CALIBRATION / EVIDENCE WORKSHEET
Check six-axis uncertainty, find missing evidence and export a review memo. Use a synthetic example or start blank.
Check the certificate and configuration together. ATI documents matched systems; do not assume identical pairing requirements for every digital or analog design.
Source basis: ATI — Force/Torque Sensor FAQ, 2020, sections 4.3–4.4 (matched systems and matrix processing). Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Laboratory competence and traceability matter, but neither a logo nor traceability alone establishes fitness for your application.
Source basis: NIST — policy on metrological traceability and fitness for purpose. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Use the stated units, coverage factor and reference basis. The worksheet limits are user inputs; GUM does not set a universal robot acceptance percentage.
Source basis: BIPM JCGM 100:2008 — uncertainty evaluation, sections 4–6. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
A release decision needs a specification and an agreed decision rule. Being within the entered worksheet limits is only a numerical comparison.
Source basis: ILAC G8:09/2019 — decision rules and statements of conformity. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Use this engineering procurement checklist alongside the actual report and referenced configuration records. These are review questions, not a claim that every item must appear on every certificate.
| Evidence | What to compare | If missing |
|---|---|---|
| Record identity | Report ID, sensor model/serial, issuer and calibration date | Ask the issuer for a serial-matched original record. |
| Measurement chain | Integrated or external electronics, gain, cables and revision | Identify the configuration actually calibrated. |
| Traceability | Reference chain and associated uncertainties | Request supporting evidence; lack of accreditation is not proof of fraud. |
| Accreditation claim | Accreditation ID and relevant quantity/range/method in the body directory | Keep scope unverified; a generic logo is insufficient. |
| Calibration configuration | Matrix or internal calibration revision; raw vs engineering-unit output | Clarify processing location and serial-specific version. |
| Results and conditions | Per-axis results, units, load directions, uncertainty basis and relevant mounting/environment | Request omitted data or document restricted use. |
Source basis: ISO/IEC 17025:2017 — laboratory competence (official scope) · ATI — Force/Torque Sensor FAQ, 2020, sections 4.3–4.4 (matched systems and matrix processing) · NIST — policy on metrological traceability and fitness for purpose. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
For a six-input linear model, w = C(s − s₀). Coefficients may be applied in the interface or host. Keep the calibration ID, channel order, zero handling and engineering units consistent; avoid applying the matrix twice.
| Review item | Required interpretation | Boundary |
|---|---|---|
| Revision | Match the physical unit and active calibration, not just the model name | Digital products may expose an internal ID rather than a public matrix. |
| Origin and signs | State the frame and use M = r × F for an origin shift | An origin change alters moments even when the sensor is healthy. |
| Condition number | Compare only after defining input/output scaling and units | No universal κ < 15 or κ > 50 quality cutoff; N vs kN changes an unnormalized matrix. |
| Higher-order model | Select on held-out residuals across the intended load domain | Complexity adds calibration effort and extrapolation risk; no universal percentage improvement. |
Source basis: ATI — Force/Torque Sensor FAQ, 2020, sections 4.3–4.4 (matched systems and matrix processing). Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Synthetic example: Fz full scale = 600 N and expanded U = 0.42% FS give U = 2.52 N. At a 20 N working load this equals 12.6% of reading. A small % FS can therefore be unsuitable at low load. This is a unit conversion, not a new uncertainty budget.
| Synthetic input or result | Value | Meaning |
|---|---|---|
| Fz reference magnitude | 600 N | Positive FS magnitude; not the ±600 N peak-to-peak span. |
| Reported expanded U | 0.42% FS, k=2 | Converted absolute U = 600 × 0.42 / 100 = 2.52 N. |
| Entered force U limit | 0.50% FS | 0.42 ≤ 0.50: within this entered U limit. |
| Working load comparison | 2.52 / 20 × 100 = 12.6% | Relative uncertainty at 20 N; application suitability remains a separate decision. |
| Coverage | k=2; approximately 95% only under suitable assumptions | Small degrees of freedom or non-normal distributions need another treatment. |
Source basis: BIPM JCGM 100:2008 — uncertainty evaluation, sections 4–6. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Type A and Type B describe evaluation methods, not a random-versus-systematic split. A budget needs standard uncertainties, sensitivity coefficients and correlations. Do not add datasheet maxima as if they were independent standard deviations.
| Contribution | What to request | Known / unknown in this worksheet |
|---|---|---|
| Reference force and lever arm | Reference uncertainty, geometry and sensitivity model | Unknown: the worksheet does not evaluate reference equipment. |
| Repeatability and reversibility | Repeated readings and direction-dependent residuals | Unknown: a single stated U does not expose repeat data. |
| Temperature, mounting and electronics | Influence coefficients, actual conditions and correlation treatment | Unknown: installed influences may differ from calibration. |
| Combination | For independent contributions, u_c² = Σ(cᵢuᵢ)²; otherwise include covariance | Not calculated here; source budget must justify its model. |
| Expansion | U = k × u_c, with stated k and coverage interpretation | Entered U is compared only after the user confirms its basis. |
Source basis: BIPM JCGM 100:2008 — uncertainty evaluation, sections 4–6. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Read the actual scope and contracted method. A uniaxial force standard does not automatically supply a six-axis combined-load protocol. No universal number of points, U limit, bolt torque or annual interval is asserted here.
| Reference | Relevant scope | Do not infer |
|---|---|---|
| ISO/IEC 17025:2017 | Competence, impartiality and consistent laboratory operation | Accreditation of this site or fitness of every sensor application. |
| ISO 376:2011 | Calibration/classification for static uniaxial testing-machine force-proving instruments | Validation of a full six-component combined-load envelope. |
| JCGM 100:2008 | Framework for evaluating and expressing measurement uncertainty | A mandatory 0.5% FS robot acceptance limit. |
| ILAC G8:09/2019 | Decision rules and statements of conformity | That comparing U alone proves error-plus-uncertainty conformity. |
Source basis: ISO/IEC 17025:2017 — laboratory competence (official scope) · ISO 376:2011 — official uniaxial force-proving scope · ILAC G8:09/2019 — decision rules and statements of conformity. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Ready to request missing evidence? Download the blank certificate checklist
Use the manufacturer drawing and laboratory procedure for fasteners, surfaces and environmental conditions. The numerical installation sensitivity of a specific sensor is unknown without model-specific evidence.
| Risk / tradeoff | Decision impact | Smallest useful next step |
|---|---|---|
| Misuse: same-family spare with old coefficients | A functioning output is not proof of the correct calibration | Match the active serial/calibration ID and recheck the measurement chain. |
| Scenario mismatch: single-axis evidence for combined use | Unverified residuals at the actual force/moment combination | Ask for relevant combined-load results or restrict the operating domain. |
| Installation mismatch: changed flange or amplifier | Calibration conditions no longer describe the installed system | Log the change and agree a verification plan; do not guess a universal correction. |
| Cost: broader calibration scope | More load cases, fixture work, transport and downtime may be needed | Quote only necessary ranges/directions first; published lead time and cost are unknown. |
| Alternative: restricted use pending evidence | May reduce immediate rework but limits valid decisions | Have the application owner document the restriction and unresolved uncertainty. |
Source basis: ATI — Force/Torque Sensor FAQ, 2020, sections 4.3–4.4 (matched systems and matrix processing) · NIST — policy on metrological traceability and fitness for purpose. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
The cited PTB documentation is version 3.3.0, not a claim about the latest release. It defines XML structure; signatures are optional. Schema validation and signature verification are different checks. This page performs neither.
| Check | PDF or structured data | Remaining review |
|---|---|---|
| Document format | PDF may be an original; PTB DCC uses versioned XML | JSON-LD page metadata is not a PTB DCC. |
| Structural validity | An XML file can be checked against its stated XSD | Valid structure does not prove truthful measurement results. |
| Signature | Verify signature and trust chain when present | DCC does not inherently require a signature or guarantee accreditation. |
| Software ingestion | Check axis order, units, serial and revision before import | Automation reduces transcription work but still needs mapping and review. |
Source basis: PTB DCC schema 3.3.0 — versioned XML documentation and optional signatures. Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Engineering derivation: at rest, F = m·g in sensor coordinates and M = r × F. With a fixed center-of-mass vector r, [F; M] = [I; [r]×]F spans at most three dimensions. More poses do not create six independent load inputs. For r = (0, 0, L), Mz is zero in every static pose.
| Illustrative static sensor-frame force | Moment for r = (0, 0, L) | What it can check |
|---|---|---|
| F = (0, 0, +mg) | M = (0, 0, 0) | An axial response at this load; sign depends on the declared convention. |
| F = (+mg, 0, 0) | M = (0, +Lmg, 0) | Coupled Fx/My response, not independent pure Fx and My gains. |
| F = (0, +mg, 0) | M = (−Lmg, 0, 0) | Coupled Fy/Mx response; no axial Mz excitation. |
| Additional fixed-r poses | Always M = r × F | Useful consistency checks; cannot independently identify all matrix terms. |
| Independent Mz verification | Requires a known nonzero axial torque | Use an appropriate reference torque or additional independent load fixtures. |
Basis: static wrench equations shown above. This derivation describes observability, not a validated field calibration protocol.
Intervals are a quality-system decision based on history, use, manufacturer guidance and risk. A calibration date is not a universal expiry date. Any suspected damaging event warrants review without waiting for an arbitrary overload percentage.
| Trigger | Immediate review | Next decision |
|---|---|---|
| Collision, overload or unexplained drift | Record event and isolate questionable measurements under site procedure | Inspect and agree reference checks, recalibration or replacement with the responsible engineer. |
| Electronics, firmware, matrix or mechanical change | Compare configuration against the calibrated chain | Determine what verification or recalibration is needed for that design. |
| Routine interval review | Use drift history, usage and criticality | Document an evidence-based interval; no universal 12-month mandate. |
Source basis: ILAC G24:2022 — determination and review of recalibration intervals (G-series directory). Engineering recommendations are the site’s interpretation; no source endorses this worksheet.
Illustrative teaching scenarios, not customer case histories or measured product results.
Premise: The preloaded synthetic record has all six expanded U values and user-entered limits.
Process: Run the worksheet: Fz is 0.42% FS, Mz is 0.52% FS, and the scope question remains Unknown.
Outcome / next step: Numbers are within entered limits; the scope gap remains. Request the relevant accreditation scope if the procurement task requires it.
Premise: Load the missing-evidence example; its matrix revision is blank.
Process: Run the same numerical comparison. Matrix, traceability and combined-load evidence remain gaps even if all numbers are small.
Outcome / next step: Request the serial-matched calibration configuration. Filling a revision only resolves that gap when the matching evidence is also marked checked.
Premise: A synthetic Fz axis has 600 N FS and U = 0.42% FS; the application works at 20 N.
Process: Convert to 2.52 N expanded U, or 12.6% of the working reading.
Outcome / next step: A 0.5% FS worksheet limit can be met while the process requirement is not. Review working-load uncertainty and the application decision rule.
Choose Load worked example, then Review evidence. The result retains one unknown scope item even when the numbers are within the entered limits.
Which unit and configuration were measured, the results, units, relevant conditions and associated uncertainty. Matrix/frame details may be in referenced configuration records. Use the checklist to request missing evidence.
No. It exports a user-entered evidence review memo. No physical test, issuer authentication or accreditation validation is performed.
No. All example identifiers and values are synthetic. Edits retain the example label. Choose Start blank review for your own record.
No. Check the accreditation body directory and the relevant quantity, range and method. This website does not verify that directory.
No. Examine its method, traceability, uncertainty and suitability for the contracted task. Accreditation may be required by your procurement or quality procedure.
This is design-specific. ATI documents matched systems, while other designs may integrate conversion inside the sensor. Record the calibrated configuration and follow its manufacturer instructions.
Axis ranges and uncertainties can differ. A single force number can hide the worst axis; the result shows each entered axis and converts its uncertainty to N or N·m.
No. That approximation depends on the distribution and effective degrees of freedom. Use the coverage interpretation stated by the laboratory.
No. Convert against the explicit reference magnitude at the relevant load first. If the basis is unclear, request clarification rather than inventing a number.
No. It compares reported U and crosstalk with your inputs, including equality. It does not evaluate measured error against a tolerance with an agreed conformity decision rule.
No. They can affect noise and zero but do not automatically remove gain, reference, hysteresis or model uncertainty. Reassess the measurement model and conditions.
Use the absolute maximum residual in percent of the affected receiving axis full scale at the relevant applied loads. Other definitions require conversion and an agreed test method.
Not necessarily. An integrated sensor may use protected internal coefficients. You still need a traceable calibration identifier, configuration, units, frame and relevant results.
Not when the center of mass is on the sensor z-axis: Mz = rxFy − ryFx = 0. With any fixed center of mass, moments remain coupled to force. Use independent known loads for independent channel validation.
No. Simulating a bridge signal does not apply known mechanical force or torque and cannot establish the full mechanical response.
No. Follow the specific drawing and procedure, including bolt and thread conditions. This guide supplies no universal fastener setting.
No. PTB defines structured XML; a signature is optional in the cited version. Validate the format, any signature and its trust chain, and the report scope separately.
Set and review an interval from drift history, usage, manufacturer advice and your quality procedure. Investigate suspected damage or configuration changes separately; no fixed interval fits every task.
Prepared and reviewed by HexaForce Metrology & Robotics. Repository record created ; public publication date is unverified. Review again by May 4, 2027, or sooner if the cited methods change.
Source and content review: . Editions are identified where cited; the PTB schema link deliberately targets version 3.3.0. Official scope pages do not replace the full standards or the laboratory’s contracted method.
No model-specific calibration records, laboratory accreditation evidence, service lead times or prices were supplied for this guide. Those remain unknown. Examples and limits are illustrative; worksheet entries are unverified. The worksheet performs input checks, evidence counting, % FS conversion and comparisons only.
Download the blank evidence checklist (CSV) or return to the local review tool. For the wider calibration method, see 6-axis load cell design and calibration.
Send the certificate ID, sensor serial, measurement-chain details and unresolved evidence list. Use the exported memo for your requirements; the worksheet does not authenticate certificates.
Continue with axis and sign checks, acquisition-chain review, or engineering request resources.
