Projector Cyanotype and Chiba python digital negative.

I've been using these, they're marked as OD 7+ from 190 to 532nm. I got them for use with a 40W 450 nm laser but they're extremely effective at blocking the 385 nm light my projector operates at.

AndrewBurns said:

I've been using these, they're marked as OD 7+ from 190 to 532nm. I got them for use with a 40W 450 nm laser but they're extremely effective at blocking the 385 nm light my projector operates at.

LG-005 | 445 nm 450 nm 455 nm 532 nm Laser Safety Glasses Green & Blue | 190 nm to 532 nm | OD 7+ LB6 | Universal Fit

The LG-005 provide excellent protection from green and blue lasers operating from 190nm through to 532nm (including 445nm, 450nm, 455nm) with an OD 7+.

www.lasersafetyglasses.com.au

Click to expand...

These are pretty good. I like that they have the transmission and OD on a chart. Its making me nervous about my yellow ones because they have no rating for higher than 360nm.

imgprojts said:

These are pretty good. I like that they have the transmission and OD on a chart. Its making me nervous about my yellow ones because they have no rating for higher than 360nm.

Click to expand...

Ah nevermind I just checked and the protection goes from 190nm to 525nm. But regardless the light seems too bright/ intense.

My second lens came in finally so I have now the two fresnels and an EL Nikkor 50. Now I have to figure out how to package these things so they fit together. I tested the EL Nikkor for significant UV drop and gladly it doesn't kill all light.


That's the fresnels together. And the Adapter for the output lens coming along.

imgprojts said:

My second lens came in finally so I have now the two fresnels and an EL Nikkor 50. Now I have to figure out how to package these things so they fit together. I tested the EL Nikkor for significant UV drop and gladly it doesn't kill all light.

View attachment 411013View attachment 411014
That's the fresnels together. And the Adapter for the output lens coming along.

Click to expand...

I use a 210mm EL-Nikkor lens for my 400W 385nm UV projector and it works just fine. I haven't quantified what it's actual UV transmission is (well, I did quickly using a UV meter and it seems better than 75% @ 385 nm) but it does work for me. I assume the 50mm would be similar. For me the first pair of fresnel lenses I got blocked significantly more UV than they were supposed to, and I had to order a second pair from a different place made with a grade of acrylic specifically designed for UV transmission.

AndrewBurns said:

I use a 210mm EL-Nikkor lens for my 400W 385nm UV projector and it works just fine. I haven't quantified what it's actual UV transmission is (well, I did quickly using a UV meter and it seems better than 75% @ 385 nm) but it does work for me. I assume the 50mm would be similar. For me the first pair of fresnel lenses I got blocked significantly more UV than they were supposed to, and I had to order a second pair from a different place made with a grade of acrylic specifically designed for UV transmission.

Click to expand...

Man, 400W! And you're using an LCD screen as negative? Won't it bleach/burn the screen?

From this quick test I don't think I even need the 50mm Fresnel, but it does make the beam more focused and more diffuse. I decided to keep the same prism because it comes with the anti reflective front coating and as opposed to a thin plate which may have a ghosting second reflection, it may work perfectly fine. I need to reduce the power by at least half to get around 20W on the DMD chip or less. Damage threshold is stated low. I finally identified the chip as 1912-7037 which appears to be a TI product without direct TI support but with an equivalent TI part number. The visible part is allowed around 20 or 30 W/cm^2 but below 395 or 410 all their chips are specified in the milliwatt region. Like 2mW/cm^2.


I bought a second garbage projector "for parts". It comes with a brand spanking new heat damaged DMD that the seller couldn't possibly know anything about. So my plan is to extract it and test long UV exposures with it just in case. It has an almost unnoticeable circle of darker pixels. I also want to see if I can see the pixel grid on the microscope.

I did a quick unscientific LED florescence measurement at the focal spot and I measured just about the same voltage I would measure at the collimation lens. So this is great so I can maximize the beam intensity as it passes thru a good 1.5" of prism glass.

As for the adapter, I need to add a helicoid to my adapter plate and then an adapter helicoid for the Nikkor. I sort of tested the magnification and was slightly disappointed. I mean they are for enlargers right? I'll probably get a 20mm or 30mm Nikkor. In the meantime (I sound like I can actually make an image), I have a vintage vivitar 20mm which I also tested for UV. Interestingly it did actually output UV light. Slightly dimmer because it is a much enlarged image. I'm gonna try to get both lenses to work on the same plate helicoid.

This is exciting. Usually this is followed by some kind of disastrous realization. In the remotely off chance all this guess work doesn't pan out, I'm going back to the original configuration and just replace the lamp with a cheap auto HID with the filter removed. That would be so much easier but someone invented the LED and here I am mesmerized by the possibilities. I just found these: 8080-SMD on aliexpress for $15.33. That's 50W of 365, 385, 395, or 405nm in a tiny chip ~12mm square. This is perfect for collimating.

Wow, 400W. I was holding the 2" 50mm Fresnel in front of the 50W LED and I could feel my fingers being in the beam. It's a weird sort of feeling because it comes with the knowledge that this is cell killing light. No funny smells like when you get a laser burn. Just that "I'm getting cancer of the finger" feeling. A 400W cob could probably cook your skin when near the source. I'm getting different eye protection too. I'm using the welding helmet for now. The image was taken on a vintage 35mm super takumar at f16 with a uv shade filter and its still saturated so you know its not something to screw around with.

imgprojts said:

That's 50W

Click to expand...

In theory. If you look at the specs they say something like 27-30V at 1-1.4A. This means 42W at the high end and I wouldn't run them that hot. I think realistically you can get 25-30W from one of those modules with proper cooling. IDK how many hours they will have in them at that level.
Btw, there are several kinds of modules you can buy presently that will yield high UV power levels in a fairly small package. The '8080' LEDs are similar to COB LEDs (different physical package); they're a 2-dimensional array of emitters inside a single physical package. Perfect collimation will be impossible, but it may/will likely be good enough for what you need.
If you're cost-sensitive you could make a table with how much power you get per buck inside an acceptably small package - make sure you include auxiliary stuff like power supply and cooling provisions into the equation.

A couple of days ago I got the lens holder and helicoid printed. I did a dry run. Instead of the DMD chip I taped a piece of white watercolor paper with some red and blue ink words. I then used a simple white light flashlight to project. Its pretty awesome that just paper with bright light will make a perfectly sharp image.

Ladies and gentlemen, my first projected UV image using a 10W, 395nm LED. The Ubuntu desktop! I will have to splurge on a 50W 10mm module, but there's at least some light at 10W and it makes the screen phosphores. Next order of business is to give it as much power as possible and make some cyanotypes. Then simulate the fans and thermocouple signals so I can remove all that junk off.

imgprojts said:

Ladies and gentlemen, my first projected UV image using a 10W, 395nm LED. The Ubuntu desktop! I will have to splurge on a 50W 10mm module, but there's at least some light at 10W and it makes the screen phosphores. Next order of business is to give it as much power as possible and make some cyanotypes. Then simulate the fans and thermocouple signals so I can remove all that junk off.

Click to expand...

Interestingly... If you actually install the 50mm lens that I was blabbering about, then the output is much brighter. So I placed an order for the 50W module at 385nm with the ~10mm source. The other 50W lamp does work brighter even without a focusing lens. Hopefully the small source plus focusing lens will get me something with a bright enough output. I can't believe I totally missed the lens.

imgprojts said:

Man, 400W! And you're using an LCD screen as negative? Won't it bleach/burn the screen?

From this quick test I don't think I even need the 50mm Fresnel, but it does make the beam more focused and more diffuse. I decided to keep the same prism because it comes with the anti reflective front coating and as opposed to a thin plate which may have a ghosting second reflection, it may work perfectly fine. I need to reduce the power by at least half to get around 20W on the DMD chip or less. Damage threshold is stated low. I finally identified the chip as 1912-7037 which appears to be a TI product without direct TI support but with an equivalent TI part number. The visible part is allowed around 20 or 30 W/cm^2 but below 395 or 410 all their chips are specified in the milliwatt region. Like 2mW/cm^2.

This is exciting. Usually this is followed by some kind of disastrous realization. In the remotely off chance all this guess work doesn't pan out, I'm going back to the original configuration and just replace the lamp with a cheap auto HID with the filter removed. That would be so much easier but someone invented the LED and here I am mesmerized by the possibilities. I just found these: 8080-SMD on aliexpress for $15.33. That's 50W of 365, 385, 395, or 405nm in a tiny chip ~12mm square. This is perfect for collimating.

Wow, 400W. I was holding the 2" 50mm Fresnel in front of the 50W LED and I could feel my fingers being in the beam. It's a weird sort of feeling because it comes with the knowledge that this is cell killing light. No funny smells like when you get a laser burn. Just that "I'm getting cancer of the finger" feeling. A 400W cob could probably cook your skin when near the source. I'm getting different eye protection too. I'm using the welding helmet for now. The image was taken on a vintage 35mm super takumar at f16 with a uv shade filter and its still saturated so you know its not something to screw around with.

Click to expand...

Nice work with the recent progress!

I've only ever driven the LED up to 350W (70V @ 5A input) as frankly the LCD screen couldn't cope with the full 400W thermally. At 350W the LED is actually emitting something like 90W of actual light energy, with the rest going to heat. Each of the fresnel lenses passes about 87% @ 385nm, so by the time that energy makes it to the LCD screen it's down to a power density of about 200mW/cm^2.

I don't really have much of an idea of what the LCD screen is actually capable of in terms of energy density, but my contact printing rig which is also LCD screen based and is designed for a 3D printer operates at 110mW/cm^2 so it should at least be capable of that. I found a paper somewhere that indicated the damage threshold for LCD screen polarising layers is on the order of 1000mW/cm^2 (using a laser). During early testing I taped a thermocouple to the LCD to get an idea of heating, and I have a very powerful and very loud cooling fan (like the fan draws ~70W) blowing a sheet of air over both sides of the LCD screen to keep it under 70 degrees.

When I originally looked at DMD's I also noticed that their rated power for UV wavelengths was practically nothing, which is why I went down the LCD route, but as long as you stay in the longer wavelengths you might be ok. I wanted to run at least 385nm because I will eventually try exposing DAS-sensitised carbon tissue which barely responds at all to longer wavelengths.

385nm is too-long of a wavelength to give you skin cancer or sunburn as I understand it, but it will definitely give you cataracts so eye protection is a must! You can obviously feel the heating of your skin if you put you hand in front of the LED though, it will melt a sheet of black plastic pretty quickly...

AndrewBurns said:

When I originally looked at DMD's I also noticed that their rated power for UV wavelengths was practically nothing, which is why I went down the LCD route, but as long as you stay in the longer wavelengths you might be ok

Click to expand...

Yeah, that's too what I gathered but now I'm not sure. I did have to reduce the optics to maximize power by getting rid of the zoom lens. That lens ate most UV power going into it. The prisms are halving the power twice but I cannot do much about that since the mirrors only tilt +-12degrees.

After some reading I learned that one other reason to pick a specific DMD is the protective window on the chip and its coating. That glass can be made to reflect or absorb UV. I think the big problem is to control heat per the specification. But then also we are removing most visible light and replacing that with UV. The power shift and reduction comes with two effects. One is that there's less heat-like waves past the 700nm but then UV is increased and most things absorb UV, thus creating heat.

That said, I have confidence that I should be able to expose cyanotype because I was previously able to do that with the high pressure lamp and the UV absorbing lens. So I know that there's definetly a useful wavelength that the chip can reflect.

The amont of light I can use on the chip is limited like Koraks said
"27-30V at 1-1.4A. This means 42W at the high end and I wouldn't run them that hot. I think realistically you can get 25-30W"
Of that power some of it is also going to heat. So let's say I'm getting 20W past the Fresnel, then 10W to the DMD and 5W to the Nikkor. I'm probably getting around 4W at the output.

One more issue with my arrangement was that when I accidentally closed the aperture on the Nikkor the effect was to crop the image LOL. So the expectation is a cone of light that focuses at the aperture so the aperture can clean the stray rays. But in my setup, at least as it is now, the beam focuses on the chip and expands towards the lens. There's not much I can do without sacrificing more power for a second Fresnel.

Yeah you lose more power with more optical elements but the benefit of the DMD is that it's WAY more efficient than my LCD screen. I'd guess the DMD would reflect like 80+% of the light going to it, while my LCD transmits around 4% of light at my wavelength best-case. The LCD is physically much larger than the DMD which means I can throw a lot of energy at the problem without making the average energy per cm^2 so high that it damages something, while the DMD is small so it can't cope with so much energy but it's more efficient so it doesn't need as much in the first place.

AndrewBurns said:

Yeah you lose more power with more optical elements but the benefit of the DMD is that it's WAY more efficient than my LCD screen. I'd guess the DMD would reflect like 80+% of the light going to it, while my LCD transmits around 4% of light at my wavelength best-case. The LCD is physically much larger than the DMD which means I can throw a lot of energy at the problem without making the average energy per cm^2 so high that it damages something, while the DMD is small so it can't cope with so much energy but it's more efficient so it doesn't need as much in the first place.

Click to expand...

I wiggled the 10W source to get it to be as bright on the image as possible. And I exposed this cyanotype. It took 3hrs and 30 minutes at 395nm. That's the longest I've run the projector with the rpi2040 ballast bypass. Its a 3X4" and I can totally see this working on shorter 385nm 50W. At least get to the same 15 minute exposures I was used to. If this doesn't work, I'm going to go to a 12 HID with the outer shield removed and a blue filter... I just need to figure out how to do that... I'm amazed at how much more power the original lamp must be dumping to get the cyanotypes exposed in 15 minutes. Or maybe my 10W lamp just sort of sucks.

So I got a 50W 380nm, a 10W 365nm and the 10W 395nm I just used. Using the 10W 365nm on my multimeter I measured a voltage when exposed to the 50W 380nm, but no voltage when exposed to the 10W 395nm. I'll be measuring the extra projector #1 with the standard lamp which I repurposed to watch movies LOL. I'll see if I get 365 or 395 power. Maybe that will guide me on what wavelength lamp to get if the smaller 50W 385 doesn't pan out. What I've read/observed is that 365 gets absorbed too much, 395 appears to be bright on the florescence screen but doesn't produced a fast cyanotype. So 380, (I could only find that lamp in 385nm) may be the sweet spot.

W.r.t. the wavelengths you're probably running into a tradeoff situation where you have to find an optimal compromise. I expect that photon absorption of ferric ammonium citrate will be vastly higher to 365nm than to 395nm. At the same time, extinction rate in the optical system of the 365nm will be much higher than for 395nm, which combats the earlier effect. Furthermore, I expect that the electrical efficiency of the light source will be better for the longer-wavelength LED.

koraks said:

W.r.t. the wavelengths you're probably running into a tradeoff situation where you have to find an optimal compromise. I expect that photon absorption of ferric ammonium citrate will be vastly higher to 365nm than to 395nm. At the same time, extinction rate in the optical system of the 365nm will be much higher than for 395nm, which combats the earlier effect. Furthermore, I expect that the electrical efficiency of the light source will be better for the longer-wavelength LED.

Click to expand...

Yup, I measured the standard 400W UHM lamp at the lens output and it had no measurable 365nm and just the tiniest bit of 395nm. So it may be that going to 385nm will be a game changer and or going to UHM or halide. There's in particular a halide lamp used for small reptile enclosures size G9 that could work well with a reflector and a blue filter. It specifically says UVA and heat.

Yeah, that 365nm peak is nice. However, regardless of what the 365nm wavelength is made by, as long as your optics attenuate it strongly, there's not much you can do with it. In that sense, the sweet spot may indeed be 385nm since things tend to get iffy for many materials at about 365nm. There's probably some kind of plausible physical explanation for it (idk, the bandgap of some common material?), but I don't know it.

koraks said:

Yeah, that 365nm peak is nice. However, regardless of what the 365nm wavelength is made by, as long as your optics attenuate it strongly, there's not much you can do with it. In that sense, the sweet spot may indeed be 385nm since things tend to get iffy for many materials at about 365nm. There's probably some kind of plausible physical explanation for it (idk, the bandgap of some common material?), but I don't know it.

Click to expand...

I decided to give the larger lamp a go. Its supposedly 50W 380nm. I don't know why I didn't try it earlier, I think it was just much messier once I saw how small the 10W modules could make the setup. I tried a few things. One of those things was to figure out if I could make a ~20mm spot. I installed a diffuser sheet in front of the COB's lens and then I used the 50mm lens in front of it. To no one's surprise there was a nice spot and it made a nice circular spot on an old dry unexposed cyanotype sheet. It went from nothing on the sheet to Prussian white in less than 10 seconds.

So I got to work on an adapter for the 50W COB. Its printing. can't wait.
The diffuser sheet will go on one of the 3 faces. probably face 3 so that its image is as fuzzy as possible. The Fresnel lens is 50mm and its too short to take the LED's image onto the DMD. Probably an 80 or 100 would be closer. but it makes a good spot. Another thing I did was to place a CD that had not been written to yet on the face of the LED for an hour. There was no discernible change at all. I spoke to a photonics college at work about the effects of UV on optics coatings. According to him the optics will get damaged at UV power closer to 250nm and its mostly the ozone doing that over a period of months. The plastics will probable degrade and yellow. The DMD can be damaged if a hot spot is projected on it, so its mostly localized heat when UV turns into heat that the surface of the micro mirrors that might damage them. Whatever it is, its unlikely to get damaged on a few tries, so I'll do that.

It works! However its still slow. But these images give me hope. The biggest part of that hope is that central hot spot. The light version is 30 minutes with a diffuser minus a lens that's probably a 50mm. The darker version is 30 minutes without the diffuser and with that extra 50mm lens. After making the darker image I went to wiggle the light to see what sort of artifacts I would see on the screen. The darker spot in the center part of the image is an enlargement of 1 of 50 LEDs in the COB!!!

1) 380nm/385nm is a good wavelength
2) I need to image the LEDs, as many as possible at the DMD chip with a diffuser

If I can accomplish that, the image should expose in around 1 minute at 3"x4" size. So far heat has not been an issue at all. Not at the DMD or at the LED.

While I wait for the more concentrated ~10mm^2 50W 385nm source, I thought to try a few more things. One of which was really interesting, my phone's flashlight can turn cyanotype sensitized paper to Prussian white in a couple of seconds when placed on it. It has me thinking. Another thing I'm going to try is printing a lenticular lens in clear resin. Then lens would then collimate every 1W beam and then easily point it on to the DMD. I was going to try a Fresnel design, but this makes more sense. It doesn't seem to be super critical for the light to be all a nice Gaussian beam with perfect mode and wave front. So I tossed one 1W source in blender using luxcore to make a proper lens using opticore. I can't believe all these things are freely available. For opticore I do wish I could adjust the lenses later. You get to create the lenses but the parametrical aspects end there. So you have to iterate a few times. From there I'll recreate the geometry on a CAD assembly with all 50 lenses pointing in the right direction and do the 3D printing. This is definetly something unique to LED sources, you can use weird lenses to cheat a little around the etendue on a per source basis. That's the theory anyway. I haven't tried this before.

Also, I did a great job as glueing my 50W LED assembly. I thought it would simply pop off after a few days lol.

Its working in blender/Luxcore! I have internal joy in what this might actually mean. Will this really actually focus most of the 50W towards the center? I gotta print it!




For one thing, this is really thin and I know its going to warp as soon as I try to wash it. Maybe I shouldn't wash it, just cure it after print. Then there is the issue of supports.

For a moment I thought it would work. Now I gotta wait a couple more weeks for the smaller 50W 385nm source. I mean, it did do something, it blocked most useful light because I could not get it to expose cyanotype. It did not focus the light to a spot. But the lenticular lens does actually look like an optical lens. You can look at a light thru it and watch the bright spots on the lens move around. So it is definetly bending the light. But maybe it needs to be higher refractive index and more UV transparent. I made it as thin as possible only .6mm at the base and maybe 2mm to the tips. And it fit right on the spots. Just not optically useful for UV LOL.

That doesn't look like a lenticular (?) looks similar to a ball lens array used in plenoptic systems.

Not sure if photopolymers are ideal, they absorb UV to polymerize after all. Have a look at this, the microlens can be made without 3D printing if you want to give it a shot.

calebarchie said:

That doesn't look like a lenticular (?) looks similar to a ball lens array used in plenoptic systems.

Click to expand...

I'm unsure of the terminology, so I did a little digging on Wikipedia.

A lenticular lens is an array of lenses, designed so that when viewed from slightly different angles, different parts of the image underneath are shown.

The most popular array is a cylindrical array used for moving images where each lenticle allows the viewing of only a certain part of an image. The image is then made up of strips from different images.

A micro lens is a lens smaller than 1mm. So my lens array is not micro but it is lenticular. Its composed of 50 individual parabolic lenses that are supposed to be Oriented at slightly different Angles such that all light collected is pointed to the spot I chose. Each of the 50 lenses is located exactly over the LED emitters on the LED COB. Each COB radiates its energy at a different Cartesian position but then the lens is supposed to point each emission towards the same final spot. I think it didn't do that because the surface finish is awful and because it absorbs most of the UV light. If it was working I would at least see some of the visible blue form a spot.

So failed lenticular millilens array? LOL. Anyway it works on the same principle as a lenticular imager. My original image being composed of 50 ~1mm squares equally separated and surrounded by areas with no light. Instead of a moving picture, the viewer would see black from all directions except the one spot.

Unfortunately this cannot work by simply tossing two cylindrical arrays on top of each other perpendicularly. The lenses must be just about the same size as the image or light source and must be perfectly aligned and scaled so each lens is aligned and spaced perfectly over each spot.

I bet I could 3D print a cylindrical lenticular lens for photos. It would be easy to polish! I think a ball lens could work too. If I could find a UV transparent resin. Maybe normal resin is more transparent to UV than 3D printing resin? Then it would be a matter of printing a pattern on paper and wacking it on a piece of aluminum or tin or cooper to use as mould. I could maybe achieved the orientation by striking from the focal direction? LOL, I might land on this after the new smaller 50W source also doesn't work.