Executive takeaway
- Every CTQ result is linked to an instrument ID with an active calibration status and due date.
- CMM touch-probe tools are qualified and graded; CTQs are not measured with “throughput” tools.
- Out‑of‑tolerance (OOT) tools trigger containment: HOLD tag, isolation, look‑back, and re‑inspection where required.
- Traceability is delivered as evidence: one indexed dossier per PO (revision control + inspection + cert packs + ship checks). See the dossier structure.
Traceability: the buyer definition
Buyers don’t need “we calibrate our tools.” Buyers need an evidence chain that stands up in receiving, audits, and failure analysis.
| Traceability layer | What is linked | Why it matters |
|---|---|---|
| Requirements | Drawing/model revision, specs, CTQs, decision rules | Prevents “built the right part to the wrong revision” and near‑limit disputes |
| Product | PO/job, lot/serial (if required), traveler, process route, material heat / cert mapping | Supports containment and root cause when something drifts |
| Measurement | Instrument ID → calibration certificate reference → inspection record | Turns inspection into proof instead of opinion |
| Shipment | Packing checklist, labels, kitting verification, document bundle index | Reduces receiving friction and “missing cert” delays |
Calibration control: how SEMIX runs it
1) Unique ID + visible status on every gage
Every measurement tool has a unique ID and a status label at point of use (calibration date, due date, and OK/HOLD state). Tools without status are not used for acceptance decisions.
2) Risk-based calibration intervals (not a fixed calendar)
Intervals are set by use intensity, CTQ impact, OOT history, and environment. Baseline starting points are practical (calipers/mics typically 6–12 months, critical gauges 6 months, CMM 12 months depending on supplier), then adjusted by performance and program risk.
3) Calibration certificates that are audit-ready
Certificates are expected to show traceability, results and error, calibration date, next due date, and measurement conditions. For export programs, this prevents the “certificate exists but proves nothing” problem.
4) Verification and gating in daily operations
Calibration is not a once-a-year checkbox. Production risk comes from day-to-day variation: collisions, probe changes, handling damage, contamination, and environmental drift. SEMIX uses verification gates to keep measurement reliable between full calibrations.
CMM touch-probe qualification gates (example)
For CMM touch-probe work, the “measurement system” includes the probe tool. SEMIX runs numeric qualification gates and grades tools for use. When a tool fails the gate, measurement stops.
| Gate | Stop threshold | Why it exists |
|---|---|---|
| Effective diameter drift | |ED − nominal| > 5 µm | Detects contamination, wear, or incorrect tip assumptions |
| RMS error | RMSError > 3.5 µm | Signals an unstable qualify; increases false pass/fail risk |
| Form (outlier / peak-to-peak) | Form > 20 µm | Flags outliers from looseness, dirt, or collision damage |
| Peak absolute residual | PeakAbs > 15 µm | Prevents CTQ measurement with a “spiky” tool |
| Grade | RMSError | Form | Allowed use |
|---|---|---|---|
| S — Master | ≤ 0.8 µm | ≤ 4 µm | System gating, drift checks, CTQ confirmation on suitable features |
| A — CTQ | ≤ 1.2 µm | ≤ 6 µm | CTQs / near‑limit features, audit confirmations, MSA/GR&R support |
| B — Production | ≤ 2.5 µm | ≤ 12 µm | Routine dimensions with moderate/loose tolerances |
| C — Throughput | ≤ 3.5 µm | ≤ 20 µm | Fast checks; not used to conclude CTQs or near‑limit decisions |
Practical rule used internally for minimum tolerance width relative to qualification quality: Ttotal,min = max(10×RMS, 2×Form). If a tolerance is tighter than what the qualify supports, the solution is not “argue the number”—it is to upgrade the tool, method, or environment.
Out‑of‑tolerance tools: containment + look‑back
When a tool is found OOT, the response is mechanical and documented:
- HOLD + isolation: the tool is tagged HOLD and removed from the acceptance workflow.
- Define the impact window: identify when the tool was last known-good and what jobs/CTQs were measured with it.
- Containment: re‑inspect affected characteristics using a verified method; hold shipment if risk exists.
- NCR/CAPA when required: if product risk exists, an NCR is opened; repeated or major issues trigger CAPA discipline.
- Prevention: strengthen the gate (verification frequency, handling rules, collision triggers) so the failure mode does not repeat.
How traceability is delivered with your shipment
SEMIX structures export deliveries so receiving does not need to reconstruct history. For most programs, evidence is delivered as one indexed pack tied to the PO.
- Revision lock (drawing/spec set used to build and inspect)
- Traveler / route evidence (what steps occurred and when)
- CTQ inspection report (format aligned to receiving)
- Certificate pack (material, finishing, calibration references as required)
- Pack/ship checklists (labels, kitting, protection)
RFQ snippet (copy/paste)
Use this checklist to prevent measurement disputes and to lock traceability requirements early.
- Define CTQs (or ask SEMIX to propose CTQs) and state which require reporting.
- State your traceability needs: lot/serial, material heat cert, and any special process certificates.
- Call out calibration evidence requirements for CTQ instruments (e.g., ISO/IEC 17025, NIST‑traceable, certificate fields).
- Specify measurement conditions for near‑limit features (temperature range, soak/stabilization expectations, decision rule).
- If you have a receiving template, share it so SEMIX maps report fields 1:1.
Proof snapshot (redacted example)
| Instrument ID | Type / range | Resolution | Cal date | Due date | Status | Cert ref. |
|---|---|---|---|---|---|---|
| G‑___‑### | Micrometer (0–25 mm) | 0.001 mm | YYYY‑MM‑DD | YYYY‑MM‑DD | OK | CERT‑___ |
| G‑___‑### | Height gage (0–600 mm) | 0.01 mm | YYYY‑MM‑DD | YYYY‑MM‑DD | OK | CERT‑___ |
| CMM‑___ | Bridge CMM | — | YYYY‑MM‑DD | YYYY‑MM‑DD | OK | CERT‑___ |
Where calibration and traceability matter most
- Near‑limit GD&T: flatness, true position, and profile disputes often trace back to method mismatch and uncontrolled conditions.
- Long-span geometry: thermal state and CTE can move numbers enough to flip pass/fail.
- High consequence programs: medical, aerospace, vacuum hardware, and semiconductor fixtures benefit from stronger evidence density.