---
title: Same 3D file, two prototype shops, two different results, why?
lang: en
source: https://mindsprt.dev/en/knowledge/research-brief-uv-curing-3d-printing-process-parameters/
---

# Same 3D file, two prototype shops, two different results, why?

*Mai Strategy Lab · 11 min read · 2026-08-03*

> You send the same STL to two vendors and the samples come back with mismatched dimensions, surfaces, and hardness. The file isn't the problem; the parameter chain behind exposure and curing is, and it never makes it onto the quote sheet. This piece breaks down four key variables in UV-curing 3D printing and gives you a question checklist you can put straight into your RFQ and acceptance criteria

**Quick answer:** You send the same STL to two vendors and the samples come back with mismatched dimensions, surfaces, and hardness. The file isn't the problem; the parameter chain behind exposure and curing is, and it never makes it onto the quote sheet

## Overview

You hand the same STL to two prototype shops. A week later two samples sit on your desk that look alike but measure apart: one holds a crisp edge with a matte surface, the other has softer corners and a slightly tacky feel when your nail catches the surface. You go back to the file. The file is fine.

This plays out every day on packaging structural samples, product appearance models, and retail display pieces. The snag is that we carry over flat-printing thinking and treat the file we hand off as the spec. In UV curing, the file only describes geometry. What the sample ends up being is shaped by exposure dose, cure rate, material formulation, and the machine's own geometry.

So if you're a designer or buyer who never touches the machine, what should you be asking and verifying to actually manage this?

## Why "same file" doesn't mean "same part"

Because in UV curing systems, geometric accuracy is a function of process parameters, not of the file.

A parameter optimization study on a UV-curing 3D printing system running a UV resin / hydroxyapatite particle composite used a Box-Behnken design to run 27 experiments, sweeping four variables together, nozzle diameter, print speed, layer height, and nozzle-to-bed distance. The best combination landed at Nozzle 2.2 mm Speed 32.52 mm/s Layer height 1.2 mm Nozzle distance 0.5 mm, with a dimensional error of 3.08% [1]

That 3.08% is the residual error measured under optimal parameters, not the number from a broken run. On a 100 mm structural sample, that's roughly a 3 mm swing. If your snap fits, tabs, or alignment holes are toleranced tighter than that, and you later hit "passed prototyping, won't go together in production," the first thing to revisit is whether your tolerance left room for this 3%.

The same study spells out the original motivation for the process: DIW extrusion cures too slowly, so the part shrinks first and then cracks, which is why UV light was brought in to speed up curing [1]. Translated into shop-floor language, shrinkage and cracking aren't material defects; they're the time gap between cure rate and stacking speed. Same physics as a spot UV varnish piled too thick on a screen-printed substrate that splits at the crease line during folding. Two versions of the same story: the material is asked to take stress before it's stable.

## What does exposure actually affect? More than "dry or not"

Most people judge exposure only by whether the surface feels dry. In practice, exposure also drives the mechanical properties of the finished part.

If it doesn't feel tacky, that usually only means the surface layer hit a high enough conversion rate; the crosslinking deeper in may still be incomplete. A study on photopolymer resin in stereolithography looked directly at the effect of UV post-curing on mechanical properties [5] — post-curing shifts the part's strength, so it can't be treated as a last-minute shine under the lamp.

Cure degree is something you can actively control. One study used cure-degree control of a hybrid UV curing resin to enable continuous printing [6], showing that "how far cured" is a knob in the engineer's hand, not a binary wet/dry. Another route uses projected UV-resin curing to achieve self-supported printing [3], letting the cure sequence substitute for part of the support structure.

Once cure degree is tunable, vendors will tune it differently. Shop A compresses post-cure to 10 minutes to hit the deadline; Shop B runs the full 30. Both samples can pass visual inspection, but drop tests, clamping, and long-term dimensional stability will diverge. On the quote sheet, that column is usually empty.

## Why is material formulation a hidden variable?

Because fillers change how deep light travels through the resin, which rewrites every exposure parameter downstream.

The composite formulation in that study was hydroxyapatite at 30% w/v with resin at 100% v/v [1]. Add solid particles and light gets scattered and absorbed, penetration depth drops, and the same exposure setting that was just right on neat resin can leave a high-filler formulation undercured. The Si₃N₄ UV resin curing-behavior optimization study treats curing performance in ceramic-filled systems as a problem that needs its own optimization pass [4].

Direct read for the design side: "switch to a different color" or "switch to a harder material" is never free. Anything that changes the resin's optical behavior, added pigments, added fillers, changed transparency, should be treated as a change that triggers re-prototyping, not a one-line swap of the SKU. UV curing resin can even be designed to produce continuous color shifts in elastomers [2], which cuts the other way too: color is bound into the resin chemistry, not a coating slapped on after.

## What can someone who never touches the machine actually verify?

The RFQ has to spell out critical dimensions, post-cure records, and batch consistency. All of this can be checked without ever approaching the printer.

Four questions to ask when sending out the RFQ

・What's this machine's layer height and nozzle diameter setting? — the study confirmed both directly affect dimensional error [1].

・How long is post-curing, and what light source? — post-curing shifts mechanical properties in a verified way [5], so it has to be a recorded parameter, not a feel thing the operator calls.

・Is this formulation neat resin or filled? At what filler ratio? — fillers change curing behavior and need their own optimization [4].

・When color or material changes, do you re-run the parameters?

Three things to measure at acceptance

・Pick 2–3 functional dimensions (tabs, holes, wall thickness) and measure them, don't eyeball the surface. Use the ~3% error range as a mental baseline [1] and check first whether your tolerance is actually tighter than that.

・Get 3+ samples from the same batch. What you're comparing is the spread between them. A single good sample tells you nothing about process stability.

・Surface tackiness, fingernail marks that linger, dimensional shift after a week of sitting, these three are common visible signs of under-curing (this is frontline experience, not a literature conclusion).

I usually call this the Mai Strategy three gates for sending out print: Gate 1, ask about parameters (who sets them, where are they written down); Gate 2, check dispersion (don't compare the best sample, compare the worst); Gate 3, track changes (any material or color swap counts as a new job). All three gates are doing the same thing: getting invisible process parameters written into the contract so they're traceable later.

## Closing: Lock down tolerance before you talk lead time

If you can only do one thing, nail down your critical-dimension tolerance before sending the RFQ and make sure the vendor knows it. Because in UV-curing 3D printing, the dimensional error after systematic optimization still sits in the 3% range [1] — which means tolerance is the design side's responsibility, not the vendor's yield problem. If you can't write the tolerance down, nobody can be held accountable at acceptance.

Scope boundary: the numbers above come from a specific DIW + UV curing system and a specific composite formulation [1] and can't be applied directly to other UV curing routes like SLA, DLP, or MSLA, those systems have different accuracy ranges and different error sources. The 3.08% doesn't transfer; what transfers is the management practice: treat parameters as variables to be controlled, leave a record of post-curing, and run change control on material swaps. Also, if all you need is a volumetric check or a shape-conversation piece, the whole acceptance rig can be skipped, how much effort to put into accuracy depends on what this sample is going to do next.

## Key takeaways

UV-curing 3D printing part accuracy is set by process parameters, not the file, the same STL under different parameters produces different results [1].

Even after systematic optimization, dimensional error still hits 3.08%, so critical-dimension tolerance has to be defined on the design side first [1].

Post-curing isn't a finishing flourish; it's a process step that changes mechanical properties and should be recorded and compared [5].

Adding fillers or pigments shifts the resin's optical behavior and curing, material or color changes should be treated as a new job requiring re-prototyping [4].

Acceptance looks at dispersion across multiple samples from the same batch and measured functional dimensions, not visual inspection of a single part.

## Further thinking

For print manufacturing, UV-curing 3D printing is pushing "parameters as spec" into the open. Print shops used to sell process steps; what they'll have to sell next is reproducible parameter sets with matching tolerance commitments, which means exposure dose, post-curing duration, and material lot all need to live on the work order instead of in the operator's head. On the design side, DfM is shifting from form-feasibility toward tolerance awareness, being able to tag functional vs. cosmetic dimensions at the modeling stage is what decides how many rounds of back-and-forth you'll have after the RFQ goes out. The entry point for AI is pretty clear: parameter optimization is fundamentally a multi-factor response-surface problem [1], and that kind of design-of-experiments and surrogate-modeling work is machine learning's home turf. A low-risk starting point would be "use historical prototype data to back out parameter recommendations." On the SaaS side there's a gap nobody's filled: there's no standard format that carries "this sample was made with these parameters." If parameter records, batch traceability, and acceptance measurements could be tied together as an exchangeable data structure, prototype quality would stop being each vendor's private kung fu and become a comparable industry metric. The open question is how a vendor that's willing to publish its parameters avoids the suspicion of tech leakage, that's both a technical and a business-trust problem.

## References

[1] Hoten, Tontowi, Herianto (2027). [Optimization of Process Parameters in a Newly Developed UV-Curing 3D Printing System for Printed (UV Resin/HA Particle) Composites](https://doi.org/10.5829/ije.2027.40.01a.16). International Journal of Engineering. DOI: 10.5829/ije.2027.40.01a.16

[2] Gao Y., Jiang Q., Liu Y. (2025). [A 3d Printing Uv-Curing Resin that Enables Continuous Color Change of Elastomers](https://doi.org/10.2139/ssrn.5093607). DOI: 10.2139/ssrn.5093607

[3] Mizuno Y., Pardivala N., Tai B. (2018). [Projected UV-resin curing for self-supported 3D printing](https://doi.org/10.1016/j.mfglet.2018.09.005). Manufacturing Letters. DOI: 10.1016/j.mfglet.2018.09.005

[4] Cao C., Wang C., Zhao Z. (2019). [Optimization of Curing Behavior of Si<sub>3</sub>N<sub>4</sub> UV Resin for Photopolymerization 3D Printing](https://doi.org/10.1088/1757-899x/678/1/012013). IOP Conference Series: Materials Science and Engineering. DOI: 10.1088/1757-899x/678/1/012013

[5] Bouchareb S., Doufnoune R. (2025). [Effect of UV post-curing on the mechanical properties of photopolymer resin in stereo-lithographic 3D printing](https://doi.org/10.1016/j.matlet.2025.138587). Materials Letters. DOI: 10.1016/j.matlet.2025.138587

[6] Kang X., Li X., Li Y. et al. (2021). [Continuous 3D printing by controlling the curing degree of hybrid UV curing resin polymer](https://doi.org/10.1016/j.polymer.2021.124284). Polymer. DOI: 10.1016/j.polymer.2021.124284

## FAQ

### Why do different vendors produce different results from the same 3D file?

Because UV-curing 3D printing part accuracy depends on process parameters, not the file itself. Nozzle diameter, print speed, layer height, and nozzle-to-bed distance all affect dimensional error, and even after systematic optimization the error still hits 3.08% [1].

### Roughly how big is the dimensional error in UV-curing 3D printing?

In a DIW + UV curing system study that ran 27 experiments under a Box-Behnken design, the dimensional error under the best parameter combination was 3.08% [1]. That number applies only to that specific system and material formulation; SLA, DLP, and other routes have different error scales.

### Can post-curing be skipped?

Not recommended. Studies have shown that UV post-curing changes the mechanical properties of photopolymer resin [5], which means post-curing time and light source are process variables that shift final-part strength. They should be recorded and compared across vendors instead of left to the operator's feel.

### Does switching resin color require re-prototyping?

Yes. Pigments and fillers change the resin's optical properties and curing behavior, and ceramic-filled systems need their own curing parameter optimization [4]. Color or material changes should be treated as engineering changes that trigger re-prototyping.

### How should a designer who doesn't know the machines accept 3D printed samples?

Measure three things: pick 2–3 functional dimensions (tabs, holes, wall thickness) and measure them rather than eyeballing the surface; pull 3+ samples from the same batch and compare their spread; and watch for surface tackiness, lingering fingernail marks, and dimensional shift after a week of sitting, early signs of under-curing.


---

> HTML version: https://mindsprt.dev/en/knowledge/research-brief-uv-curing-3d-printing-process-parameters/
> MINDS — 麥思印刷整合有限公司 · https://mindsprt.dev