The ice-point check
A properly made ice bath gives 0 °C to within a few millikelvin — a cheap, repeatable way to watch a PRT for drift between calibrations.
An ice point is a slurry of shaved ice made from clean water, in an insulated container, with just enough water to fill the gaps and no free water at the bottom. Made carefully it realises 0 °C closely enough for most industrial purposes.
Using it
- Immerse the probe to sufficient depth in the slurry, away from the walls, and allow it to stabilise.
- Compare with the value recorded at the last calibration (or with the certificate's 0 °C correction).
- A shift larger than the check's uncertainty indicates drift or damage and should trigger investigation.
The check does not replace calibration — it tests one point — but a PRT that holds its ice point is unlikely to have drifted grossly elsewhere, and a record of checks strengthens the case for the calibration interval.
Making one that is actually at 0 °C
The ice point is cheap and repeatable, and it is also easy to make badly. A glass of water with ice cubes floating in it is not an ice point; it is stratified water somewhere between 0 and 4 °C, and a probe in it can read half a degree high. The difference between a good bath and a bad one is entirely in the preparation.
| Step | Why | Error if skipped |
|---|---|---|
| Shaved or finely crushed ice, not cubes | Small particles pack without voids and keep the whole volume at the transition | Up to several tenths of a degree |
| Pure water for the ice — distilled or deionised | Dissolved solids depress the freezing point | 0.01–0.1 °C, more with tap water |
| Insulated container (a Dewar) | Slows melting and stops the outside warming the slurry | Drift over the measurement |
| Pack tightly, then add just enough cold water to fill voids | Air pockets are insulators; free water is not at 0 °C | 0.1 °C or more |
| No free water at the bottom | Denser water at 4 °C collects there and the probe sits in it | Up to 0.4 °C, and always in the same direction |
| Adequate immersion, away from the walls | Stem conduction from a 20 °C room | Tenths of a degree with a short probe |
Made properly the ice point realises 0 °C to within a few millikelvin, which is better than almost any working reference in the building, for the cost of a Dewar and some ice.
Reading it as a trend
A single ice-point reading proves very little. The value comes from charting it: the same probe, the same procedure, monthly, plotted. A drift then becomes visible as a trend long before the annual certificate reports it, and the chart is also the evidence for extending or shortening the calibration interval.
Set a decision threshold in advance. A shift larger than the certificate uncertainty at 0 °C is the usual trigger to investigate; a shift approaching the tolerance is a trigger to withdraw the instrument and recalibrate. Deciding the threshold after seeing a worrying reading is the same mistake as choosing a decision rule after seeing a borderline result.
What it does and does not tell you
- It tests one point. A PRT that holds its ice point is unlikely to have drifted grossly elsewhere, but unlikely is not evidence for a certificate.
- It is a check, not a calibration: no uncertainty is evaluated and no traceability link is created. It cannot extend an interval by itself.
- It catches the failures that matter most — a knock, a thermal shock, moisture ingress — because those move the ice point first.
- For a thermocouple it is far less informative, since the reference junction and the wire's thermal history dominate and an ice point exercises neither.
Frequently asked questions
- How do you make an ice point?
- Shave or crush ice made from distilled or deionised water, pack it in an insulated container, add just enough cold clean water to fill the gaps, and immerse the probe deeply, away from the walls, until stable.
- How accurate is an ice point check?
- A careful ice bath is within about 0.002–0.005 °C of 0 °C. Common errors — tap water, large cubes, free water, shallow immersion — can add tens of millikelvin.
References
- [1]ISO/IEC 17025:2017 — General requirements for the competence of testing and calibration laboratories
General technical guidance written against the cited sources. It is not regulatory or legal advice and does not replace the applicable standard, guideline or a qualified reviewer's judgement.
Have a question on this topic?
Ask ValiTrac AI and see the evidence and calculation behind the answer.
Ask ValiTrac AIFree calculator: Certificate checker
Paste the text of a calibration certificate and screen it against ISO/IEC 17025:2017 clause 7.8: uncertainty with k, traceability, as-found and as-left, the decision rule behind a pass, and every other required element, with the matched text shown for each.
Open the calculatorCalibration on ValiTrac
Support calibration workflows, uncertainty budgets and certificate generation.
Open the calibration worksheetRelated articles
Selecting a reference thermometer
Pick a reference for stability first and certificate uncertainty second: one that drifts between its own calibrations is worse than a larger honest number.
How to set calibration intervals
There is no universal interval. Set it from risk, manufacturer data and — above all — the instrument's own as-found history.