Resolution vs. channel size
A printer’s pixel pitch is not the smallest channel it can make. The gap between the two is where most first-chip failures live.
Why a channel needs several pixels
Section titled “Why a channel needs several pixels”A channel is a void. To leave it open, the machine has to not cure resin there — and several things conspire against that.
Cure-through. Light penetrating past the exposed layer cures resin below and beside it. The void shrinks from every side.
Pixel edges are not sharp. An LCD pixel has a finite aperture and the light diffuses through the FEP and the resin. The boundary between exposed and unexposed is a gradient a few pixels wide, not a step.
Resin does not leave. Even where nothing cured, uncured resin has to actually flow out. At sub-millimetre scale surface tension is strong and viscosity is high.
Together these mean a channel drawn at one pixel wide prints as nothing at all, and a channel at three pixels prints unreliably.
Five to six pixels is a conservative working rule for a feature that survives
washing and curing. That is where the shipped
printer profiles get their
minFeatureSizeMm.
| Pixel pitch | 5–6 px | Shipped minFeatureSizeMm |
|---|---|---|
| 20 µm | 0.10–0.12 mm | 0.12 mm |
| 35 µm | 0.18–0.21 mm | 0.2 mm |
| 46 µm | 0.23–0.28 mm | 0.25 mm |
Height is a dimension too
Section titled “Height is a dimension too”For a rectangular or trapezoid channel, height can be the limiting dimension
just as easily as width — and the
channel-below-min-feature
check measures both.
Height is built from layers rather than pixels, so it is governed by layer height rather than pixel pitch. In practice a 0.5 mm channel at 50 µm layers is ten layers, which is plenty.
Where height gets interesting is a wide, shallow channel. A 2 × 0.15 mm channel is three layers deep, and cure-through from the layer above will close it.
The printed dimension is not the drawn one
Section titled “The printed dimension is not the drawn one”Expect the printed channel to be narrower than drawn, typically by 0.02–0.08 mm per side depending on resin and exposure.
Two consequences:
Pressure. Fourth-power resistance means a channel 0.05 mm narrow on a 0.4 mm design is about 40 % more resistive than the estimate says.
Balance. In a gradient generator, the error is roughly constant in absolute terms, so it affects narrow branches proportionally more than wide ones. A network that is balanced on paper is less balanced in resin.
The fix is to measure your offset once and design with it. It is repeatable for a given resin and process.
Aspect ratio
Section titled “Aspect ratio”Tall narrow channels are harder than wide shallow ones. The side walls of a deep channel are exposed to scattered light from many layers, and cure-through accumulates.
Below roughly 1:3 width-to-height, expect the channel to close from the sides. If you need a deep channel, a two-piece split helps: half the depth in each piece, bonded to full depth.
Practical floors
Section titled “Practical floors”Rough, for a typical grey standard resin:
| Feature | 35 µm machine | 20 µm machine |
|---|---|---|
| Channel width, reliable | 0.3 mm | 0.15 mm |
| Channel width, with care | 0.2 mm | 0.1 mm |
| Channel height | 0.2 mm | 0.15 mm |
| Wall between channels | 0.4 mm | 0.3 mm |
| Port bore | 0.8 mm | 0.5 mm |
“With care” means measured exposure, a fresh FEP, a pigmented resin, and a test print. Do not design a whole chip at those numbers on the first attempt.
The wall figure is the default for
channels-too-close, and it is
the one people are most often optimistic about.
Measure it yourself
Section titled “Measure it yourself”The single highest-value hour you can spend. See A first test-print protocol for a coupon design and a procedure.