This is a ready-to-use worksheet. It records the reasoning behind a calibration tolerance so that two years later, when an inspector asks why an instrument tolerance is what it is, the answer is on file rather than in someone’s memory. Work top to bottom: process requirement, then instrument tolerance with margin, then standard accuracy and the ratio, then the guard-band decision. Replace every <<FILL: ...>> placeholder. A filled specimen follows. This content is general educational reference, not legal or regulatory advice.
Instrument and measurement
| Field | Entry |
|---|---|
| Instrument ID / description | <<FILL>> |
| Measured parameter | <<FILL: e.g. shelf temperature>> |
| Operating point(s) the process actually runs at | <<FILL>> |
| Classification | Critical / Non-critical |
Step 1: Process requirement (the tightest real need)
| Field | Entry |
|---|---|
| What the process needs the measurement to be within | <<FILL: e.g. validated range -42 to -38 degC>> |
| Source of that requirement | <<FILL: validation report / registered condition / spec, with ID>> |
| Allowable process band | <<FILL: e.g. ±2 degC>> |
Step 2: Instrument tolerance (with margin retained)
| Field | Entry |
|---|---|
| Allocation of the process band to the instrument | <<FILL: e.g. no more than 1/4 to 1/3>> |
| Derived instrument tolerance | <<FILL: e.g. ±0.5 degC>> |
| Basis of tolerance | Percent of reading / percent of full scale / absolute |
| Test points the tolerance applies to | <<FILL>> |
| Margin retained for control variation | <<FILL: process band minus instrument tolerance>> |
The instrument tolerance must be tighter than the process requirement, because the instrument error consumes part of the available margin. Setting the tolerance equal to the process band leaves nothing for normal control variation, and is a common finding.
Step 3: Standard accuracy and the ratio
| Field | Entry |
|---|---|
| Reference / working standard proposed | <<FILL: ID, description>> |
| Standard accuracy or full uncertainty | <<FILL>> |
| TAR = instrument tolerance / standard accuracy | <<FILL>> |
| TUR (using full measurement uncertainty), if used | <<FILL>> |
| Meets 4:1? | Yes / No |
If the ratio is below 4:1, either select a better standard or complete Step 4.
Step 4: Guard-band decision (only if ratio below 4:1 or a decision rule is required)
| Field | Entry |
|---|---|
| Guard-band applied? | Yes / No |
| Decision rule | <<FILL: e.g. accept only if error within tolerance minus standard uncertainty>> |
| Effective acceptance limit after guard-band | <<FILL>> |
| Justification for the reduced ratio | <<FILL>> |
| Reference | ILAC G8 (decision rules and statements of conformity) |
A guard-band shrinks the acceptance limit by a portion of the measurement uncertainty, so you only pass instruments comfortably inside the tolerance. It reduces the risk of accepting a bad instrument at the cost of occasionally rejecting a good one; the rule must be written, not improvised.
Step 5: Consistency and approval
| Check | Result |
|---|---|
| Tolerance identical in the master list, the procedure, and the certificate | Yes / No |
| Process requirement recorded alongside the tolerance | Yes / No |
| Derivation approved | <<FILL: name, signature, date>> |
Acceptance criteria
A tolerance is correctly derived when: it traces to a documented process or product requirement; it is tighter than that requirement with margin retained; the standard achieves at least 4:1 (or a justified, guard-banded lower ratio); and the same tolerance appears in the master list, the procedure, and the certificate.
Filled specimen
The following shows the worksheet completed for a lyophilizer shelf RTD. Numbers are illustrative.
| Step | Entry |
|---|---|
| Instrument | TI-2041, RTD, lyophilizer shelf, critical |
| Operating point | -40 degC (process runs at the target, near mid-range of the sensor) |
| Step 1: process requirement | Validated shelf range -42 to -38 degC (band ±2 degC), from PV report PV-LYO-07 |
| Step 2: instrument tolerance | ±0.5 degC (about 1/4 of the band), absolute basis, applies at -40 and at range checks; margin retained ±1.5 degC |
| Step 3: standard | Reference dry block ±0.1 degC; TAR = 0.5 / 0.1 = 5:1; meets 4:1 |
| Step 4: guard-band | Not required (ratio comfortably above 4:1) |
| Step 5: consistency | Tolerance ±0.5 degC matches master list, WI, and certificate; requirement recorded; approved R. Gomez 01 July 2026 |
Here the tolerance is derived from the validated band, not copied from a datasheet, the margin for control variation is explicit, and the standard beats 4:1 so no guard-band is needed. Contrast a case where the only available standard is ±0.2 degC: the ratio drops to 2.5:1, and Step 4 would set an effective acceptance limit near ±0.3 degC with the reasoning recorded.
Common inspection findings this worksheet prevents
- A tolerance equal to or looser than the process requirement, with no margin.
- A tolerance copied from a datasheet with no link to what the process actually needs.
- A standard no more accurate than the instrument it checks, with no TAR/TUR justification.
- The tolerance differing between the master list, the procedure, and the certificate.
- No record of why a tolerance is what it is, two years after it was set.
How to adapt this worksheet
- Complete one worksheet per critical instrument (or per instrument family with a common measurement) and file it with the calibration record.
- Where you standardize tolerances across a fleet, keep one derivation per family and reference it from each instrument.
- Set your house allocation rule (a quarter, a third) in the SOP and apply it consistently.
- Confirm the ratio and guard-band references against the current published versions before you rely on them.