Eccentricity & Linearity Check

Eccentricity (also called 'off-centre load' or 'corner load' error) tests whether the balance reads the same value regardless of WHERE on the weighing pan the load is placed. Place a single reference weight at the centre, record; move it to the front-left, record; front-right, record; rear-left, record; rear-right, record. Compute the maximum deviation from the centre value. The maximum deviation is the eccentricity error. The test catches mechanical issues a centre-placed calibration weight cannot catch: load-cell asymmetry, pan misalignment, levelling-foot uneven adjustment, vibration-platform tilt, and physical damage from operator drops or knocks.

Linearity tests whether the balance's response is proportional across its used range. Place reference weights at multiple points across the range (typically low + mid + high + replicate), record each reading, compute the deviation from the nominal at each point, and the deviation differences across the range. Pure 'span' calibration at the high end can mask offset at the low end; pure two-point calibration can mask bow in the middle; multi-point linearity catches all three. The test catches calibration drift at one end of the range, non-linearity introduced by load-cell ageing or temperature, and offset issues from electronics drift.

The standard eccentricity test (per OIML R 76-1 + USP <1251> + EURAMET cg-18) uses a single reference weight equal to 1/3 to 1/2 of the balance maximum capacity, placed at five positions on the pan: centre, front-left, front-right, rear-left, rear-right. The reading at each position is recorded after stability dwell, the centre reading is the reference, and the maximum deviation from centre (in absolute mass units or as a percentage of the test load) is the eccentricity error.

Acceptance criteria are typically derived from the operating tolerance the balance is used to enforce. A common rule is eccentricity ≤ 1/3 of the smallest operating tolerance, or ≤ 0.05% of the test load, whichever is tighter. For a balance enforcing ±2 mg charge tolerance, eccentricity ≤ 0.7 mg at the test load is defensible; ≥ 1 mg indicates a corner-load issue that will produce variable weighments depending on operator placement. The acceptance criterion is per-balance + per-use-case, captured on the balance master + PQ protocol, and change-controlled.

The standard linearity test uses reference weights at multiple points across the balance's used range, typically five points: 0% (zero check), 25%, 50%, 75%, and 100% of the maximum used capacity (not the maximum balance capacity if the balance is over-specified for the application). At each point, place the weight, record the stability-dwelled reading, compute the deviation from nominal. The linearity error is the maximum absolute deviation, OR the deviation difference (max - min) across the range, depending on the acceptance specification chosen.

Acceptance criteria pair with the operating tolerance and the capability-ratio discipline. A typical rule is linearity error ≤ 1/3 of the tightest operating tolerance applied across the used range. For a balance enforcing ±2 mg charge tolerance in the 50-500 mg range, linearity ≤ 0.7 mg across that range is defensible. Tighter is better; the acceptance must be at least one capability tier (3×) finer than the operating tolerance — otherwise linearity-pass / operation-fail becomes possible.

Test frequency is risk-based per ICH Q9(R1) and is per-balance-class + per-use-case. A 5-place analytical balance dispensing potent APIs runs eccentricity weekly + linearity monthly + both at every formal calibration. A bench balance dispensing wide-tolerance excipients runs eccentricity monthly + linearity quarterly + both at every formal calibration. A receiving-area pallet scale runs eccentricity quarterly + linearity annually + both at every formal calibration. The frequency is captured in the balance qualification SOP + the PQ protocol + the calibration plan, and change-controlled.

1. Test technician opens the routine-qualification record in V5; system pre-populates balance ID + test plan (positions + test-points + acceptance criteria) + reference weight set IDs + last test results for trending.
2. Pre-flight: reference weights cert valid + balance level confirmed + ambient temperature + humidity within qualified range + warm-up time elapsed; any mismatch is a hard-block.
3. Eccentricity test executed: reference weight placed at centre, stability dwelled, reading recorded; moved to front-left, dwelled, recorded; front-right, rear-left, rear-right; max deviation computed against acceptance.
4. Linearity test executed: weights placed at each of the five (or as-defined) test-points, with replicates as required; readings recorded; deviation from nominal computed at each point; max deviation + deviation-difference computed against acceptance.
5. If eccentricity OR linearity OOT: instrument status flips to 'out-of-service pending investigation'; deviation auto-opens; back-impact workflow auto-queries every dispense_result since the previous passing test of the failing dimension; per-batch QRM scoring drives disposition.
6. If both pass: technician e-signs the record; reviewer (independent) e-signs; instrument status remains 'in-service'; results written to balance master; next-test-due recomputed.
7. Audit trail captures every reading + position + test-point + acceptance + disposition + signature with UTC timestamp; raw readings immutable; corrections (rare) e-sig'd separately.
8. Results feed the SPC trending dashboard: eccentricity max-deviation trend + linearity max-deviation trend per balance per test-frequency; control-limit breaches page the lab supervisor + open an investigation candidate before formal OOT.
9. Quarterly product review aggregates per-balance eccentricity + linearity trends; sustained drift triggers PM-frequency revisit + load-cell-life evaluation + environment-stability assessment.
10. PDF rendered + stored in regulated-reports bucket scoped by tenant; cross-referenced from balance master + from every dispense_result that consumed the qualified balance.

1. Eccentricity skipped entirely — calibration plan covers only centre-point readings; balance has corner-load asymmetry that produces operator-placement variability; first inspector question 'how do you know the balance reads the same regardless of placement?' has no answer.
2. Linearity tested only at calibration weight points — calibration done at 100 g + 200 g + 500 g but the balance is used for 25 g charges; low-range linearity uncalibrated; charge weighments at 25 g may be systematically biased without anyone knowing.
3. Eccentricity / linearity pass logged with no raw readings — record shows 'pass' without the position-by-position or test-point-by-test-point readings; trend data lost; data-integrity finding under MHRA DI 2018.
4. Test frequency not risk-based — every balance tested annually regardless of use-case; potent-API analytical balance tested at same frequency as receiving floor scale; risk-justified frequency missing; §211.68 + ICH Q9 deficiency.
5. Acceptance criteria not tied to operating tolerance — eccentricity acceptance ±10 mg on a balance enforcing ±2 mg charge tolerance; acceptance is 5× the operating tolerance, meaning eccentricity error can be 5× the tolerance the balance is asked to enforce; the test is meaningless.
6. Environment-change trigger ignored — bench moved, HVAC serviced, new equipment installed nearby, none triggers re-test; eccentricity degrades silently; chronic variable weighments emerge weeks later as 'operator technique' false root-cause.
7. Eccentricity OOT 'recalibrated and continued' with no back-impact — corner-load error discovered, balance re-levelled, no investigation of weighments produced between the previous pass and this OOT; §211.192 + §211.100 deficiency identical to as-found OOT skipping.
8. Reference weight nominal used as 'actual' for deviation calculation — instead of using the weight's cert-stated actual mass, the technician uses the nominal (e.g. '200 g' instead of '200.0017 g per cert'); systematic offset in linearity calculations; ISO/IEC 17025 traceability deficiency.

• Eccentricity test compliance rate — fraction of scheduled eccentricity tests performed on or before due date; target ≥98%; misses force kiosk warning + scheduled review.
• Linearity test compliance rate — fraction of scheduled linearity tests performed on time; target ≥98%.
• Eccentricity max-deviation trend — per-balance trend of max corner-load deviation across consecutive tests; positive slope predicts upcoming OOT + load-cell wear; SPC chart with control limits.
• Linearity max-deviation trend — per-balance per-test-point trend; identifies whether drift is uniform across range (calibration issue) or localised (load-cell wear or electronics issue).
• Routine-qualification OOT rate — fraction of routine tests that fail acceptance; target ≤2%; sustained higher rate indicates frequency-too-low or environment-degraded or instrument-end-of-life.
• Back-impact assessment cycle time — median hours from eccentricity / linearity OOT to deviation closure with back-impact disposition; target ≤72h.
• Reference-weight cert-currency rate — fraction of routine tests performed with reference weights inside their cert valid-through period; target 100%.
• Environment-change trigger compliance — count of qualifying environment changes (bench moves, HVAC service, equipment installed nearby) followed by re-test within defined window; target 100%; near-zero indicates change-control SOP not capturing environment triggers.

1. Balance master carries eccentricityTestPlan + linearityTestPlan (positions + test-points + acceptance criteria + frequency) + nextEccentricityDue + nextLinearityDue + last-N-results for trending; all under change-control with QA approval for plan edits.
2. Routine-qualification record schema enforces required fields: reference weight set IDs + cert valid-through + uncertainty (hard-block if expired), per-position eccentricity readings + per-test-point linearity readings (hard-block on save if any reading blank), acceptance evaluation engine-side, two e-sigs (technician + independent reviewer).
3. OOT detection is engine-side: any position deviation outside the eccentricity acceptance OR any test-point deviation outside the linearity acceptance auto-opens a deviation + auto-queries every dispense_result since the previous passing test of the failing dimension + pre-populates the back-impact workflow.
4. Instrument status state machine: in-service → routine-test-due (soft-warning) → routine-test-due-hard-block (grace expired) → out-of-service-pending-investigation (OOT detected) → in-service (only after deviation closure if OOT, or after pass if routine).
5. Routine-qualification kiosk widget runs the technician through the position / test-point sequence in order; live readings tracked against acceptance; replicate readings prompted as defined.
6. Environment-change SOP integrates with the routine-qualification trigger: HVAC service tickets, bench-move events, nearby-equipment-install events all auto-trigger a re-test requirement on affected balances within a defined window.
7. SPC dashboard: per-balance eccentricity max-deviation trend + linearity max-deviation trend + linearity per-test-point trend + routine-qualification OOT rate; control-limit breaches page the lab supervisor.
8. Vendor-calibration coordination: if vendor calibration does not include eccentricity (common with field-service contracts), internal eccentricity test is required on receipt before return-to-service; vendor scope gaps logged + flagged at scorecard.
9. Routine-qualification PDF generated from the record + cross-referenced from balance master + from every dispense_result that consumed the qualified balance; PDF rendered through @react-pdf/renderer + stored in the regulated-reports bucket scoped by tenant.
10. Quarterly product review (ICH Q10 §3.2.5) auto-aggregates per-balance eccentricity + linearity trends + OOT rates + adjustment events; sustained drift triggers PM-frequency revisit + load-cell-life evaluation + environment-stability assessment; chronic OOT triggers balance-replacement business case.