Exposure Dose Calibration for Wax-Filled Photopolymer Resins
Exposure calibration on a wax-filled casting resin cannot be done by eye. The material is heavily filled and largely opaque, so the visible difference between a correctly cured layer and one that is a few percent short of correct dose is close to zero on the build plate — and decisive in the burnout oven three days later. What follows is a measurement-based procedure that produces a dose window and a dose sensitivity, rather than a single number copied from a supplier sheet.
The distinction matters because a single number is only valid for the exact combination of light engine, release film, resin batch, layer thickness and shop temperature it was measured on. Change any one of those and the number is a guess. A window plus a slope survives those changes.
What dose actually is
Dose is energy delivered per unit area at the image plane:
E = I × t
where I is irradiance in mW/cm² and t is exposure time in seconds, giving mJ/cm². For the bulk of the layer, two settings that produce the same product behave nearly identically, which is why "I shortened the exposure and raised the power" usually changes nothing measurable. The exception is the inhibited zone within a few micrometres of the release interface, where oxygen and residence time interact; that zone is thin and mostly matters for adhesion to the previous layer, not for feature size.
The description that is actually useful for a filled photopolymer is the working curve:
Cd = Dp · ln(E / Ec)
Cd is cured depth, Dp is the depth of penetration of light in that specific resin, and Ec is the critical dose below which no gel forms. For a clear unfilled resin Dp can be in the hundreds of micrometres. For a wax- or ceramic-filled casting resin the filler particles scatter light strongly and Dp typically collapses to tens of micrometres. That single change has a consequence most operators miss: the absolute error in cure depth caused by a given percentage error in dose is proportional to Dp. A 10 % dose error shifts cure depth by roughly 0.095 × Dp. If Dp is 60 µm, that is about 6 µm of cure depth — small. If Dp is 250 µm, the same error moves cure depth by 24 µm, which is a whole layer at 25 µm settings.
Two conclusions follow. First, highly filled resins are more forgiving of dose error in the Z direction than clear resins, which is why the same exposure time that undercures a clear resin can overfill a filled one. Second, because the same scattering that limits Dp also spreads light laterally, the XY error grows even where the Z error looks acceptable.
Why the target is a band, not a depth
A layer must cure deeper than its own thickness, otherwise it will not bond to the layer below and the part delaminates during the build or splits during burnout. It must not cure so deep that light bleeds past the pixel boundary and welds features that should be separate. The usable band sits between those two failure conditions and is narrower than most operators assume — on filled casting resins it is often no wider than ±10 % of the nominal dose.
A practical starting target is a cure depth of about 1.2 to 1.4 times the layer thickness, verified by measurement rather than by assumption. At 50 µm layers, that is 60 to 70 µm of cured depth. At 25 µm layers it is 30 to 35 µm, and the same absolute dose error that was tolerable at 50 µm now represents a much larger fraction of the layer.
The coupon, and what to put on it
Do not calibrate on a customer's model. Calibrate on a coupon that measures the three things that matter: positive feature width, negative feature width, and inter-layer bond strength. A working coupon carries all of the following on one plate, printed in the orientation the shop actually uses:
- Four bars nominally 0.500 mm wide and 8 mm long, two along X and two along Y, so directional bias is visible.
- Three holes nominally 1.000 mm diameter, at least 3 mm deep, spaced far enough apart that scattered light from one does not reach the next.
- One thin rib, 0.300 mm thick and 10 mm tall, standing free — this is the delamination indicator.
- One slot pattern with gaps of 0.100, 0.150, 0.200 and 0.300 mm — this is the bleed indicator.
- One step wedge from 0.2 to 1.2 mm in 0.1 mm steps, for a coarse cure-depth sanity check.
Nominal dimensions must be checked in the slicer's output, not in the CAD file, because slicers apply their own compensation and can shift the nominal by tens of micrometres before the printer is involved.
Running the ladder
- Freeze every variable except dose. Same layer thickness as production, same orientation, same support style, same resin batch, same room temperature. Record the temperature; a shop that swings ten degrees between morning and afternoon will produce a ladder that cannot be read.
- Print five coupons at dose multipliers of 0.80, 0.90, 1.00, 1.10 and 1.20 relative to the current setting. Five separate builds, one coupon each, is slower but cleaner than one plate with five zones, because exposure uniformity across the image field would otherwise confound the result.
- Measure green. Blot the surface, wait a fixed interval — ten minutes is a workable standard — then measure each 0.500 mm bar at three points with a micrometer and each 1.000 mm hole with a pin gauge or an optical comparator. Write the values down before post-curing; green dimensions are the only clean read on what the printer did, before shrinkage and secondary cure blur it.
- Post-cure all five identically, and measure again after the parts have stabilised at room temperature for an hour. The post-cure delta is itself a diagnostic: a part that moves more than about 0.3 % during post-cure was not fully cured in the printer.
- Test the rib. Flex the 0.300 mm rib sideways. A correctly cured rib bends and springs back. An undercured rib splits cleanly along a layer boundary, and the split face looks dull and slightly tacky.
- Inspect the slots. The smallest gap that stays open tells you the lateral bleed distance directly. If the 0.100 mm slot closes at 0.90× dose but opens at 1.00×, your bleed is on the order of 50 µm and any dose increase will cost you fine detail.
- Plot deviation against dose multiplier for bar width and hole diameter separately, and fit a straight line to each. The point where bar width equals nominal and the point where hole diameter equals nominal are almost never the same — and that gap is the real calibration problem, not a measurement error.
Reading the two lines
Overcure grows positive features and shrinks negative ones, because the cured region extends past the intended boundary in every direction. So bar width rises with dose while hole diameter falls. If your two fitted lines cross at the nominal dimensions, the process is well behaved and you can sit at the crossing. If they do not cross — bars reading +15 µm while holes already read −15 µm — the light engine is bleeding more than the geometry can absorb, and no single dose fixes it. The remedies are geometric or optical: revise the slicer's XY compensation so that bars are drawn narrow and holes drawn wide, reduce the bleed by improving collimation or contact, or accept a different feature-size tolerance class for holes than for bars.
The slope of those lines is worth recording in its own right. It tells you how much a future dose change will cost. A slope of 2 µm per 10 % dose on a 0.500 mm bar means you have room to trade surface quality against detail; a slope of 8 µm per 10 % means every adjustment is a re-qualification.
Failure modes and how to tell them apart
| Observation | Under-cured | Over-cured |
|---|---|---|
| Bar width, green | Below nominal, often by 10–30 µm | Above nominal by a similar margin |
| Hole diameter | Above nominal | Below nominal; small holes close entirely |
| Surface after blotting | Dull, resin continues to weep from the surface | Glossy and dry, with a hard skin |
| Layer boundary | Visible, and the rib splits along it | Invisible; the rib snaps instead of splitting |
| Fine slots | Open, sometimes wider than drawn | Bridged or welded shut |
| Post-cure movement | Large and non-uniform between thin and thick sections | Small, but accompanied by curl on thin flat sections |
| Support removal | Supports tear away with a smeared contact point | Contact points chip, leaving craters |
| Burnout residue | Higher than usual, sometimes with visible carbon in the sprue | Normal |
| Cast result | Rough surface from incomplete pattern removal, gas porosity | Lost detail, rounded edges, occasional shell cracking from expansion stress |
The burnout and cast rows are the ones that matter commercially, and they are also the slowest feedback. This is exactly why the coupon exists: it gives the same information in ninety minutes instead of three days.
When the window moves
A calibration is a snapshot of a specific state. The window shifts when the resin batch changes, when the light source ages, when the release film is replaced with a film of different thickness or oxygen permeability, when layer thickness changes, and when shop temperature departs meaningfully from the calibration temperature. Re-run the ladder after any of those events. If the shop cannot control temperature within a few degrees, calibrate at the temperature the machine actually runs at, not at the temperature the datasheet quotes.
Resin age matters in a way that is easy to miss: settled filler and slowly reacting photoinitiator both reduce the effective dose delivered to the monomer, and the symptom is a batch that suddenly needs more exposure than the previous one. If a fresh batch requires a dose change of more than about 10 % to hit the same dimensions, suspect the material or its storage rather than the machine.
What to write down
A calibration record that is actually useful contains: date, resin product and batch, layer thickness, nominal exposure time and irradiance, measured cure depth from the step wedge, green bar width and hole diameter at each dose multiplier, post-cure dimensions, slot bleed distance, room temperature, and the two fitted slopes. With those fields, the next operator reproduces the setting instead of rediscovering it, and drift in any one number points at a specific cause.
The output of calibration is not a number to type into the slicer. It is a band within which the process is stable, a sensitivity that tells you how far you can push it, and a record that makes the next adjustment an engineering decision instead of a guess.
Ready to discuss your project?
Tell us what you need — our engineering team replies within 24 hours.