Artificial Sunlight (official topic)

As a seattle resident I am 100% behind this project. I just started off the build with a small satellite dish that my neighbor is getting rid off and tried two ways to make it reflective, using the reflective mirror vinyl tapes and the adhesive disco-ball mirrors.

The disco-ball mirrors are surprisingly easy to install, conformed to the curve and are collimating the light way better than I expected. Here is a picture with both installed half-and-half.

Another thing I noticed is that the LNB for the satellite dish is almost solid aluminium and very conductive. I will try to use it for both mounting the LED and attach the heatsink to see if I can get the LED in the perfect spot.

@nolo Hello Nolo! I live in Northern Sweden and I've chosen this project for my master thesis (except I'm in college, if that makes sense?). In two weeks I'm visiting the company SSAB in Luleå to hopefully lay my hands on a sonification device. Do you happen to have any titanium dioxide left and willing to send some? That would be a relief!

I'm very excited for this project since the sun doesn't rise for almost three months here and seasonal depression is a big thing. Also at my school everyone comes from the southern parts of the country which makes it even worse. If this works, I really believe it could make an impact here.

Hi Alex, I have sent you a private message.

@nolo Hello, I read the reply. Unfortunately I can't reply since I haven't made five posts yet. I'm working on it

Has anyone thought of microwaving the nano particles to disperse them?

@alex981 Not yet, but I don't think this would work.

Is the Tyndall effect affected more by the density of the nanoparticles in the epoxy sheet or the amount per surface area unit? Say I make the epoxy sheet 0.5 cm instead of 1, should the total amount of nanoparticles still remain the same given that the surface area is still the same or should there only be half as much?

If it only has to do with amount per area unit, a more versatile formula for the nano particles would be 1,3 grams/square meter instead of 0,01% of the total weight, which has been discussed earlier.

Old formula: Surface area in cm2 * 1,3 * 0,01% = weight of particles, in grams

1,3 g/cm2 is a common density of epoxy resin.

Correct me if I'm wrong, this is simply a thought I had.

@alex981 I think, that the amount of particles should be still the same, regardless of the thickness.

If you would use less, then the scattering would become weaker till only very tiny amounts of blue wavelengthes are being scattered. It also depends on the size of the partcles. I read somewhere, that the nanoparticles in our sky are below 1 nanometer, but as the sky has a very long path, it can result in the same effect as a thin epoxy sheet with larger particles i.e. 20 nm or bigger.

The amount and size of particles are therefore the most important factors. Also there is a maximum size of particles. I read that the particles or particle clusters should not exceed around 150 nm, otherwise the blue rayleigh scattering would turn in to "Mie" scattering, which clouds for example do, as the particle size and density is too high and sunlight gets reflected in white color (all wavelengthes of the light source are being scattered equally) instead of blue and some green.

This is the reason why we need to work with an ultrasonic probe device, which can seperate clusters of nanoparticles into reasonable small sizes, so that rayleigh scattering can take place.

If I did write something wrong, someone may correct me, this is all what I gathered from reading and some experimenting.

https://m.youtube.com/watch?v=mfAGivG9Koc

Instead of buying lots of lenses, apparently it's pretty easy to duplicate lenses with silicone and epoxy.

If I ever get the time, I'll try to duplicate my large fresnel lense with ti-infused epoxy. It would be an all-in-one solution to the problem.

I overlooked the fact that employing an achromatic lens could potentially eliminate all chromatic aberrations, adding a glassy touch to the optical perfection.

As for the projector, its construction is considerably simpler compared to a video projector since its primary task is to gracefully transmit the LED COB's radiance through a glass lens onto the wall. While I lack firsthand experience in crafting one, I believe it to be a feasible endeavor. Nevertheless, the initial step involves fabricating a PMMA sheet devoid of any bubbles, resembling a pristine piece of Rayleigh scattering glass.

Does anyone have the scoop on where I can locate the video or get the scoop on whipping up a simplified rendition of the artificial sun project? You know, the one with a box that's as enticing as a fresh batch of sjokoladebiter cookies, sporting a blue interior and a dazzling Fresnel lens. I heard whispers of its existence in the initial YouTube video release.

@lilon the simplest version is a white box, a fresnels lense, one or more waterproof printing sheet and a lightbulb with a neutral color temperature.

Regarding the nano particles, how important is it that it's titanium dioxide specifically? I found some ferrofluid I figured I'd give a chance, since it is made of iron nano particles.

This could be a shot in the dark since ferrofluid only contains about 5% nano particles, meaning we have to take into account how to handle the rest of the components. BUT, to prevent clumping caused by the magnetic force of the iron, these particles are thoroughly coated with a surfactant, which is a dispersion agent. This could mean removing the need for a ultrasonic dispersion device, completely changing the game.

Do any of you guys have any input on this before I get started with the epoxy resin?

@alex189 the titanium oxide is important, since it's the cheapest material with high refractive index and high transparency. It's easily available since it's used as a food dye, paint color and sunscreen active component. It's also not affected by UV light. I doubt iron oxide will display the same effects.

It's a specific crystal form too that is the most useful, "rutile" I believe.

The dispersant/surfactant could actually help, since it prevents the clumping that ruins the opacity.

All the challenge is in having a high enough Tyndall scattering while not losing too much translucency.

@alex189 I also tried sulfur nanoparticules and copper nanoparticles, to very limited effects even in water. In any solid form it would perform worse, because of the lower difference in refractive index.

@alex189

I achieved this effect by mixing colloidal silica with epoxy. There was no need for sonification. Just straight forward mixing.

Hey everyone, I am looking for a bit of guidance as LEDs and power supplies are 2 new topics for me.  I am looking to make my artificial sun as bright as the one featured in the video (aka 500W).

This is the LED I am looking at (the 500W one). They have this listed as their recommended power supply: https://www.ebay.com/itm/183700735555

I ideally would like to use the same one Matt uses in the video due to the water resistance but neither that one or the one recommended above list having a high enough voltage for the LED I have above.

Am I missing something in terms of the requirements for the LED driver? Or would there be a better 500W LED with a lower voltage?

Thanks as I really appreciate any guidance as this project is very important to me.

Hello!

I am in process of building my own and I have some questions.

First of all, on the links of the YouTube video I didn't find anything about the led driver? From his other videos I assume he's using constant current driver as you should, but where do I find one with capacity this high?

My plan is to use one or two 300W leds because then I could on days of prohibitely expensive electricity (which happens especially in the darkest months of the year when this is most needed) run it, or one of them, from an 48-52V ebike battery, so 60V/5A per led is the driver I'm after.

Dual 300W also helps with cooling, it's easier to actually keep them under 60C with a regular CPU water block than a single 500W one, which helps with longevity.

Thanks!

Hello everybody!

First of all, thank you Matt, and everybody else for sharing your wonderful projects and doing so much research!

I've been obsessed with recreating a fake sun luminaire for multiple years now, especially after Matt's video. I've been lurking for a while and the experiments with the screen printing mask foils were especially exciting as it offers a great compromise on usability and complexity for somebody like me who has very limited opportunity for experimentation and building (meaning I dont think it would be appreciated if I spent a grand on a sonicator and started blasting it in our living room 🤪 )

So recently I came across a circular Fresnel lens with a shallow focal point of 120mm and decided to finally stop passively lurking and do some actual experiments. So here is me sharing my janky attempt while i wait on proper LEDs to arrive:

- The source is just an iphone flashlight 😬 but it is quite cold in comparison and offers more in the blue end of the spectrum than my Pixel phone.

- The Fresnel lens is held at the ~120mm focal distance but this is not that exact as I kept on messing with it to get close to a collimated beam but have a little more distortion to mask the exact shape and image of the phone. The shadows to remain basically constant.

-There are 3 sheets used in the photos 2 are cut to the lens size and the 3rd one on top to evaluate the optimum ammount. I used a silk screen exposure film for inkjet printers, I am not sure how water resistent it is however. I was trying to avoid spending too much and also the shipping hassles from china ao I ordered a small pack locally, based on the product photos that showed a bluish tint. Possibly there are far better foils out there. In summary the films do exhibit some ammount of Rayleigh scattering as well as quite a bit of Tyndall/Mie.

-The lens is certainly not of best quality and there are plenty of aberrations but it does offer some illusion of the "sun" at infinity.

My wish and plan is to design a bit more decorative light fixtures that would still replicate some aspects of sunlight and atmospheric interractions.

Next is trying out variable color temperature LEDs that go as low as 2400K for evening light and 6500K for daylight. Given that the foils diffuse the light to some extent as well it offers a nice bluish sky illusion. Stacking multiple circular fixtures offers interesting design possibilities.

I will attemp to post more updates as I move forward. Any tips and comments are of course very welcome!

Bestest wishes y'all 😉 

I had this led panel for a while and finally found a good use for it

Hi everyone!

I've been looking for a way to create sunlight-alike light source for my photography studio, and also for my home. For this reason, it needs to be somewhat compact and portable. Like theo above, I'm intrigued by the idea of using Fresnel-lens(es) to collimate light.

Using a large (900-1000mm) Fresnel works, but their focal length is too long, making them hard to angle. downwards from above. Because of this, I'm considering using multiple smaller Fresnel lens in an array formation. From what I gathered, this technique is used in solar cell industry, but I didn't find information of this being used in lighting. This might be simply because it's not a great idea 😆 

My plan consists of 36 individual 150x150x150mm Fresnel units, each housing a 9W point source LED. The units are grouped in 6x6 formation. Each unit has a thin-walled enclosure, with the Fresnel lens glued to the front, and a LED, cooling and wiring attached to the back. To improve LED's point source quality, a shroud could be made inside the unit to trap stray light. Beside the light it self, I plan to build a separate control unit which houses the power supply (Mean Well HLG) and the possible water cooling equipment.

I have few concerns, to which I hope you could comment on.

1. How uniform would the outcoming light be? It doesn't have to be perfect by any means, but I hope it could cast a good-looking light. The lenses are spaced within a millimetre from each other, but I think ultimately this depends on the quality of the cheap Chinese lenses.
2. Cooling. From what I gathered, 10W LED's require at least 4C°/W heatsinks. 36 x ~120g heatsinks add to quite a lot, which is why I'm considering water cooling. Downside for water cooling is the complexity it requires, but it's manageable. I'm really torn on this question.
3. I asked ChatGPT what it thinks about my plans. It thinks this is a great idea (and had some good advice as well!), but I still have my doubts. Do you think this could work?

Thanks!

-Niila

I have seen enough wrong schematics about coelux. here is the real one. there is no curved mirrors. it's just 2 mirrors. 

now where can I buy a proper scattering screen? 

this plan is for this light

https://www.aliexpress.com/item/1005005908516791.html

Hey everyone, here's my short update:

• Got mystelf some Lumitronix Tunable White Leds and CC drivers. Got to unpack my Arduino clone kit to try sending a PWM control signal to the drivers as they support pwm dimming. I am absolutely speechless with ChatGPT's help on this as it provided me a working code on the first try! (I am extra dumb when it comes to code, this thing is life changing)
  

  

• Mounted everything on a ryzen heatsink in a terribly janky manner and now I have a dimmable source that also changes color temperature. Future plans are to get an IOT Arduino Nano board and set this thing up to be wireless as now I see there is a realistic chance for me to get the software running.
• Once expectations were tempered, I am rather happy with the outcome of the light, the bluish "sky" glow is there, the light gets collimated to a reasonable degree and I can get away with being a bit out of focus with the lens as it gives a nicer light spot and if not overdone, can still give paralel-ish light.
  

  

  

• My next step is making a proper enclosure. For the first prototype I am going with a sphere with, honestly, a low design effort. The whole thing is multi-part and importantly gives some wiggle room in focus adjustement. There is a "thread" between the lens holder and the led-heatsink assembly so the precise distance can be dialed in so im not just projecting a 3x4 cob led matrix image on the wall. The heatsink I'll be using is just some cheapo cpu cooler for some old Intel socket. On the back of the led-heatsink assembly there a mount holes to use the cooler mounting screws to attach it to the assembly and hold the led in place.
  

  

  

• Printing this out on my ol' reliable Prusa is quite a slog as I must split the parts into 4 segments each for them to fit. Gluing, sanding, smoothing and painting is honestly not something I'm looking forward to!
  

  

  

  

• A quick parts list:
  - 1400 mA Tunable white led with separate power to different color leds (12W total)
  - 700 mA CC driver with pwm dimming x2- one per color
  - 60w 12V dc wall wart power supply ( should be enough to drive a set of 3 lamps)
  - Old cpu cooler
  - Arduino clone and a bunch of spare parts to use some 10k pots for PWM output
  - 300mm diamater fresnel lens with a folca length of 120mm
  - (waterproof?) Inkjet printing foil for silkskreen printing masks

Still a prettly long way to go but I'm pleased with how this is turing out. Having decided on the goal of a "sun lamp" and not a full-on skylight simulator lets me compromise on some accuracy of the physics simulated and so I'm feeling less demotivated when the cheap and easy solutions are not giving the exact perfect impression of sun and sky. For the last two weeks I've been running this cludge of wires, copper, foamboard and tape next to my desk when working from home and honestly the effects of sunny day are certainly there to some extent.

Will keep you all posted as I make further progress!

@niila

-I am not entirely sure that the light uniformity is that great from the fresnel lenses. They vary in quality widely. In my case I get more uniform output by defocusing the lens a bit but then I'm sacrificing paralel output more and more. But your array idea seems really interesting! But that especially could be really tricky in regards to uniformity. However, fresnel lenses seem to be the most accessible options for many of us.

-Cooling is certainly a problem but maybe in a 36 array 10w are a bit mutch? If you went with ~5w maybe you could get away with lighter cooling. I would not bother with water cooling. IMO in an array it will be even heavier that a bunch of heatsinks. Think about all those chonky waterblocks, hoses, clamps, etc. So many failure points! And don't forget the drivers can also get toasty!

-In my case I believe my heatsinks are a massive overkill for my 12W leds but only with a fan. My test setup with the Ryzen heatsink was getting really hot but the thing is designed to run with a fan, and for now I did not have it plugged in. Perhaps one option would be to use smaller heatsinks and design an airflow path over groups of them where a single large fan could feed air to multiple heatsinks at the same time?

Cheers!

-Theo

Hey @theo your light looks very cool and kinda retro! Looking forward to see how it performs! I think a "sun lamp" is a perfectly reasonable middle ground. If this works out, you can always build other lamps to increase the overall light output.

Thank you for your comments, I've been thinking about them a lot.

Posted by: @theo

↑

-I am not entirely sure that the light uniformity is that great from the fresnel lenses. They vary in quality widely. In my case I get more uniform output by defocusing the lens a bit but then I'm sacrificing paralel output more and more. But your array idea seems really interesting! But that especially could be really tricky in regards to uniformity. However, fresnel lenses seem to be the most accessible options for many of us.

Yes I'm bit worried, but I tested out some rectangular fresnel lenses I had laying around, and the uniformity was actually pretty OK. Only the very edge produced aberration, and on the middle there was a slight hotspot. Having a gap between lenses might not be such a problem, I actually like the classic window panes -look. Further diffusion also helps.

Having an adjustable focal distance could be really handy feature, thanks for the idea!

Posted by: @theo

↑

-Cooling is certainly a problem but maybe in a 36 array 10w are a bit mutch? If you went with ~5w maybe you could get away with lighter cooling. I would not bother with water cooling. IMO in an array it will be even heavier that a bunch of heatsinks. Think about all those chonky waterblocks, hoses, clamps, etc. So many failure points! And don't forget the drivers can also get toasty!

-In my case I believe my heatsinks are a massive overkill for my 12W leds but only with a fan. My test setup with the Ryzen heatsink was getting really hot but the thing is designed to run with a fan, and for now I did not have it plugged in. Perhaps one option would be to use smaller heatsinks and design an airflow path over groups of them where a single large fan could feed air to multiple heatsinks at the same time?

Yes indeed 9W is plenty and then some, but I don't have to run them full power. I've been contemplating all kinds of heatsink configurations, but the sheer number of heatsinks required and the goal of having a reasonably portable light is a difficult combination. Your idea of having a single fan for multiple smaller heatsinks is very compelling! I could, for example, divide units to 3x3 groups, and utilize a single cooling fan in the middle to pull air through surrounding heatsinks.

Your point about water cooling is a valid one. I'm slowly drifting away from that approach, but the the greater cooling capacity combined with reasonable weight still attracts me. But I have to admit I have a tendency to over-complicate things I build. 😆 

Thanks again, and keep us posted about your project!

-Niila

@bohnje Hi bohnje, we have another kind of AC110/220 to DC0-220V 0-70A 1000W Adjustable Knob Switching Power Supply for our 100W, 200W, 300W and 500W COB LED, here is the link:
https://www.ebay.com/itm/185772867519

and we have another kind of 2700K+5600K Double Color 300W+300W COB LED, here is the link: https://www.ebay.com/itm/175651507543

hope you like our led products.

Hey there, did not have too much time this week so only a small picture update to show progress.

Due to all the camera magic it's not really visible, but sadly the white cone background really diminishes the blueness. Anything blue that gets scattered gets overwhelmed by the general diffusion of the plastic.
Cooling wise the smaller cheapo heatsink does amazingly well and does not even get warm running at 7V instead of 12V even now in open air its basically inaudible.

My next prototyping step is going to be trying a small condenser lens over the LED to try and control the cone of light more towards the fresnel lens. If that works, and most of light gets directed towards the lens I will try painting the cone black or dark grey and see if the diffused blue light becomes more visible. With the light cone controlled, perhaps the brightness loss might not be an issue. Looking at, admittedly scarce, pictures of the disassmebled chinese skylights I noticed that the insides are black. Not the most important point but I don't like the idea of wasting lumens by absorbing them 🤓 

Overall, the light is quite pleasant but I wish to find the precise focus for sharper shadows. But to do that, I still need to print the outer threaded lens mount and that print is predicted to be 36h long 😰 

I can only wish the blue color would be as vivid as it is in the video! Not even close in real life 🙁

Me and my wife recently visited an artist friend of ours and she had these amazing samples of handmade glass. Now I'm getting all these distracting ideas about using something like this as an outer layer to protect the inkjet film 😵 :

Posted by: @niila

↑

Yes I'm bit worried, but I tested out some rectangular fresnel lenses I had laying around, and the uniformity was actually pretty OK. Only the very edge produced aberration, and on the middle there was a slight hotspot. Having a gap between lenses might not be such a problem, I actually like the classic window panes -look. Further diffusion also helps.

Having an adjustable focal distance could be really handy feature, thanks for the idea!

@niila Yes adjustable focus is really handy when setting up the lenses! Maybe playing around with masking the edges of the lenses could be an option? Only a few milimeters but maybe these "apertures" could help control the overlap?

Cheers y'all,

Theo

@zero An update regarding the 3M film. It seems you can now purchase it on some online stores !

It is called "radiant mirror film", it is far from cheap from what I see but the extra efficiency (99% reflective) is important for smaller and more compact designs 

If links work, here is an example: https://www.sirvisual.com/lighting-films/3m-specular-film-df2000ma!0-4269

Edit: I also found it under the name of 3M™ Specular Film DF2000MA

Hey yall, here is my version using a 60 cm satellite dish covered in acrylic vinyl, a watercooled high CRI 100W LED and some of that waterproof screenprinting paper.

I decided not to try to hide it;but make it look like a movie projector. It's a 60cm cube of plywood atop a staked walnut/cherry/maple tripod. Window is 50*50cm area. The middle of the cube stands around eye level, and is hinged on the tripod with a wooden locking mechanism for vertical alignment of the beam.

Very pleased with the results, thank you all for your research and input! It considerably changes the mood in the room where we live and work in northern Europe, the nights are long these days...

Happy to share more details or pics as needed. I also ended up having to buy a whole roll (24" * 100ft https://www.ebay.de/itm/304693614142) of that special waterproof plastic sheet; but I don't need it all. Happy to send some chunks of it for a few euros + postage (within EU), just message me!

@nolo Do you happen to have any photos of your Nanlite projector operating inside / with an inkjet sheet?  Do you find that it has to be placed far away from its target due to its relatively narrow 19 degree beam?  It is very tempting to get something like this, but they are quite pricey.