How Are 3D Printed Products Rated? A Practical Guide
A rating is not one number. It is a set of measurements taken against a drawing: layer height, dimensional deviation, surface roughness, material state and inspection coverage. This guide shows how we rate custom 3D printed parts at GreatLight and what you should ask a supplier to show you.

Key takeaways
What a rating actually measures on a 3D printed part
When an engineer asks how are 3D printed products rated, the honest answer is that no single score exists. A rating is a set of measured values compared against the drawing you supplied. Four values carry most of the weight: layer height, dimensional deviation on critical features, surface roughness, and material verification.
Layer height is the first number to pin down. FDM machines run from about 0.1 mm to 0.3 mm per layer. SLA and DLP resin printers go finer, commonly 0.05 mm to 0.1 mm. A 0.05 mm layer resolves small text and thin walls; a 0.3 mm layer prints three to five times faster and leaves stair-stepping you can feel with a fingernail.
Dimensional deviation is where most arguments start. On a well-tuned FDM machine, expect roughly ±0.3 mm on a 100 mm span, and worse on tall parts because the bed cools unevenly. Resin printers often hold ±0.1 mm on small features. Neither figure is a guarantee until you measure the part on a granite table.
Surface roughness is usually reported as Ra. As-printed FDM surfaces sit around Ra 3.2–12.5 μm depending on layer height and orientation. Resin parts are smoother out of the machine but need support removal, which leaves witness marks. Only post-processing, sanding or tumbling pushes a printed surface toward Ra 1.6 μm.
- 1Pin the drawing firstA rating without a tolerance callout on the drawing is an opinion.
- 2Measure critical features onlyDatum faces, bores and mating surfaces decide fit.
- 3Record the instrumentCalipers, micrometers and CMMs give different confidence levels.
How layer height, orientation and material change the rating
Orientation changes strength more than any slicer setting. A part loaded along the layer plane can lose 30–50% of its tensile strength compared to the same material loaded across layers, because the bond between layers is weaker than the filament itself. Rotate the part so the highest load runs in-plane, then re-check the support marks it leaves.
Material grade is the second lever. ABS, PC, PMMA, POM, PA, PEEK and carbon fibre blends all print differently. PEEK needs a heated chamber above 200 °C; most service bureaus will not run it. If your drawing calls for a specific grade and the shop substitutes a look-alike, the rating is void even if the part looks right.
Shrinkage is the quiet failure. Large flat ABS parts warp as they cool, so a 200 mm plate that measured flat on the bed can bow 0.5 mm overnight. Design ribs, keep walls uniform, and let the part cool in the chamber before removal.
When a printed part has to hold ±0.005 mm or a fine Ra 0.8–1.6 μm finish, printing alone will not get there. That is the point where we machine the printed blank, or switch the part to 5-axis CNC. We run 16 simultaneous 5-axis centers and 127 high-precision CNC machines across 7,600 m² in Dongguan, so the transition from print to cut happens under one roof.
- 1Load in-planeRotate the model so stress does not pull the layers apart.
- 2Match the gradeSubstituting resin or filament voids the rating.
- 3Control coolingUniform walls and slow cooling cut warpage on flat parts.
Why 3D printed products rated for fit still need machining
Printers build geometry; they do not hold metal-class tolerance. A printed bracket can be dimensionally right and still fail on a bearing bore, a threaded hole or a sealing face. Those features need a cutting tool, not a nozzle.
A common workflow: print the housing to near-net shape, then face the mating surface and ream the bores on a 3-axis mill. The part keeps the printed internal channels that would be expensive to machine, and gains the ±0.005 mm fits that make it assemble. Cycle time on the finishing cuts is usually minutes, not hours.
The same logic applies to prototypes in aluminium 6061, 7075, 316L stainless or Ti-6Al-4V. If the end use is a machined part, prototype it in the same alloy so your test data transfers. Printing in resin gives you geometry feedback, not mechanical feedback.
Ask the supplier which features were printed and which were cut. If nobody can answer, the rating on the certificate describes a different part than the one in your hand.
- 1Print for geometryInternal channels and organic shapes are cheap to print.
- 2Cut for fitsBores, threads and sealing faces need a machine tool.
- 3Prototype in the final alloyResin test data does not transfer to metal parts.
How to rate a 3D printed part in 6 steps
Follow this order on the shop floor
- 11. Read the drawing before the partList every dimension with a tolerance and mark which ones are critical to function. If the drawing has no tolerances, agree on them in writing before printing.
- 22. Record the build parametersWrite down layer height (0.05–0.3 mm), nozzle or laser spot size, material grade and batch, print orientation, and support strategy. These five items explain most deviations later.
- 33. Measure critical features, not the envelopeUse calipers for reference, micrometers for outside diameters, pin gauges for holes, and a CMM or height gauge on a granite table for datums. Measure at 20 °C; printed plastic moves with temperature.
- 44. Check surface roughness the right wayCompare against a Ra comparator or run a portable roughness tester across the print direction. Note whether the reading is as-printed or after sanding, tumbling or bead blasting.
- 55. Verify material and bondingConfirm the resin or filament grade against the certificate, check for layer delamination with a light twist test, and look for voids on a cut cross-section on one scrap part.
- 66. Write the report and set the re-check intervalLog the values, the instrument, the operator and the date. For production runs, re-measure the first part of every batch and one part per shift, and keep the data with the shipment.
Typical values by process and feature
Use these ranges to set realistic expectations before you quote
| Feature or process | Typical value | How to verify | When it is not enough |
|---|---|---|---|
| FDM layer height | 0.1–0.3 mm | Slicer log plus visual step check | Small text and thin walls below 0.8 mm |
| Resin layer height | 0.05–0.1 mm | Slicer log plus caliper on a step | Large flat panels that warp |
| FDM dimensional spread | About ±0.3 mm per 100 mm | Calipers or CMM on datums | Bearing bores and threaded holes |
| Resin dimensional spread | About ±0.1 mm on small features | Micrometer or pin gauge | Mating faces under load |
| As-printed roughness | Ra 3.2–12.5 μm | Ra comparator or portable tester | Sealing faces and sliding contact |
| Machined finish | Ra 0.8–1.6 μm | Profilometer trace on the face | Optical and cosmetic surfaces |
| Tight machined tolerance | ±0.005 mm | CMM with temperature control | Non-critical cosmetic geometry |
| Inspection coverage | 100% before shipment | Signed report with instrument list | Nothing on a functional part |
The rating is only as good as the drawing behind it
Fix the tolerances and the material grade first, then measure the critical features and keep the data. If a printed feature has to hold a metal-class fit, machine it.
Questions engineers ask about ratings
Is a 3D printed part rated the same way as a machined part?
No. A machined part is rated against a tolerance and a surface callout that the process can hold directly. A printed part is rated against the same drawing, but the achievable values are looser, so the drawing usually needs a separate set of tolerances for printed features.
Many teams keep two drawings: one for the printed prototype and one for the production machined version. That avoids arguing about a ±0.05 mm callout on a feature that will only ever be printed.
What tolerance can a supplier realistically promise on a print?
For FDM, about ±0.3 mm per 100 mm is a fair working number, and it gets worse on tall builds. For resin, roughly ±0.1 mm on small features is common.
Anything tighter should be machined after printing. We hold ±0.005 mm on our CNC equipment, so a printed blank can be finished to that level on the features that matter.
Does layer height affect strength or only appearance?
Both. Thinner layers usually bond better and raise tensile strength slightly, but orientation matters far more. A part loaded across its layers can lose 30–50% of its strength regardless of layer height.
Set orientation first, then choose the finest layer height your cycle time allows.
How do I check a printed part without a CMM?
Use a granite surface plate, a height gauge, micrometers and pin gauges for the critical features. Compare against the drawing at 20 °C.
For surface finish, an Ra comparator set is enough for a pass or fail decision. Send the part out for a CMM report when the feature is functional and the risk of a bad fit is high.
When should I stop printing and switch to CNC?
Switch when the part carries a tolerance below about ±0.05 mm, when it has to seal or slide, or when the material is a metal alloy that has to match production.
Printing still helps for the geometry. We print the near-net shape, then machine the datums, bores and sealing faces on the same project.
What paperwork should come with a rated part?
A dimensional report listing the measured features, the nominal and actual values, the instrument used, and the date. Add a material certificate for the resin or filament batch.
GreatLight inspects 100% of parts before shipment and provides reports on request, with raw material checks, in-process monitoring and final inspection recorded.
Send us the drawing and we will rate it
Upload your model and tolerances. We return a quotation and a free DFM analysis within 12 hours, with no minimum order quantity.
12-hour quoteNo MOQ100% inspection