QETPP study materials
QETPP study resources

QETPP: Calibration, Measurement Uncertainty, and Records

QETPP · Work through your exam, one topic at a time.

All guide sections

Calibration, Measurement Uncertainty, and Records

Domain focus

This domain tests whether a candidate can produce measurements that survive an audit: traceable calibration, a stated uncertainty, a documented decision rule, and records that cannot be altered unnoticed. It is the domain where "the meter read 0.48" is not yet an answer.

Calibration and traceability

An instrument is calibrated when its readings are compared, through an unbroken chain, to a national standard. ISO/IEC 17025 accreditation is evidence that the laboratory's chain and uncertainty statement are competent. Calibration is not verification (checking a reading against a limit) and not adjustment (changing the instrument to reduce error).

Uncertainty and decision rules

Every measurement carries an expanded uncertainty U = k·u_c, usually with k = 2 for roughly 95% confidence. Under simple acceptance, a result passes if it lies inside the limit. Under guard banding, the acceptance zone is narrowed by U (or by whatever shared-risk rule the specification names), and results inside the guard band are reported as indeterminate.

Worked example

A bonding continuity limit is 0.50 Ω. The instrument reads 0.48 Ω with U = 0.03 Ω at k = 2. Simple acceptance: pass. Guard-banded acceptance limit: 0.50 − 0.03 = 0.47 Ω. Because 0.48 Ω exceeds 0.47 Ω, it sits in the guard band and must be reported as indeterminate - retest with a better instrument or investigate the joint.

Data integrity

Records follow ALCOA+: attributable, legible, contemporaneous, original, accurate - plus complete, consistent, enduring, and available. Corrections are single-lined, initialed, dated, and the reason stated. Never overwrite, back-date, or transcribe readings later onto a clean sheet.

Traps that cost marks

  • Treating the certificate's "accuracy" figure as the measurement uncertainty.
  • Ignoring k when comparing U with a limit.
  • Accepting a certificate that states only pass/fail with no data.
  • Using an out-of-tolerance instrument and not estimating the impact on prior results.