Alarm accuracy
The IEC 60846-1:2009 §8.13 alarm accuracy evaluation — exposure at 0,8 and 1,2 times the alarm level, the timing requirements for dose rate alarms, and the effect of the expanded uncertainty of the conventional true value on the exposure conditions.
Last reviewed
The alarm is a safety function. Unlike the indication, it is not read and interpreted — it either occurs or it does not, and an operator acts on it without further judgement. This test establishes that the alarm level is where the instrument declares it to be.
Requirements
Two conditions, at each alarm set point:
- at 0,8 × Hₐ, where Hₐ is the alarm level, no alarm occurs;
- at 1,2 × Hₐ, the alarm occurs.
Both must be satisfied for a pass. The tool returns no verdict until both are answered, so an incomplete evaluation and a passed evaluation are distinguishable in the report.
Number of set points
At least two set points are tested: one near the upper end of the effective range, and one near the upper end of the second decade. A single set point characterises the alarm at one level only, and alarm thresholds implemented in firmware frequently behave differently across decades.
Expanded uncertainty of the conventional true value
The conventional true value is the dose equivalent rate assigned to the reference field by the calibration laboratory. It carries an expanded uncertainty, U_rel, normally quoted at a coverage factor of k = 2.
That uncertainty must be included in the exposure conditions, otherwise the test is not conclusive. Exposing at exactly 0,8 × Hₐ when the field itself is uncertain by several percent means an instrument whose threshold is correct may still alarm, and one whose threshold is wrong may not.
The exposures are therefore widened:
Enter U_rel in the tool’s settings; it is a property of your calibration field, not of the instrument, and it applies to every alarm row. The tool shows the resulting exposure values on each row so the conditions actually used are recorded rather than inferred.
Dose rate alarms: the timing requirement
For a dose rate alarm the criterion is not simply whether the alarm sounded. The alarm state is observed over a ten-minute exposure at each level:
- at the lower level, the alarm must be active for less than 60 s in total across the exposure;
- at the upper level, the alarm must be active for more than 540 s in total, and must first activate within 10 s of the exposure starting.
The cumulative form of the requirement exists because a dose rate alarm near its threshold does not switch cleanly. Statistical variation in the indication causes it to cross the threshold repeatedly, and the alarm chatters. An alarm that is active for 45 s out of 600 s at the lower level is behaving acceptably; one active for 300 s is not, even though neither is a clean on or off.
The 10 s first-activation requirement is separate and is not covered by the cumulative total. An alarm that eventually accumulates 560 s of active time but took 40 s to activate has failed: the operator was in an over-threshold field for 40 s without warning.
Dose alarms
For a dose alarm — one that triggers on accumulated dose rather than rate — the observation is simply whether the alarm occurred at each of the two exposures. There is no timing component, because the alarm condition is monotonic: accumulated dose does not decrease, so the alarm does not chatter.
Record which type each set point is. The tool carries the type per row because the evaluation differs.
Failure patterns
Alarm occurs at the lower exposure
The effective threshold is below the declared level. This is the conservative direction of error and is often accepted in service, but it produces nuisance alarms, and nuisance alarms train operators to disregard the alarm. Record it rather than dismissing it.
No alarm at the upper exposure
The effective threshold is above the declared level. This is a safety-relevant failure: the instrument does not warn at the level the operator has been told it warns at.
Chattering at the lower exposure
Cumulative active time between 60 s and the pass condition indicates the alarm threshold sits within the statistical variation of the indication at that dose rate. The instrument’s usable alarm setting is higher than the one tested, and the report should say so.
Delayed first activation
First activation later than 10 s with an otherwise compliant cumulative total normally indicates that the alarm is evaluated from an averaged indication with a long time constant. The behaviour is related to the response time result, and the two should be reported together.
Reporting
Record Hₐ, the alarm type, the U_rel used, and the exposure values actually applied. An alarm result without the exposure values cannot be reproduced, because the widening from U_rel is laboratory specific.