Artificial Sunlight (official topic)

@smjedison Nice work! Are you going to print it and see how it works? I'm guessing it would be possible to use a single emitter to keep it simple (or smartphone light) for testing purposes.

@diyperks that's the plan... I think I'll need to grab some things first though 🙂

@gavindiy That's a great writeup! I was just thinking, would it be possible to use two mirrors that rotate while being parallel? That way the light source wouldn't need to move (which is definitely complicated). Not sure how that would affect the light distribution, but in theory the light intensity could compensate...

It... worked? It's not as clean as I was hoping, but I wonder if that had to do with me not being careful when bending the mirror. Here's a shot of it projected across our kitchen, about 20ft/7m (note, my phone made it a lot bluer and sharpened up the edges, I was using a headlight):

Here's a few pictures of it:

I had some crinkles appear in the bending and gluing process, I wonder if these could be reduced by applying more distributed force when gluing it:

I 3D printed all of the frame pieces, and cut the acrylic mirrors with a band saw. For the bent mirrors, I epoxyed the back to the frame I printed. To put it all together I just used hot glue. The process could definitely use refining, but I think it's a good start!

Fascinating!! I had a go at this too and the light projection looks very similar. I'm actually kind of confused about it. In theory, if it's properly collimated, shouldn't it result in the same sized projection as the exit pupil of the mirrors? I couldn't get shadows to be uniform in size irrespective of distance, either 🙁

@diyperks those were the same issues I had. I'm not quite sure how I managed to produce both nearly collimated light and yet have so many strange shadows.

The second parabola (the bigger one) doesn't seem to have any issues, it seems to be collimating light correctly. There's something about the sandwich mirror though that I need to mess around with more. Playing around with the 2D light simulator has been insightful, the result I got from it almost exactly matches what I've seen:

What kind of mirror setup would be needed on this first parabola to keep the output light evenly distributed along the Z axis? or does this setup have a "focal point" of sorts that we'd need to account for?

EDIT: one other consideration is that the light source in the simulator will be coming from multiple different places on the X axis, since the parabola will be bouncing light from multiple different distances from the opening. Perhaps the distance from the opening is also important?

Here's an idea: what if you start with a small 1x1" beam of collimated light? Step one, use an linear spherical reflector (or lens) to spread the light out, but only in one direction. Step 2, collimate that light. Step 3, spread out the light with a spherical reflector, but only in the other direction. Next, collimate that light, and presto, out comes a much bigger beam! It's pretty much using a combination of spherical and parabolic mirrors to scale up the light, one dimension at a time. Not sure if it would be any more compact, I'll need to mess around with it first...

EDIT: add picture example

I am almost done with my first prototype sunlamp, which uses an array of 100 LEDs with 3° lenses with the printer sheets mentioned above in front. I really don't have space for anything very deep, so I thought this was my best shot. Initial testing seems okay - it's bright and has a decent beam.

The main issue is that when looking directly at the window (from the side) the individual lenses are visible. So I'm still looking for a better solution. Smaller lenses would be good; mine are 37mm; the smallest I've seen are 22mm. Square lenses would also be good, if that's possible.

One possibility I came across is this report on Pixel LED headlamps: https://mdpi-res.com/d_attachment/applsci/applsci-11-03368/article_deploy/applsci-11-03368.pdf

Somehow each pixel lights up a 0.25° degree beam. For headlamps, one wants the total FOV to be pretty wide, but if they can control the beam of each pixel so well, I assume they could make them directed forward.

This technology looks close to ideal for a sunlamp. Does anyone know more about this? Is it possible to buy these or something similar?

not sure if anyone's seen this, but this went on indie gogo a few months ago and seems to work great.

Hello everybody,

this is Carsten from Germany and since i watched that youtube video about the artificial sunlight I am totally hooked. I ordered some material to make my own try on this. I want to go with an array of fresnel lenses and many LEDs.

I ordered a square fresnel lens with 1.96"x1.96" and a very short focal length of only 0.39". I chose to put the LED very near to the lense to use almost the complete light angle and to minimize loss of light. Hopefully the angle of divergence in front of the lens won't be too big. If it is, I will have to try a longer focal length. But if this works, I could achieve a very short overall depth.

My plan is to produce copies and arrays of lenses with silicone and resin. The refractive index of the ordered fresnel lens and resin is very similar, around 1.5.
For the scattering of blue light I might order the waterproof inkjet film, when I checked everything else works as intended. I may put it some inches away from the lenses to hide the visibility of single lenses.

I will make updates here and provide photos of this project as I proceed.

Thank you all for sharing all your knowledge!

@carsten Looking forward to watching your DIY Project

Hey what happened to the second part of the video?! There used to be a section on how to build a smaller version with a fresnel lens and some blue vinyl. The product links are still listed below the video but the part on how to build it is gone.

Have had the supplies on a shelf for a while now, finally went to build it and the instructions were gone 😝

Unfortunately I made a mistake in that part, so I took it out with the intention of revisiting it in a future vid (which I'm presently working on coincidentally)

@diyperks, ah that makes more sense now, thanks for clarifying, was starting to doubt my memory. I’m halfway into the project from what I recall. What was the mistake that needs correction? Need a “faux sun” for a small filming project and was hoping this would do it.

…also why do I have a large roll of black felt tape? Was that to finish the outside of the box? 😆

I'll be busy for the next while, so I don't think I'll be able to build my new prototype. I figured I should share the design I got so far though, so here's the scaffolding for the design (the only thing missing is a 45° mirror reflecting the output from the first parabola on the second spherical reflector, as well as a light collimator).

Is there any reason why you left a link for a blue reflective vinyl in the YouTube description? Is this to replace the standard reflective material and scattering material to provide a blue quality to the light?

I'm a little late to the game. This thread has been very enlightening and I've got some milky white inkjet sheet on order.

 If I backtrack a bit to the earlier TiO2 discussions, I was thinking that a readily available source would be things around the home that use the. TiO2 for it's white colour, such as sunscreen or white household paint.  Anyone ever did anything with these to try and make a paint?

I tried a little bit of sunscreen in a disposable cup and added water, and it seemed to scatter the light, but I couldn't say by how much. This could work if we found a readily available product around the world (eg banana boat sunscreen?) that just happened to have the correct type of tio2 nanoparticle size and was water soluble

@smjedison 

I think a problem lies there that we are not dealing with point light sources. I drew a sketch of the problem to make it easier to understand. If our parabolic mirror has its focal point at point p1 then the parabolic mirror will successfully collimate the light from p1. The problem is that we of course don't have point light sources. For simplicity it is okay to assume this anyway if our parabolic mirror is large and thus its curvature small compared the width W of our emitting light source. Then moving the point light source over our LED width will only introduce a very small angle beta as drawn in the sketch. Here Beta is proportional to the curvature of the parabolic mirror and thus ultimately also somewhat anti proportional to the size of the parabolic mirror. Furthermore it is trivial that W is proportional to Beta too.

This error is negligible in the video from DIY Perks as the satellite dish is huge and the size of the emitting light source W is small. But if we use a reflector then W is the front diameter of the reflector. Thus using a reflector should increase W and thus beta. Furthermore scaling down the parabolic mirror will also increase beta. Just from the images it seems to be a particular big problem for the parabolic mirror 3 in your sketch as the light that hits there comes from those two sandwiching mirrors that one can see in your images of the constructed prototype so W which is the distance between these two sandwiching mirrors is quite big compared to the parabolic mirror 3 which is rather small.

Hello! I'm so glad to have found this thread, I've been working on this for a while too, here's what I tried for the Rayleigh scattering part:

* Fumed silica : doesn't work.

* Silicon nanoparticules solution: does work, but I'm not messing with the risk of silicosis. I will link the source if asked, but Jesus Christ guys don't do it.

* Glass micro-bubbles : readily available, doesn't work. Also bad for your health.

* TiO2 : works! Kinda. Since I didn't know about hypersonic machines, I tried different solutions, such as smearing specific sunscreens on acrylate with a lot of hand pressure then locking it with lacker. Kinda works, but not really.

* Non waterproof printer sheets. Works. Posted about it on reddit a while ago, so that might be how it got here, otherwise it's just parallel discovery. Tried transferring or collecting the powdery film on something else, didn't work.

* Best results yet: optical aerogel. BuyAerogel.com offers different types. It's expensive, but looks incredible. The granule type has to be contained though. Doesn't work if placed before collimation of the light, which is too bad, because containment would be easier.

* Sulfur nanoparticules. Works and easy to make, but the Tyndall effect is reduced by the piss-yellow color of the sulfur. Smells like egg farts.

* Copper nanoparticules: doesn't work.

* Polished HDPE: need to try more. I found bottles of gogurt that have extremely high Tyndall scattering for some reason, but I need to see if they can be melted/flattened then polished for better translucence.

* Pearly/milky nail polish : some nail polish brands have absolutely incredible blue dispersion. But nail polish is hard to source in large quantities.

* Cheating with phosphorescent blue paint: i thought this could work, just add a thin layer of glowing blue paint to the fresnel lense. However the blue light is a single wavelength, and pretty ugly. Also requires lamps with unsafe uv output to activate the paint.

* TI deposition with a magnetron. Didn't actually try that one, but found an old patent that mentions it, so I presume it works. If we could find a specialist in optical traitements of glass it would be so nice

For collimating light :

* Obvious one is light boxes with a bunch of cheap fresnel lense, but I was never satisfied with the result.

* Best result yet: truck headlights. You can get them at your local dump, the replacement lamps exist in standardized mounting formats for 5 to 50 bucks, they take standard 12v, and some of them have active cooling. Only problem, there light is not parallel, but about 10° divergent.

@blalbu123 thanks for pointing out the issue with collimating non-point sources, that'll be really important to remember going forward with my next iteration! I ended up abandoning that first design (with the light sandwiched between two mirrors), it had too many issues, especially after trying it out in a light simulator. My current idea is to collimate light in a 1-2" square, and use a combo of spherical and parabolic mirrors to amplify that collimated light into a much larger space (that way the X and Z directions can be amplified one at a time, without dispersion on the other axis). Collimating the light into a 1-2" will be even trickier though, because like you said we're not dealing with a point source, but a plane.

@fhr just want to say that's a crazy impressive write-up! It's a fantastic summary of all the random parts that have (and haven't been) discussed. I particularly enjoy your description of sulfur, lol. Also, car headlights is an intriguing direction to pursue (need to be very bright and collimated).

EDIT: A bit of a stupid idea, but could you put the aerogel particles inside of a laminating sheet to suspend them?

@smjedison the aerogel is a bit annoying to work with, because it's so light and clings to stuff, but also gets blown by drafts. Think polystyrene foam beads.

It could just be put into a transparent box, and be a slightly simpler version of the water version, or crushed more and mixed with epoxy maybe.

What I've tried was putting it into the car headlight. It makes for a pretty lamp, but because the diffraction happens before the collimation, it's just a very nice diffuse lamp.

Anyone has experience with the dirt-cheap led lense-reflector combos? They probably have some pretty intense artifacts, but they're very compact

https://a.aliexpress.com/_EGSxUrt

I found this object which exhibits Tyndall scattering. It's sold in tourist shops as "opalite" or "moonstone". This is a synthetic gemstone made from glass. Keep in mind that there are actual precious stones also called by these names.

@chroma99 another name for Tyndall scattering-like effects is opalescence 🙂

Fire opals have the effect closest to what we're looking for, but I've never seen synthetic opals sold in sheets, and usually they're optimized to be more colorful and less transparent

That said, since they're spheres, they're already lenses?

They're cheap though € 1,31 37%OFF | Natural Stone Opal Beads Opalite 4/6/8/10/12MM Fit Diy Make Up Charms Beading Beads For Jewelry Making Accessories Free Shipping
https://a.aliexpress.com/_EJCUq4F

@fhr Thanks for sharing your results! A few responses:

1. How did you get the aerogel to work? I bought aerogel particles (also from BuyAerogel) and when I suspended them in epoxy they were far too opaque.

2. It's the waterproof printer sheets that work, not the non-waterproof ones.

3. I've tried a bunch lenses like the one you linked from alibaba (ilenstech) with 3-5° beam angles. They are not cheap though - it's like 50¢ / each, while the LED + driver are only about 10¢. A few give pretty good results without artifacts; the problem is that they are round so there is dead space in between them when looking directly at the window.

What I'm trying to do now is find cheaper and smaller diameter options to reduce the dead space.

@akvadrako I'm currently using this scheme as well, and I think I'll be able to make a sample around July.

@akvadrako for the aerogel, I think the best solution is to leave them between two acrylate sheets, and avoid any epoxy. The surface effects would be too hard to handle. Maybe with a vacuum pump?

For the printer sheets, I don't actually know which one I had, but the powdery film on them was very sensitive to water or touch, so I assumed non-waterproof.

For the lenses, it might be possible to buy them in bigger bulk. I'll look into it.

Hi there

I'm about to build an apartment in an old barn with a saddle roof, where the room height will be 4.20m in the centre of the room. My plan is to create an artificial sun using a welded steel frame, onto which I would attach fiber-reinforced plaster boards (Fermacell Firepanel A1) that can be plastered in the end. A diameter 1800mm satellite dish would reflect the 500W 5600K U-home LED light, possibly through a filter to raise the CCT, as suggested by @Nolo.

For the fake window (1200x1200mm), I plan to use a polycarbonate sheet and a Rayleigh scattering panel that can be purchased on 1688.
While it's complicated to use 1688 yourself, some Chinese guys offer to buy all the stuff you want from different suppliers to their address in China, repackage it if necessary (fee USD7), and send a slow air parcel to Europe, which surprisingly isn't that expensive.

The massive assembly could be hung from the ceiling beams using a few steel cables that go down along the wall to two manual winches.

I am concerned about heat, as this lamp is not exactly efficient. I plan to install a thick 560mm radiator, that I already have from another project, on the outside of the box, and a ~420mm radiator inside the box to cool the air inside. The ESP32 used to control the "sun" would have a temperature sensor and shut it off at a certain value. Installing a fuse that melts at 95°C on the water block would also make sense.

What do you think? Is this project worth pursuing? What do you think about heat management?

Edit: formatting

I’m so glad to find that there’s been more discussion on this project. I started on my own version of the artificial sun lamp back in 2021, it was actually my first foray into electronics of any kind! I got sidetracked by learning about arduinos and going down rabbit holes of different features I wanted to add, but eventually I came back around and got the lamp part into working order. I never solved the blue sky effect though, so I ended up just bringing the light inside from my workshop and it’s been adding some nice “daylight” to my home office for the past year.

I’m really excited that a cheap solution has been found for our Tyndall effect using those screen printing sheets! I’ll definitely be ordering some for myself soon to experiment with.

With this project on my mind, I just bought a light meter last week to see what kind of light output I was getting from my lamp. Was pretty shocked to discover that it measures only a fraction of the lux I measure outside in the sun! The light seems very bright, but I’m in a basement suite and the office faces north, so I guess it’s the lack of comparison. Someone upthread mentioned that it wa a probably unnecessary to aim for the 120k–140k lux of bright sunlight, and I think they’re probably right.

It’s not much to look at, but here’s a couple photos of my current setup. I took these at night: