Purpose and scope

This guide provides a validation sequence for openly documented, 3D-printed laboratory tools that measure, hold samples, route fluids, position optics or automate an experiment. It is not a medical-device approval guide and cannot turn a research prototype into a clinically suitable product. The aim is to demonstrate that a specific assembly is fit for a defined laboratory task and produces repeatable, traceable results.

1. Define the measurand or controlled function

Validation starts with intended use, not the slicer profile. State what the device must do, over which range, with which sample, environment and acceptance limits. A syringe holder may depend on transferred volume and leakage; an optical mount on positional stability and background scatter; an incubation cell on flow and temperature. 'It works' is not a measurable requirement.

Assign a verification method and decision rule to every requirement. If volume error must stay within ±2%, record the reference, number of repeats, fluid and temperature. The decision rule must include measurement uncertainty; a result extremely close to the limit should not automatically pass.

2. Open source means more than publishing an STL

The Open Source Hardware Association requires source in the preferred form for modification. For a mechanical project, that means editable CAD rather than only a tessellated export. A complete package may also include a bill of materials, tolerances, drawings, firmware, schematics, control software, licence, assembly instructions and tests. CERN OHL 2.0 offers three variants with different reciprocity levels; name the selected licence explicitly.

Version every release. Record the CAD version, source commit, bill of materials, firmware, slicer and profile version, and manufacturing date. Without that chain, another laboratory cannot know whether it tested the same configuration.

3. Build a risk breakdown

Decompose the tool into functions and failure modes. Mechanical risks include creep, fracture and backlash; fluidic risks include leaks, dead volume and bubbles; optical risks include absorption, fluorescence and misalignment; electrical risks include noise, ground loops and overheating; biological risks include adsorption, cleanability and cross-contamination. Risk ranking determines test depth.

Evaluate material against the actual medium and cleaning process. PLA, PETG, ABS, resins and other polymers differ in chemical resistance, water uptake, thermal stability and sterilisation compatibility. Layer lines and internal porosity can retain liquid or contamination. A supplier datasheet is a starting point; test the printed part made with the actual process.

4. Qualify the manufacturing process

Record the machine, nozzle or optics, material lot, drying, orientation, layer height, walls, infill, exposure, supports, post-curing and cleaning. Measure designated critical dimensions across multiple parts and builds. The goal is not to inspect every surface point but to establish stability of the characteristics that control function.

Choose sample size according to risk and the decision being made. A first internal jig may need few repeats; a published measurement instrument intended for multiple laboratories should cover more machines, operators, days and material lots. Report failed parts and exclusion reasons so successful specimens do not create a misleading picture.

5. Calibrate the complete measurement chain

NIST describes metrological traceability as a property of a measurement result linked to a reference through a documented, unbroken calibration chain, with every step contributing uncertainty. A calibrated sensor alone is insufficient. Printed fixture, mechanical transmission, transducer, amplifier, digitiser, software and analysis equation form one measurement system.

Calibrate across the real range and environment. Test zero, linearity, hysteresis, repeatability, short- and long-term drift, temperature dependence and, where relevant, frequency response. Preserve raw data, fit, residuals and the valid range of the calibration.

6. Estimate measurement uncertainty

List uncertainty sources: reference uncertainty, resolution, repeatability, temperature, geometry, sensor drift, sample handling and data processing. Some terms come from repeated observations; others from certificates or bounded specifications. Document the combination method and coverage attached to the final result.

Uncertainty makes comparisons meaningful. A 1% difference between two devices does not prove superiority when each has 3% expanded uncertainty. If the acceptance band is narrower than the system can resolve reliably, improve the design or method.

7. Verify intended use

After component tests, run the complete system with real or representative samples. Compare it with an accepted commercial or independent reference method where possible. Test range extremes, foreseeable operator errors, assembly cycles, cleaning, sustained use and whether failures are detectable.

Software needs test cases, version locks and logs. A spreadsheet, microcontroller firmware or analysis script can fail as readily as mechanics. Use known inputs and expected outputs, test limits, and store processing parameters with each result.

8. Reproduce the build in another laboratory

Repeatability in one laboratory does not establish that an open design is reproducible elsewhere. Ask at least one independent group to build from the documentation and measure the key characteristics with the same protocol. Record where instructions required interpretation or correction.

A substituted component creates a new configuration. Nominally equivalent sensors, fasteners or filaments may behave differently. The bill of materials should name manufacturer, part number, allowed alternatives and the test needed to accept an alternative.

9. Change control and revalidation

Define revalidation triggers: CAD revision, material or supplier change, new printer, profile update, firmware or library update, different sterilisation, or changed measurement range. Risk determines whether the retest is partial or complete. Keep each result visibly linked to the version it supports.

Use a release checklist for every unit or lot. It can include visual inspection, critical dimensions, leak test, zero, reference check, software version and labelling. Set calibration and maintenance intervals from drift and usage evidence rather than an arbitrary calendar number.

10. What to publish

A minimum package includes intended use and exclusions; editable source files; licence; bill of materials; manufacturing and assembly instructions; calibration and test protocols; raw and processed data; uncertainty budget; acceptance criteria; known limitations; version history and an issue-reporting channel. A good photograph and low material cost are not substitutes.

Quick decision sequence

1. Intended use and measurable requirements. 2. Risk and material-compatibility analysis. 3. Versioned, modifiable source package. 4. Manufacturing process and critical dimensions. 5. Complete-chain calibration. 6. Uncertainty and decision rule. 7. System test in the real environment. 8. Independent reproduction when distributing the design. 9. Release, maintenance and change control.

Editorial conclusion

3D printing can democratise laboratory instrumentation by making mechanics locally adaptable. Scientific value, however, comes from the documented measurement chain rather than the printed object. An open licence enables modification; calibration, uncertainty, reproduction and change control make the result dependable.

Sources

- Umeå University and DiVA — Daniel Nilsson's doctoral work on open biophysical instrumentation, 2026. - Open Source Hardware Association — Open Source Hardware Definition 1.0 and sharing practices. - CERN — CERN Open Hardware Licence Version 2. - NIST — Measurement Uncertainty; Metrological Traceability FAQ.