Identify the dimensional error before choosing a correction

Hello everyone. An undersized part, a tight hole and lifted corners are different problems. Unit errors, proportional shrinkage, constant dimensional offsets and warping require different responses. To achieve accurate dimensions with a 3D printer, first establish where the error begins: in CAD, file transfer, slicing, printing or post-processing.

Rule out a millimetre and inch mix-up first

Standard STL files do not specify physical units. If coordinates created in inches are interpreted as millimetres, the imported model can appear 25.4 times too small. The reverse interpretation can enlarge it by the same factor. Confusing centimetres or metres with millimetres can produce factors of 10 or 1000. With discrepancies this large, investigate unit handling before material shrinkage.

3MF can carry unit information; the core specification defaults to millimetres. Even so, compare the imported X, Y and Z dimensions with CAD. Check automatic fit-to-bed operations and any previous scaling. A displayed scale of 100 percent does not, by itself, establish that the source units were interpreted correctly.

Does the exported geometry match the CAD design?

For parts designed to specific dimensions, define the critical features first: outside dimensions, hole diameters, distances between hole centres and mounting faces. Mesh formats such as STL approximate curves with triangles. A coarse export can turn a circular hole into a polygon. Inspect mesh refinement and the slicer preview before printing; enlarging the model later will not restore missing curve detail.

For complex geometries, checking only the overall bounding box is insufficient. A model with correct outside dimensions may still contain an incorrect channel, thin wall or hole position. Choose export refinement to suit the part; an unnecessarily large file does not guarantee dimensional accuracy.

Is the error proportional, constant or a change in shape?

Proportional error: Suppose nominal lengths of 20, 60 and 100 mm measure 19.90, 59.70 and 99.50 mm along the same axis. Each is approximately 0.5 percent undersized. This suggests a proportional effect, but you must still establish whether it comes from material shrinkage or the movement scale.

Constant offset: If those targets measure 20.20, 60.20 and 100.20 mm, the error is approximately 0.20 mm in every case. Investigating contour and extrusion effects is more useful than immediately applying a global percentage. A constant 0.20 mm error represents a different percentage on a small part than on a large one.

Warping or instability: Lifted corners, unexpectedly unequal diagonals or different measurements from repeated prints cannot be explained by one scaling percentage alone. Investigate mechanical play, bed adhesion and thermal conditions first. In industrial production, the target is a repeatable result within a defined tolerance, rather than a single successful part.

How can you make the measurements reliable?

Measure after the appropriate post-processing and temperature stabilisation. Check the caliper zero, clean measuring faces and applied force. A 0.01 mm display increment does not mean the instrument is accurate to 0.01 mm. Small bores or tight tolerances may require suitable pin gauges, a micrometer or another measurement method.

Start with a test piece containing several lengths, outside dimensions and holes. Record X and Y separately and measure Z independently. Several repetitions help reveal random variation. Use the same material, orientation, profile and finishing conditions throughout. Thin layers or high resolution do not replace measurement.

When is scaling appropriate, and how is it calculated?

If several measurements show a consistent proportional deviation and the process is stable, use this relationship for a trial correction: New scale percentage = current scale percentage × target dimension / measured dimension. For example, a 50.00 mm target printed at 100 percent repeatedly measures 49.80 mm. The calculation is 100 × 50 / 49.80 = approximately 100.40 percent.

This is an initial correction for that example, not a universal printer or material setting. Print and measure again afterwards. Applying different X, Y and Z factors can also change the geometry; a circle can become an ellipse. Use separate axis factors only when supported by measurements, and save them with the relevant material profile.

Why can holes be tight when the outside dimensions are correct?

Hole dimensions can be affected by mesh resolution, extrusion, line placement, cooling and print orientation. Suppose a part's outside width correctly measures 40 mm, but its nominal 10 mm hole measures 9.7 mm. Enlarging the whole model by approximately 3 percent also increases the outside width to about 41.2 mm. Applying global scaling to a local hole error can spoil dimensions that were already correct.

Some slicers provide separate hole compensation and outer contour compensation; OrcaSlicer is one example. Check your software's sign convention and whether its value refers to a boundary offset or a diameter change. Geometrically, moving a circular boundary outwards by 0.10 mm increases its diameter by 0.20 mm. That does not mean every software interface expects the same input value.

Assembly clearance must also be designed. Do not assume a 10 mm shaft will slide freely in a nominal 10 mm hole. A 0.20 mm diametral clearance gives a 0.10 mm radial gap on each side in an ideal concentric assembly; this is a geometric example only. Actual clearance depends on loading, temperature, movement and tolerances. Avoid weakening part durability by leaving excessively thin walls while improving the fit.

What if only the base is wide, or the Z dimension is wrong?

If the bottom layers bulge out while the upper body is correct, the problem may be elephant foot, a local expansion near the bed. Check the first-layer distance, bed setup and temperatures, then try elephant foot compensation if necessary. Shrinking the whole part also shrinks the already-correct upper section.

For a Z error, assess the first layer, layer heights, the model's final sliced layer and the Z mechanism separately. A nearly constant height shortfall across different models is different from an error that grows with height. Removing material from the base during surface finishing can also change the measured height.

Separate motion calibration from extrusion adjustments

Axis movement scale is related to belt and pulley geometry, screw lead and driver configuration. Resolve mechanical play, loose pulleys, misalignment or lost motion first. Klipper's documentation advises against setting X/Y/Z movement values using a simple measure-and-adjust procedure; it explains determining them from the hardware geometry where necessary.

Extruder feed calibration and the slicer extrusion multiplier are separate settings. Excess flow can thicken walls and narrow holes. However, arbitrarily reducing flow just to open a hole can cause gaps or poor bonding between layers. Preserve mechanical properties alongside dimensional accuracy.

How do material and orientation affect dimensions?

Printing materials do not cool and shrink identically. Global scaling alone will not remove bending in a long, thin ABS part. Appropriate environmental conditions, bed adhesion and the manufacturer's temperature and cooling profile matter. Filament properties and moisture condition can also affect flow consistency. Revalidate a saved correction when changing to another spool.

Changing orientation changes how holes and thin surfaces are formed. Support structures can leave marks or deformation on critical mating faces. In resin printing, assess orientation, supports and exposure together with subsequent washing, drying and recommended post-curing. The same compensation value or processing sequence does not apply to all 3D printing technologies.

Define finishing and acceptance checks in advance

Drilling, reaming, sanding and coating alter the final dimensions. If a critical hole will be machined afterwards, plan the machining allowance and choose a suitable material and process. Acceptance dimensions for the finished part must be distinguished from intermediate measurements. A resin part measured before processing is complete may not represent its final service condition.

Record the model revision, material and spool, printer, orientation, profile, finishing steps and critical measurements together. Validating a small sample before a large one-off print can reduce wasted time and cost. Change one measurable variable at a time so that you can understand its effect.

A practical sequence to follow

Verify the units and CAD dimensions first. Then inspect the export and slicer preview. Print under stable conditions and measure several lengths and holes. Separate proportional errors from constant offsets and changes in shape. Apply the appropriate correction and measure again. The success criterion is a finished part that repeatedly meets the required dimensions and function, rather than a particular percentage on screen.