Documentation
Tutorials
Workflows
Kebab actions
Daily Work
Dashboard & Reports
Quality Control
ISO 17025
Reagents & Storage
Logs
Contacts
Billing
Analysis Setup
Settings
Initial Setup
Help & Reference
Architecture decisions
Architecture decisions
0:000:00

Why three different QC checks

Calibration, Control Sample and Replicate sound like they do the same thing ("check

that the result is right"), but they check three different levels: the instrument, the

method, and the individual result. You need all three — one doesn't replace another.

The scale example

  • Calibration = you calibrate the scale with a certified reference weight, so it's

traceable to the official kilogram. → "Is the INSTRUMENT trustworthy and traceable?"

  • Control Sample = each day you weigh a known control sample and see whether it lands

where it should. → "Is the METHOD in control today?"

  • Replicate = you weigh the same sample twice and see whether they agree. → *"Is THIS

result repeatable?"*

Calibration (ISO/IEC 17025 §6.5)

Establishes the metrological traceability of the instrument to reference standards (ultimately

to the SI International System).

  • Subject: the instrument.
  • Frequency: periodic; the instrument is locked out of new analyses when its calibration expires.
  • Without a traceable instrument, everything else is unreliable.

Control Sample (ISO/IEC 17025 §7.7.1)

A known control material you run regularly to monitor that the method stays in statistical

control (Shewhart control chart).

  • Subject: the method / the work batch.
  • Judged against the Control-Chart Limits (target ± limits).
  • Out of limits → stop and investigate (instrument, reagents, operator) before releasing results.

Replicate (ISO/IEC 17025 §7.7.1f)

Re-testing the same customer sample independently, to check the repeatability / precision of

that particular result.

  • Subject: one individual result.
  • The replicate shares the same sample as the original, but has its own analysis and report number.
  • Judged by RPD (relative percent difference) against the RPD limit (on the Control-Chart Limits).
  • Large divergence → approval is blocked until it's investigated (approval guard).

At a glance

Calibration (§6.5)Control Sample (§7.7.1)Replicate (§7.7.1f)
Checksthe instrumentthe method todayone result
Question"Traceable?""In control?""Repeatable?"
Againstreference standardsControl-Chart LimitsRPD limit
On failureinstrument lockoutstop & investigate the batchapproval blocked

The practical rule

Calibration = the tool is right. Control Sample = the method is right today.

Replicate = this result is right. Three levels, three separate checks.