How UV Printing Actually Works

1. How a drop gets out of the head

A printhead is thousands of microscopic ink-filled chambers, each ending in a nozzle. There are two ways to get a drop out. A thermal head — your home HP or Canon — flash-boils the ink at around 160 °C and lets the steam bubble pop a droplet out. It is cheap, but the ink has to survive boiling. A piezo head uses a piezoelectric crystal, a material that physically flexes under voltage, as one wall of the chamber. It works like a micro-pump: drop the voltage and the chamber expands and pulls ink in; wait microseconds for the fluid to settle; drive the voltage and the chamber squeezes and a 3–4 picolitre drop breaks off. Drop formation takes tens of microseconds; the flight to the surface takes about 500.

UV ink would cure inside a thermal head, so every UV printer ever built is piezo. That is why all six machines we compared run an Epson head — same supplier, same physics, two different parts.

The “waveform” both companies talk about is the exact voltage-versus-time curve driving the crystal. Get it wrong and you get satellite droplets that drift in the airflow and land as speckle. Both Morpho and HeyGears showed us in-house drop-watcher rigs — a strobe and a camera — for tuning it.

2. The honey rule, and who does the heating

Heat makes liquids runnier, not thicker — honey from the fridge versus honey from the microwave. Viscosity falls as temperature rises. That matters because stronger cured parts need higher-viscosity chemistry, in HeyGears' words: the lower the viscosity, the worse the mechanical performance. But thick fluid will not jet through a nozzle. The trick is to heat the ink so it temporarily runs thin during jetting, lands, cools and cures strong.

The milkshake version, which is the one worth remembering: each head is a kid with a straw. The F1080 only drinks thin shakes; the i3200 slightly thicker ones. A small difference — except that a thick shake means strong plastic, and it can be warmed until it is temporarily sippable and then thickens again on landing. That is the whole difference between printing paint and printing plastic.

But be precise about the microwave, because this is where almost everyone including us gets it wrong. Epson's own I-series spec table lists “Heater: Not available” and “Ink Recirculation: Not available” for the i3200, and publishes no viscosity range at all. The head has no heater.The machine builder heats the ink path and the reservoir so the resin arrives in-window at the nozzle. Say “the machine heats the ink,” never “the i3200 has a heater.”

Three separate jobs make viscosity the binding constraint. The push: the pressure needed to force fluid through a nozzle rises steeply with viscosity and with the fourth power of nozzle narrowness. The pinch-off: the drop has to neck and snap free — surface tension snaps, viscosity resists, and too thick means the ligament stretches and slurps back or tears into satellites. Fluid dynamics gives every head a jettable window that no amount of waveform tuning escapes. The refill: chambers refill by capillary suction between fires, about 30 microseconds at 30 kHz, and thick fluid crawls. An unrefilled chamber starves and misfires. So viscosity caps drop quality and firing speed, and heating helps both at once.

3. UV ink and UV resin are not the same fluid

They are the same chemical family — acrylate liquids built from a pre-polymer backbone, a monomer solvent, a photoinitiator, pigment and additives, which crosslink into solid plastic under UV. What differs is what they are engineered for.

UV ink is pigment-dense and designed to be a decorative film tens of microns thick on top of a substrate. Its strength is scratch resistance and adhesion; the substrate does the structural work. UV 3D resin is engineered for bulk mechanical properties, because the print is the object: higher viscosity, different pigment loading, and deliberately more transparent on a machine like the G1X because soluble support strips surface colour.

On a dual-mode machine these are literally different bottle sets, different hubs, different modes. You cannot 3D print out of 2D ink any more than you can build a chair out of paint.

4. Why overhangs are impossible

A 2D UV printer's droplet always lands on something. In 3D mode you are stacking thousands of layers in open air, and a jetted droplet is a low-viscosity liquid until the lamp hits it. Any drop with nothing directly beneath it falls or slumps. HeyGears' engineer: you cannot even have a millimetre of overhang, because the liquid viscosity is too low.

That is genuinely unlike the two processes people compare it to. SLA and MSLA form the part inside a pool of resin, so the surrounding liquid and the cured layer below hold everything together and supports are sparse sticks you snap off, leaving scars. FDM extrudes semi-molten plastic with enough body to bridge small gaps. Material jetting gives every voxel a floor, so the machine co-prints a full support cocoon under and around everything.

Follow that one decision downstream and you have most of the ownership experience: support consumption is huge, hollowing saves no material because the hollows get filled with support, the surface is matte wherever support touched it, and the part then soaks for six or seven hours. The upside — zero-scar, zero-labour support removal — is the same physics read from the other side.

5. Slicer versus RIP, and why 3D colour is hard

Two different worlds have two different words for turning a design into machine instructions. A slicer, from 3D printing, turns a mesh into layer contours and toolpaths: intersect a horizontal plane with every triangle, chain the resulting segments into a closed outline, repeat every 10–20 µm of height. Pure geometry. A RIP— raster image processor, from 2D printing — turns an image into per-channel dot patterns. Its core step is halftoning, which is dithering: a nozzle cannot fire “40% cyan,” it fires or it does not, so continuous intensity becomes a spatial dot pattern the eye averages. Before that comes ICC colour mapping, squeezing screen RGB into what these inks on this material can actually make.

Full-colour jetted 3D needs both fused together, and two things break in the process.

The Z-neighbour problem. In 2D, a dot mixes optically with the dots beside it, on one flat plane, over a white base — colour is a reflection off the white underneath through tens of microns. In 3D the colour skin is many stacked layers of translucent resin: light enters, scatters and is absorbed through the stack, bounces off the white core, and exits. What you see at any point is the optical sum of a column of drops, so the dither pattern has to be computed volumetrically. No off-the-shelf RIP does this.

The flatness constraint. In 2D a sparse dot pattern just looks lighter. In 3D a sparse pattern is a thinner layer, and thickness error compounds over thousands of layers into grooves and valleys. So the halftoner carries a second simultaneous constraint: every cell gets roughly equal total ink volume regardless of colour, with transparent resin acting as blank filler where colour is sparse. The colour pattern and the physical shape are one decision.

That is why a 3D colour RIP had to be built in-house. HeyGears started theirs in 2024 and says openly that there are a lot of ideas they have not implemented yet — which is the software-maturity risk in one sentence. It also explains the pale prints, through a single causal chain: support must dissolve scar-free, so it chemically intermixes with the colour surface, so the colour resin must be translucent to survive contact, so translucent colour over a white core makes colour volumetric and thickness-dependent, so halftoning gets harder and prints come out paler. One chemistry decision, four consequences.

6. The detail that gives the game away

Some of these machines run fibre-optic data lines to the printhead. Nothing to do with curing light — they carry data. A 3D job runs the carriage back and forth continuously for 24 hours or more; copper ribbon cable fatigues after a few hundred thousand flex cycles, while fibre survives millions and is immune to motor EMI. It is a quiet tell that 3D duty cycles are a different reliability regime from 2D printing, and the kind of engineering choice that only makes sense if you expect the machine to be printing for days.