in the right shape is a serious challenge.
Another interesting idea, perhaps styrofoam could be used. A large block of styrofoam, then cover with a sheet of something like cling film, then heating it to melt the foam to conform with the deformation of the sheet. Perhaps filling the sheet with a hot relatively dense liquid like molten wax to form the parabolic shape.
Maybe a sheet of Perspex could be simply suspended from a hoop of metal and heated until it sags in the middle.
Unlike what others here wrote, I think this could work, at least upto some usable degree.
There is such a thing as diamond turning for PMMA, in which lenses are made on a CNC lathe. And it’s possible to get very near to a working lens without any polishing afterward.
But it all hinges on the quality of the lathe. A regular cheap chinese lathe has no chance. but the DIY lathe from Dan Gelbard can probably do it. The main spindle of his lathe has a second hand spindle from some wafer manufacturing device with air bearings, and the rest is made from granite blocks (also air bearings).
What about stretching a rubber sheet [ something really elastic like a wubble bubble ball perhaps ], over a large sealed container and then pulling out some of the air. Maybe using something like low temperature wax to cast the shape. Maybe just use a mylar space blanket and create the mirror directly. Although not so good for IR.
I thought I remembered it on here, found it.
https://hackaday.com/2016/07/26/pressure-formed-parabolic-mirror-from-a-mylar-blanket/
That was indeed a nice article. But the comment noted:
So that makes it good enough for focusing some sunlight, but nowhere near enough to build a telescope.
IIRC it was used to cast the blanks for large telescope mirrors, using a rotating liquid-glass oven. A few tons of oven and molten glass – quite formidable !!! The blanks still had to be precision ground and figured though. I wonder if a liquid tin bath could be incorporated, so the glass floats on the tin as in the float-glass process – at a guess it would still be a parabolic surface at the interface ?
Confession : I did make a parabolic surface once using polyester resin, rotating it on a record player as it set. It was parabolic as far as it went, but the surface was too textured to do much with and the whole lot was only a few inches wide. Fun though.
Glass on tin would be a neat way to make a convex blank. I suspect they are so simple to make using conventional techniques it’s just not worth the added complexity.
I, too, did the record player attempt, but using wax. Dumb, but I learned.
Spinning a pool of mercury (even with its toxic properties) to create a parabolic mirror has been experimented with (16th century ?) and is in use today.
https://www.popsci.com/science/international-liquid-mirror-telescope/
If spin casting a mirror form with something other than molten glass could be achieved then there is still the problem of achieving a surface that can be smoothed or ground and made uniform to prevent “Hubble” vision or other aberrations. What about active deforming using actuators or even heating resistors to correct the inaccuracies?
I was thinking why say ” parabolic-like shape” and made a list of why and the forces that keep it from being ideal. First, the derivation of the nature of the curve assumes an infinite flat Earth, not the real gravitation which behaves as if it converges to a point source at the center of mass of the Earth. To that error you can add the Coriolis forces, precession, nutation, and tidal forces from the Sun and Moon. Coriolis is probably the strongest though the rotation speed for a typical telescope f-ratio is quite low.
A telescope mirror is very small compared to the distortion in the gravitational field due to Earth being an effective point source. Even gravitational variations due to different densities in different locations won’t have any practical effect.
A big problem is the change in volume of the material as it solidifies, whether by cooling or curing. If it’s not uniform, the shape will be distorted. Adhesion to the container and maybe Van der Waals forces are going to play havoc with good shaping.
That spinning mercury mirror telescope was featured in a 1934 pulp scfi story “Old Faithful” by Raymond Z. Gallun
http://www.technovelgy.com/ct/content.asp?Bnum=994
I think the idea was described and tested with mercury in the early 20th century. Check “Amateur Telescope Making” by Ingalls. It didn’t work very well then, either. A good telescope mirror is accurate to at least 1/8 wavelength of sodium light, or 70-odd nanometers. Vibrations alone ruin the image.
I wonder if the surface imperfections might also be a result of slight vibrations from the motor and the mounting of the spinning container used. It might be an interesting idea to use an old record player for spinning instead, those were manufactured for a pretty smooth turning process in order to keep audio playing perfectly.
Anyone tried using UV-setting resin ? Eg as used in 3d printing … the rotating liquid would form the surface,
which is then hit with UV light and solidifies. Maybe repeated, so that the final surface is just a thin film
on top of a bulk support.
In fact just dipping something into acrylic resin monomer and letting it drain should leave a glossy surface as it polymerizes, as in glossy clear polyester lacquers. Not near optical-mirror quality I guess.
Kind of ironic, tin was used as mirrors way back (as alloy with copper, speculum metal) , but it tarnished a lot.