Projector Cyanotype and Chiba python digital negative.

Today I looked into calibrating UV focus. Holy moly! Look at that! My focus was at zero in visible light but in UV it's shown at 16 or more in this chart.
The card was at 45 degrees, the distance to 16 mark from 0 is 76mm so it means that I have to focus my projector 54mm in front of the actual photo. And I just did my first image with these changes and boy! wow! its nice!

But one thing I noted was that the focus 54mm away is around 10%. Every footnote i've read says that UV focuses 5 to 10% in front of visible light. Here for whatever reason, the UV light is focusing 10% behind visible. I guess I should not really expect a particular number or direction since lens makers can pretty much swing the focal plane any way they want. Any comment or opinion on this? The lens on this projector is weird. It has this sort of central pinch to it. Like the center part of the lens looks almost flat if not reversed inward. I thought it was maybe parabolic or hyperbolic or something. but if you look from the side you don't see the typical parabola shape, you see this center pinch. The same with the light's lens. I think they are playing some sort of focal plane correction from both sides of the optical chain.

OK So last week I worked on the focus thing. Today I gave up on processing the images locally on the little thinkcentre computer. Instead I went back to Mr ChatGPT and I asked it to build me an HTML page and serve it locally using podman. The benefit of this is that you can have a big powerful computer do the image crunching and a tiny computer do the projecting. It can be the same computer it doesn't matter. To use it:

You will need Ubuntu or some other Linux. I don't do windows programs sorry for you if you must use windows:
first allow the network traffic :
sudo ufw allow 5000
Then
podman run -d -p 5000:5000 my-flask-app

it looks like this:

then just open an email file, hit process image. That will start the threshold as a series of images in a cache folder on the server. then when it is done, go hit start or stop. Oh and input your exposure time. Its really working finally!! like if you set it to 12minutes it will expose for 12 minutes!


One last bit. On Firefox (friends don't let friends use chrome anymore) go to : about:config and then search for "full-screen-api.ignore-widgets" double click her to set it to true. Now if you press F11, the top menus of your browser will disappear and you will have a single window with your image ready to go. Hit start then take the pointer and put it somewhere else so the menus autohide. Its working OK. It probably needs some TLC but I am not complaining.

my projector is 1080p, but you can change the native resolution in the python app called app.py:

I'm following this recipe: https://lemmy.world/post/26591859 and I'm getting photographic level resolution on my cyanotypes. On textured paper its interesting because you can Illuminate it directly overhead and see the resolution you saw when it was wet, but as soon as the illumination comes from elsewhere it looks like a regular slightly blurry cyanotype. On glass and plastic as well as flat photo paper it keeps its full resolution. The formula comes from the latest experiments by Habib Saidane but are modified and a completely separate thing. The tissue poured with the sensitizer doesn't actually dissolve away when developed in water. the image is embedded inside of it. The formula seems to clear to paper-white unless you use some sort of staining liquid.

Finally! After solo many months of wandering around with all sorts of electronics and ideas I got my DLP projector to post and then go to desktop.

As they say, necessity is the mother of invention so when the normal lightbulb went out I started desperately trying to reverse engineer the signal to the lamp so I could play it back. Well as it turns out it cannot be simply played back unless you got exact timing. Digging deeper into the burnt ballast I found that 3 of the wires went into the LED side of an optocoupler and the 4th wire went into the receiving end of an optocoupler. So the 4th wire was the signal back from the lamp's ballast. Now I don't need the lamp or the ballast. I get the clock signal, the command signal and the video on signal and basically I replay the response digits and the exact time needed. Now I need to write a protocol to send responses to the commands and to force the video signal on.

The whole thing is just so I can try removing the last mirror so I can stick a UV LED there to project it down to make prints. Man, when I saw my Linux desktop on the DMD chip I was so freaking happy. But there are so many open issues between now and being able to make images. The biggest for me is what the DMD chip will project when I shine UV light. The high voltage halogen lamp gets focused into a rectangular glass rod with mirror faces. The result being that the light gets mapped into infinite geometrical replicas. Then the light passes thru the Color wheel which is no doubt synchronized using the clock signal. I bet the lamp flashes bright as the correct color passes and then the DMD aligns the correct set of mirrors for that particular color at that particular time step. After the wheel the light reflects off a final mirror and into a lens that is bonded to a beam splitter prism. So the image is focused by the lens on to the DMD, then the reflection bounces off into either the lens or a little beam dump. The reflection angle is definitely not 45degrees. So there's the question... After all this work am I just going to get a weirdly parallelogramed image worth less than a fly's posterior?

I just like over complicating things I guess.

Today I put together the TX and RX for the projector and it works beautifully!

I should now be able to remove the first mirror and replace it with a UV LED...of some sort. So the project goes from normal projector with 400W high pressure metal halide lamp cyanotypes to software bypass to optical work. I bet I'm going to get just awful images

The DMD chip is sort of hard to see thru the lens and it flickers with my phone's lamp.... Hmm I just realized that the flickering is due to my phone's LED syc to the DMD color drive....without any interaction with the Color wheel, it's just the chip moving the mirrors at that frequency. Wild!

I took a picture of the DMD running. You can sort to see the Ubuntu 20.04 desktop update window.

That was like a year of smashing my forehead up against UART walls left and right. Every single post out there is using an Arduino listening to 8bit UART. As soon as you realize the signal is 9 bit, all hell comes down on you. There is no standard 9bit UART code for arduino or the chip I'm using the RP2040 Zero. But finally!

At the top on Channel 0 you can see the beautiful binary going from 0 to 1 for video on. Below on CH1 we see the clock in brown. This signal appears to be a tachometer for the color wheel. The red and yellow are queries from the projector and the mid orange are the replies to keep the "lamp" on. Of course, there is no lamp in the projector and the ballast is disconnected.

Hilariously I can now use my ryobi shop LED lamp to power the projector lamp. Interestingly the projected image is crisp. I expected a blurry mess. I also tried with my 380nm lamp. Currently that lamp is not focused, it had a 120degree spread so the image is visible but very dim. I think I might keep the glass integrator rod and color wheel for now. If things are too dim for use I'll remove those things and try without.

Here's the projector projecting a photo using the ryobi lamp. That lamp makes a pretty large circle but the glass integrator tube take only a few mm^2 of all that. The lens is a pinch lens that tosses the light towards the center in a tiny spot. The integrator rod is a rectangular glass with mirrors on all 4 lateral faces such that if you were to ...it's a 4 sided Kaleidoscope. So if you stick a light into one end you see hundreds of replica images in a quadrilateral matrix. I assume that is how they make the light homogeneous when it hits the DMD chip......man, I'm so far from making a cyanotype ever again with this thing. But I'm just extatic to have a working DMD mirror.

The 25mm square 50w cob is not going to make a good focused image. There's no way around the collimation problem. If your source is large and has large divergence it will not collimate well. But I just found some LEDs that are 10w at 5mm diameter. There are also 3w 365nm lasers that could work.

Playing around with my 50w 380nm cob, I can make Prussian white within 5 to 10 seconds near the LED emitter. But the effectiveness seems to drop very quickly. 380nm is probably the way to go with the DMD chip. Not only is it more efficient than a 365nm but the light will be less absorbed by the optics. For optics my plan is to use two convex fresnels. One will be 50mm and the other will be 150mm.

I have to get over my lazy mood and go cut some plastic to rehouse the projector into a useful form. I also purchased a third projector as a parts backup in case I burn off the DMD chip for example. Speaking of which, I've been thinking of what material I could test to see if the UV will quickly delaminate the DMD chip. Well what about a writable DVD disk? So I exposed one for a few minutes to the 50w 380nm COB and I got nothing. No damage.

I've been turned off and uninspired lately by the lack of actual prints and the dull search for a good light source. I gotta get over the hump. Finally today I found some lenses and the 10w LED....but $30 more bucks grr! This is truly starting to feel like a hobby.

You could try to add a collimator to your COB LED. Perhaps if you root around a bit it's possible to find a combination of lenses that will effectively collimate the light. The net usable output will probably still be more than what you get with a 10W single emitter.

Your remarks about 380nm vs. 365nm make good sense to me. I can't advice on the delamination testing; my primary concern btw would not so much be the UV as such but mostly the heat.

koraks said:

my primary concern btw would not so much be the UV as such but mostly the heat.

Click to expand...

I did measure a black piece of metal under the beam intensity with a contact thermometer and it did not go over 45C. The chip likes staying near 25C but can tolerate higher temps and it comes with its built in fin cooler and fan.

These are two images of the cob and a single emitter using the same lens. The single emitter would probably damage my phone if I made the spot smaller. The cob on the other hand won't get any smaller. My DLP chip is 16mm diagonal so the collimated beam needs to be around the size of the single emitter image. I'll still be removing some of the power using a rectangular aperture. Reading up on DMD chips, they can only handle high intensity beams on the reflective surface. Everything around may absorb too much power too quickly for the cooler to be effective.

Anyone reading this should be aware that I am using yellow polycarbonate glasses for UV and a welding helmet for beam intensity. Do not, and I repeat, do not play around with your eyeballs. I used to go burning ants as a kid and I got lots of floaters, I wonder where those came from. Wear your ppe.

This projector was a good find. The light engine is all one piece and the lamp, color wheel and integration tube are all removal. Here you can see what I'm left with. Its basically a lens with a dichroic beam splitter into the DMD, that's it. So you input a beam and by default it gets tossed to the side into a beam dump plate. There is a beam on the screen, but I just realized that was a reflection from a piece of tape covering the cavity. My work area at home is very dusty at times when I cut stuff on one of the saws.

Well, I got my second wind after work. I decided to make a permanent mockup of the baseplate out of wood. It'll work just fine.

Its gluing now. Tomorrow I'll add the holes and then figure out how to mount the various fans and thermocouples. The color wheel needs to spin too.

Finally did a dry test. Before, I used a 1st surface mirror, and the tiny mirror gave me enough power to expose cyanotype like 10 times faster. But then with the DMD mirror I got like 100 times less power. That was using my tiny LED+multimeter tool.

I'm going to remove the lens to see if the lens is what's eating the power. I'll have to replace it with a UV enlarger lens.

I probably also need to get the light source closer to the DMD chip. Maybe I should remove the beam splitter prism.

imgprojts said:

Finally did a dry test. Before, I used a 1st surface mirror, and the tiny mirror gave me enough power to expose cyanotype like 10 times faster. But then with the DMD mirror I got like 100 times less power. That was using my tiny LED+multimeter tool.

I'm going to remove the lens to see if the lens is what's eating the power. I'll have to replace it with a UV enlarger lens.

I probably also need to get the light source closer to the DMD chip. Maybe I should remove the beam splitter prism.

Click to expand...

Hmm maybe I have to keep that lens and prism in place. Yeah, the only way forward that is practical is to get an LED that is slight smaller than the DMD and to get it as close as collimated as possible to the size of the DMD. Maybe going to blue light only would be better because I can get a 40W LED of 5mm diameter.

Wow, I just ran the thing without the lens and boy, the resolution of the chip is incredible at that size.

Also, yeah the lens pretty much attenuates all UV and unfortunately so does that prism. Hmm maybe this whole time I was only using a combination of green and blue light. At those wavelengths its okay to dump uddles of power on the DMD... ~22W/cm^2 max between 410nm and 800nm.

I just dumped another $50 into this project I'm using 2 fresnel lenses, a 28mm and an 80mmm to focus a 5mm wide 265nm to 400nm 10W LED onto the ~16mm wide DMD.

Interesting, looking forward to seeing how that fresnel setup pans out.
What software did you use for the ray tracing simulation?

imgprojts said:

Hmm maybe this whole time I was only using a combination of green and blue light.

Click to expand...

You don't get much response from ferric salts in that part of the spectrum, though.

koraks said:

Interesting, looking forward to seeing how that fresnel setup pans out.
What software did you use for the ray tracing simulation?


You don't get much response from ferric salts in that part of the spectrum, though.

Click to expand...

It'll be a couple of weeks from aliexpress lol. I used https://phydemo.app/ray-optics/. It's 2D and simple. In blender.org you can add luxcore 3D and simulate a beam using simulated fog. I've designed a few things that way without needing zemax.

Thanks, that's interesting; I should look into this one day!

The incident light into the DMD must be at 12degrees (DLP670RE 0.67 WUXGA). For this, they use a lens and a beam splitting prism, but that is right off the bat more than 1/4 of the beam gone! I wonder what I could do here.