Technical reference
Automatic doesn't remove the need for trust in the measurement. It moves it. A person's reading can be challenged on the spot, but a sensor's reading is relied on from a distance, so the evidence has to show the device was accurate, in the right place, reporting, and able to reach someone.
Air temperature moves faster than food temperature, because a large mass of food warms and cools slowly. A short air spike after a door opening usually says little about the food, while a slow, steady climb is what a failing unit tends to look like.
That's why alarm delay is a design choice: too short and people stop responding, too long and the food has lost its margin. Set it for the unit and test it against real records.
Parameters
| Hazard | Food left unsafe because a unit failed and no working sensor or alarm showed it |
|---|---|
| Instrument (Australia) | Readily accessible, measures to ±1 °C ( cl. 22) |
| Calibration guide (Australia) | At least every 12 months; replace or adjust if correction is over 1 °C () |
| Silence | No reading is not a pass. Cover it with manual readings |
| What a sensor shows | Usually air temperature, so check the food when the reading matters |
| Records (Australia, category one) | Logger graphs can be a record. Keep for 3 months (FSANZ guidance) |
Records that prove it
- Register tying each device to its unit, batch, certificate and warranty
- Installation report, with photos and locations
- Alarm recipients confirmed in writing, and the test results
- Quiet-unit cases: since when, cause, what the site was told
- Manual readings during the gap
- Accuracy results for each unit, and swaps
- Recall record for a faulty model or batch
Reference temperatures
Values from , a code for mass catering. Your own law or plan may set different ones.
CCP or prerequisite?
That depends on the for your process. The number, if it has one, belongs to your plan.