Most resin printers cure an entire layer at once through an LCD mask or a galvo-steered laser. A new build from the YouTube channel Diffraction Limited abandons that model and does something closer to what an FDM printer does: it moves a single point of ultraviolet light along a toolpath, one line at a time, and builds parts so small that a Benchy comes out 150 micrometers long. The video, Micro-Fabrication with a Fiber Tip – A Low-Cost 3D Micro Printing Method, was picked up by Hackaday on September 6, and it reads as three separate projects finally converging into one machine.

The three ingredients are an open-source stepper-driven micromanipulator the same maker published a year ago, a home-built fiber-coupled 405 nm laser source that Hackaday covered at the end of June, and a jar of turmeric powder. That last item is not a joke. It turned out to be the fix for the problem that initially capped the printer's layer resolution.

A fiber tip as a print head

A 405 nm laser diode is coupled into a single-mode optical fiber with a 3-micron core. Rather than focusing that beam onto the resin from a distance, the maker treats the bare fiber tip itself as the print head. The micromanipulator holds the tip just in front of a build plate that has been coated with photopolymer resin, then drives it along toolpaths generated by an ordinary slicer, in the same way a hot end traces perimeters and infill on a desktop FDM machine.

Resin cures in a narrow cone where the light exits the fiber. According to Hackaday's writeup, because the fiber core is only three microns across, the cured material shears cleanly away from the tip as the head moves on, so the fiber is not dragging a growing blob of solid resin behind it. Reusing slicer output is the clever economy here: the FDM world had already solved the toolpath problem.

The layer-height problem, and the spice rack

The first prints ran into a limit that anyone who has fought overexposure on a masked-SLA printer will recognize: ultraviolet light passed through previously cured resin too easily. When the tip exposed a new layer, the light kept going and cured material it was not supposed to touch, which put a hard floor under how thin a layer could be.

The maker's answer was curcumin, the yellow pigment in turmeric, used as a UV-absorbing dye dissolved in the resin. Extracting it required nothing more exotic than alcohol, which pulls the dye out of turmeric powder easily. With the absorber in the mix, light from the fiber tip is soaked up within a much shorter distance, the cure stays confined to the intended layer, and the minimum layer height drops.

Curcumin also leaves the finished parts slightly fluorescent under UV light. Locating a 150 µm Benchy on a build plate under a microscope is a genuine problem, and fluorescence makes the parts easier to find and pick up.

The results, per Hackaday, are a series of Benchies 150 micrometers long, a Stanford bunny that is dwarfed by a human hair, and a handful of other microscopic pieces.

The hardware underneath

None of this works without positioning far beyond what leadscrews or belts can deliver.

The micromanipulator, which Hackaday covered in September 2025, uses three stepper motors under closed-loop control from magnetic rotary encoders. Three pairs of parallel linkages, swiveling on ball joints rather than flexures, connect a crank on each motor to a small mobile stage. The encoders resolve 100,000 steps per revolution, and an array of alternating-polarity magnets on the rotor acts as a kind of magnetic gear that pushes the effective resolution to roughly 50 nanometers over 23 mm of travel on each axis. In the original demonstration, run slowly to limit thermal drift, the maker traced the first layer of a scaled-down Benchy inside a 20 µm square. The controller is a Raspberry Pi Pico 2 with three motor drivers on a double-sided, through-hole carrier board, and the whole design is on GitHub as MicroManipulatorStepper. At the time, the stated target applications included maskless lithography, electrochemical 3D printing, focus stacking, and micromachining.

The laser source was the second piece, and Hackaday covered it on June 30. A 405 nm diode is coupled into single-mode fiber through an aspheric lens, all held in a CNC-machined brass housing. The fiber is fixed in place with UV-set adhesive, and two set screws under the fiber's mounting pin exist for a reason that will sound familiar to anyone who has glued optics: the adhesive shrank as it cured, pulled the fiber out of alignment, and killed the light output until the screws brought it back.

What It Means for Makers

A single maker, working from an open-source stage and a home-built optical source, has demonstrated a low-cost route to micron-scale printing using the same slicer workflow that drives a desktop FDM machine.

Three things stand out for anyone considering a replica.

First, the positioning stage is the intimidating-looking part, and it is already published. The MicroManipulatorStepper repository covers the mechanics, the electronics, and the Pico controller firmware. The mechanical parts are designed to be machined on a 3-axis CNC mill, though the repository also supplies STL files and notes that 3D-printed parts work at the cost of some thermal drift. The motors are off-the-shelf NEMA 17 steppers, the electronics use commonly available driver and MCU modules on a board that needs no SMD soldering, and the ball-jointed linkages are within reach of a well-equipped home shop.

Second, the optical source demands machining and patience but no exotic components. The lesson about adhesive shrinkage is already documented, sparing the next builder a frustrating afternoon.

Third, the resin chemistry problem that would normally send a hobbyist hunting for specialty photoblockers was solved with a kitchen ingredient and a solvent. Light penetrating deeper than the layer it was meant to cure is the same phenomenon that costs detail on conventional masked-SLA printers, and a UV absorber is the tool for fighting it. Curcumin is not going to replace formulated additives in production resins, but it is a cheap, visible way to experiment with cure depth.

What the summary coverage does not tell us is how fast the machine prints, how repeatable the results are across many parts, or how the tip and resin behave over long runs. Those details live in the video, which anyone weighing the approach should watch in full. As a proof of concept, though, it is hard to argue with a Stanford bunny you need a microscope to see.

Sources