The Trials of Turning a Digital Light Projector Into an Ultraviolet 3-D Printer

An enticing idea turns out to be devilishly complex


The basic idea is so seductive. Forget about forming an object from spools of plastic melted and laid down by a tiny nozzle as most consumer 3D printers do. Instead, use an optically reactive resin and modern display technology to zap an entire layer of solidified material into place in one shot. Then the next layer, and the next, until the object is complete.

After all, some professional stereolithography printers, such as the Form 1 and 2 and the Nobel 1.0, do exactly that. But could I, one person working at home in his spare time with mostly surplus parts, build such a machine? The answer is—spoiler alert!—almost. Like a tiny home-brew analogue of fusion power, my printer seems to stay just another month or two of concentrated work away from operation, no matter how many months go by.

My machine is based on a DLP printer, with the DLP standing for Digital Light Processing, a Texas Instruments technology that uses an array of individually controllable micromirrors. Ultraviolet light from a mercury bulb or an LED is directed at the array. Each micromirror sends light either to a beam dump or to the projector lens to focus an image corresponding to a layer of the object. The image shines through the bottom of a vat to cure a microscopically thin layer of UV-sensitive resin. Instead of spreading the beam onto a wall, the system focuses the image onto an area maybe 10 centimeters across, so the physical resolution of the resulting 3D object is a small fraction of a millimeter. A movable stage lifts the cured layer up a fraction of a millimeter, more resin flows underneath, the projector displays the pattern for the next layer, and so on.

To construct this setup, I bought a DLP projector from an Internet overstock site for US $300. I tentatively dubbed my machine Lobachevsky (after the old Tom Lehrer song) because almost every part was lifted from some other project. The CPU for driving the projector is from an old digital picture frame, the aluminum extrusions from a disassembled ER1 robot, the stepper-motor drivers from my attempt at building a conventional 3D printer, and so forth.

After a few false starts, I figured out how to choreograph communications among the computer I was using to handle the 3D object data, the picture-frame CPU, and the microcontroller operating the stepper motors that raise and lower the stage. Then the real problems started.

Every step in the simple description of UV projection stereolithography I gave above hides a pitfall. For example, a standard consumer DLP projector uses a spinning filter wheel to subdivide each video frame into different time periods. During each period only red, green, or blue light passes, which also blocks UV light. So I had to get rid of the filter wheel. But I couldn’t just unplug the wheel from the projector’s motherboard, because the firmware will provide power only for the projection lamp after it’s detected that the wheel is spinning at the right speed. Oh, and when you open up the projector to start messing with its internals, there’s more firmware that reads a pressure switch that turns everything off, so you have to defeat that interlock too.

Meanwhile, the bottom of the vat where the printing happens must be transparent to UV and also completely unwilling to adhere to freshly cured resin. Other stereolithography home brewers have tried covering the bottom with a layer of PTFE film; a silicone gel used to encapsulate solar cells; various soaplike substances; and even carefully calibrated salt- or sugar-water solutions that are denser than the resins and so pool beneath them. When I tried the liquids, I was treated to an amazing display of surface chemistry: As long as I had just a tiny pool of resin on top of my solution, it floated in the middle of the vat just fine, but as soon as any resin touched the edge of my container, it would pull the rest of the pool over to the edge and then down to the bottom.

The encapsulant seemed to work better; objects formed, and sometimes the cured resin successfully detached from it as my stage lifted up. But that’s when I discovered an entirely different problem: The resin wasn’t actually curing as a result of the UV exposure from the carefully focused images coming from the projector. Instead, the pattern of light from the projector was heating small areas of the resin, increasing its sensitivity to ambient UV from light leaking out of the projector. As soon as I cut off those leaks, I stopped getting any curing at all. Further investigation showed that the projector optics included a previously unsuspected UV filter incorporated into a block of glass that protects the rest of the system in case the projector’s bulb explodes. So I replaced it with a simple block of heat-resistant glass of precisely the same dimensions.

But while I was acquiring that block and installing it, time passed. My UV-sensitive resin, as complex chemical mixtures will, lost some of its sensitivity. It still curdles in response to a UV LED, but the light from the projector does nothing for it at all. I have more resin on order.

While I wait, there’s another, potentially much simpler method that’s attracting my interest. I took the backlight off an old LCD and now I can shine whatever I want through it. A UV LED produces a perfectly legible first layer at least, just like the contact prints 19th-century photographers made from their page-size glass negatives. The resolution of the old screen is miserable, and I know the ultraviolet will eventually damage the liquid crystals in the display, but it was due for recycling anyway. And if this version works, maybe I can convince someone to give me an out-of-date tablet or mobile phone, with 10 or 20 pixels to the millimeter. I’m sure it will be a simple matter of software to control the display…

This article appears in the July 2016 print issue as “DIY Light-Based 3D Printing.”