Engineering research
How to detect sensor drift after calibration
Recheck a stable reference over time and plot the difference. Record the instrument, environment, and reference condition to help identify the source of changes.
Gather raw readings, reference values and uncertainties, calibration-coefficient versions, and environmental records. Keep timestamps and retain the raw readings.
Calibration describes instrument response at the time of calibration. Independent check observations and repeated measurements help track changes across the working range.
How to do it
Set a baseline and check points
Use reference points across the working range. Record settling time, units, and calibration order. Keep an independent check standard or observation outside the fitted calibration.
Record repeatability and bias separately
Replicates under the same conditions show short-term scatter. Compute indicated-minus-reference differences on corrected readings and define the sign and unit.
Track change over time
Keep the check procedure fixed and plot differences by date. Apply the prespecified tolerance and appropriate control rules. Check temperature, reference condition, and setup before attributing a change to the instrument.
Trace the affected data
When a check exceeds limits, identify the last passing check, the first abnormal check, affected batches, and applicable coefficients. Preserve raw values and record dispositions for later analysis.
Worked example
Synthetic example using a 50.0 °C reference with a ±0.5 °C working tolerance. Bias is corrected reading minus reference.
| Time | Corrected reading | Bias and action |
|---|---|---|
| Calibration day | 50.1 °C | +0.1 °C, within tolerance |
| One week later | 50.2 °C | +0.2 °C, within tolerance |
| Two weeks later | 50.7 °C | +0.7 °C, repeat the check and review the last week of data |
Reference and measurement uncertainties affect decisions near a limit. Follow the decision rule specified by the lab or study.
Common questions
Can I rezero the sensor at each calibration?
Yes, with a record of the pre-adjustment values and coefficient version so you can trace when the response changed.
Do stable replicates rule out drift?
Replicates describe scatter over a short period. Compare reference checks across periods to examine long-term change.
Verified sources
The official sources below were checked on 2026-11-13. Access conditions and update dates are rechecked before publication.
RigorMed workspace
Review my sensor check records
Bring authorized timestamps, reference values, raw readings, and environmental data. Organize a bias table, time plot, and list of batches to review.
