Before a custom 3D printing batch starts, define what must pass, how it will be checked, how often it will be checked, and what happens when a part does not pass. At minimum, the release baseline should identify the approved file revision and material, critical fit or function, dimensional and cosmetic limits, inspection method, sampling level, packaging requirements, and the disposition path for borderline or nonconforming parts.
Do not ask for “good quality” or “production quality” and assume the shop sees the same standard you do. A useful QC plan turns buyer expectations into observable pass conditions without forcing tight measurements onto features that do not affect the job.
Minimum batch-release baseline
- controlling file name, revision, units, quantity, variants, and material;
- critical-to-function features and the real pass test for each;
- acceptable appearance by surface or viewing zone;
- inspection method, part condition, gauge or mating reference, and sampling frequency;
- rules for rework, replacement, concession, partial release, and records; and
- packaging, labeling, lot separation, and evidence required before shipment.
Start with the failures the buyer cannot accept
Acceptance criteria should begin with use, not a generic list of print defects. Ask what would stop installation, create a safety or reliability problem, cause a customer return, or make the batch unusable. Those outcomes define the critical features. Surface marks and minor dimensional variation may matter too, but they should not outrank the interface that decides whether the part works.
A mounting bracket may have three different acceptance lanes: the hole pattern must align to the fixture, the loaded arm must not crack in the approved orientation, and the customer-facing face must not show support damage. Each lane needs a different check. One caliper measurement cannot prove all three.
Write the failure in plain language first: “the cable must pass without abrasion,” “the latch must close by hand,” or “the insert must resist the specified installation torque.” Then decide whether a dimension, functional gauge, mating part, visual standard, or load test is the cleanest evidence.
Rank requirements by consequence. A critical requirement stops use or creates unacceptable risk. A major requirement reduces function, fit, or customer acceptance. A minor requirement is still worth controlling but does not make the part unusable. The names matter less than the response: the shop and buyer should know which failure stops production, which failure holds a subgroup, and which issue can be documented without rejecting sound parts.
Build one controlled release table
A short table is usually easier for the buyer, operator, packer, and receiving team to use than requirements scattered across emails. Keep noncritical preferences out of the critical column, and identify who owns the final decision when a result is close to the limit.
| Requirement | Pass condition | Method and condition | Frequency | If it fails |
|---|---|---|---|---|
| Mounting fit | Seats on Fixture Rev C with all four supplied screws started by hand | After support removal and 24 hours at room condition | First and last part from each build | Hold that build and notify buyer |
| Visible face | No cracks, deep gouges, or support scars visible at agreed distance and light | Zone A face, clean and dry | 100 percent visual | Segregate for review or replacement |
| Quantity and variant | Count and color match the release manifest | Count at pack-out; label by part number and revision | Every carton | Stop shipment until reconciled |
This table does not have to become a formal quality manual. It does have to be specific enough that two people can reach the same decision from the same part.
Freeze the file, material, and process assumptions
QC is meaningless if the inspected part cannot be tied to a controlled baseline. Record the CAD or mesh filename, revision, units, material family and grade when relevant, color, approved quantity, and variant map. If orientation, support contact, insert type, secondary finishing, or conditioning affects function, identify it as part of the released method.
A sample made in PLA does not validate a production batch in PETG or nylon. A fit approved after drilling or sanding does not validate an as-printed feature unless that operation is included in the production scope. A supplier substitution may be reasonable, but it should trigger the agreed review when it changes fit, appearance, durability, or traceability.
Use the fit, tolerance, and file-version guide when the revision package is still unclear. The QC plan should inspect one released job, not try to reconcile several competing file stories after production starts.
Define fit and dimensions with the real inspection condition
For every critical dimension, state the nominal or allowed range, the feature being measured, the reference or datum when needed, the tool, and the condition of the part. Temperature, cooling time, moisture conditioning, support removal, installed hardware, and access can change the result.
Functional checks are often stronger than over-specified decimals. If a printed housing must accept a buyer-supplied board, test the board and fasteners. If a clip must engage a rail, use the controlled rail sample or a go/no-go fixture. Calipers remain useful, but they may not capture hole location, flatness, assembly stack-up, or local deformation.
Do not let “within tolerance” stand alone. Identify whether the shop checks one feature, a feature relationship, or the complete assembly. When a gauge or mating part controls release, label and preserve that reference so wear, damage, or an undocumented replacement does not move the standard later.
Define cosmetic acceptance by surface and viewing condition
Layer lines are normal on FDM parts. Support contact, seams, small color variation, gloss changes, and witness marks may also be process-normal while still being unacceptable on a customer-facing surface. Divide the part into viewing zones instead of demanding a flawless finish everywhere.
- Zone A: customer-facing or presentation surface with the strictest visible-defect rules.
- Zone B: visible during use but not a presentation face; normal layer texture may be allowed.
- Zone C: hidden, internal, or installation surface where cosmetic marks are acceptable unless they affect fit.
Name the viewing distance, ordinary lighting, and whether the part is inspected before or after cleanup. Avoid microscope-level review unless the application truly requires it. Photos can document examples, but an approved physical sample usually represents gloss, texture, edge condition, and support cleanup more reliably. The surface-finish guide helps separate process-normal texture from a finish requirement that needs added work.
Use the first article as the reference for production
A first article should prove the released revision, material, method, and inspection plan before the remaining quantity is committed. Record what passed, what failed, what was changed, and whether any manual operation was needed. “Looks good” is too weak when the sample becomes the evidence for a larger order.
If the buyer approves a compensated feature, support cleanup method, insert process, or cosmetic limit, carry that decision into the release record. If the sample is approved only for geometry learning, say so; it should not silently authorize production material, packaging, or repeatability.
The first-article approval guide covers the proof and release decision in detail. Store the approved sample ID, file revision, approval date, approver, photographs when useful, and any limits on how long the reference remains valid.
Choose inspection frequency from risk and process evidence
Checking every feature on every part sounds safe, but it can add cost without improving the decision. Use 100 percent inspection when a fast functional or visual check catches a high-consequence failure. Use sampling when the process is stable, the check is slower, and a defined response catches drift before the batch escapes.
- Every part: missing hardware, gross damage, critical pass-through, simple go/no-go fit, or customer-facing cosmetic defects.
- Each build: first and last part, one part per plate, or another position-based check when build-to-build drift matters.
- Per lot or material change: dimensions or appearance that may move with spool lot, color, moisture state, or supplier.
- First article only: design questions that do not need repeated verification once the production method is controlled.
Define what happens after a sampled failure. Does the shop expand the sample, inspect the entire build, hold the lot, or replace the subgroup? Sampling without a reaction rule merely delays the disagreement.
Keep variants, builds, and lots separable
Mixed colors, mirrored parts, size variants, or revision lanes should not be treated as one anonymous pile. Give each subgroup a part number or manifest line, and decide whether a failure in one subgroup holds only that lane or the entire shipment.
Traceability can be light. A carton label, bag label, build record, or packing-list line may be enough. The goal is to identify which parts share the same file, material, process window, and inspection result when an issue appears. Without that separation, one failed variant can force unnecessary sorting across the whole batch.
For multi-build orders, ask whether the inspection record identifies build, machine, date, operator, or material lot when those facts matter to recovery. Do not demand records the job cannot use. Capture enough to contain a real problem and protect the unaffected quantity.
Define borderline, rework, replacement, and concession rules
A good plan covers parts that are not obviously good or bad. If a measurement is close to the limit, define whether a second tool, second operator, conditioning period, mating test, or buyer review decides the result. Prevent repeated measurement until one favorable number appears.
List permitted rework such as deburring, support cleanup, drilling, reaming, insert replacement, or controlled sanding. Reworked parts should be rechecked against the same acceptance rule. If the operation could change strength, appearance, fit, or interchangeability, it needs buyer approval before it becomes the recovery method.
A concession is a documented buyer decision to accept a specific known deviation. It should identify the affected quantity, lot or serial range, deviation, reason, evidence, use limitation, approver, and whether the exception applies only once. A concession should not quietly rewrite the next reorder baseline.
Connect QC to packaging, shipment release, and receiving
A technically conforming batch can still fail in transit or at receiving. Confirm whether parts must be bagged individually, separated by revision or color, protected from rubbing or bending, counted by carton, labeled with part and revision, or shipped with loose hardware in a controlled kit.
Define the release evidence the buyer expects before shipment: a simple pass summary, photographs, measurement sheet, count confirmation, certificate, or notice of approved deviations. Match the record to the risk. A consumer utility part may need only count and visual confirmation; a fit-sensitive production component may need first-and-last results per build.
Use the packaging, labeling, and inspection guide to control pack-out, and align it with the receiving checklist. The supplier inspection and buyer receiving check should test the same released story.
Ask the supplier to restate the release plan
Before production, a serious supplier should be able to confirm the controlling revision, material, quantity and variants, critical checks, cosmetic zones, sampling frequency, reference fixtures, allowed rework, hold-and-notify rule, required records, and shipment-release condition.
That restatement is a useful competence check. It exposes gaps before material and machine time are committed. The supplier should also identify requirements that are ambiguous, unusually expensive to inspect, poorly matched to additive manufacturing, or impossible to verify with the provided access and tools.
The production-readiness guide covers broader supplier questions. JC Print Farm should feel like a production partner when it can translate buyer function into an observable release plan, not when it promises perfection without defining the check. For an operator-minded review of a controlled batch request, start with JC Print Farm.
Copy this buyer-ready QC note
Please quote and release against Housing Rev D, 120 black PETG parts. Critical acceptance: the supplied PCB and four M3 screws must install without forcing after support removal and 24 hours at room condition. Check the first and last part from each build with Fixture Rev B. Zone A is the logo face: no cracks, deep gouges, or support contact visible at 24 inches under normal room light. Normal layer texture is acceptable on Zones B and C. Hold the affected build and notify us before rework or substitution. Bag 10 per labeled pack with part number, revision, quantity, and build ID. Send the per-build fit results and final count before shipment.
Adjust the material, quantity, feature, method, frequency, cosmetics, response, and pack-out to match the real job. If the quote has not yet captured those controls, use the quote-approval checklist before authorizing production.
Common acceptance-criteria questions
Do buyers need to list every possible print defect?
No. Define failures that affect function, fit, safety, customer-visible appearance, count, identity, and handling. Use representative examples or an approved sample for normal process marks. An endless defect dictionary can be harder to apply than a clear surface-zone and function-based standard.
Can an approved sample replace written criteria?
Not completely. The sample is strong evidence, but it still needs an identity, revision, material, approval record, and explanation of which properties it controls. It may not establish batch sampling, packaging, labeling, or disposition rules.
Should every critical dimension be inspected on every part?
Only when the consequence, process behavior, and speed of the check justify it. A fast go/no-go test may suit 100 percent inspection; a slow dimensional layout may be better assigned per build with a defined expansion rule after failure.
What if the buyer and supplier measure differently?
Agree on the tool, feature access, datum or reference, part temperature and condition, number of readings, and resolution path before production. For functional interfaces, a controlled mating part or gauge may remove more ambiguity than comparing isolated caliper readings.
Bottom line
Define acceptance before the batch starts: one released file and material, the few outcomes that must pass, an observable method, risk-matched inspection frequency, cosmetic zones, a failure response, and pack-out evidence. That is enough to make QC usable without turning an ordinary custom-print order into unnecessary bureaucracy.
If the file, material, quantity, critical checks, and release rules are stable, send them through the tracked quote intake. For repeat work, preserve the accepted baseline with the reorder-consistency guide so the next batch does not restart the quality conversation from memory.
Affiliate availability note (checked July 31, 2026): The General Tools vernier-caliper listing previously linked here was unavailable when rechecked. The footer now points to a current same-brand 0-6 inch digital caliper listing; it is a different instrument, so define the zero, measurement points, repeatability check, and pass/fail method in the acceptance plan.