Diagnostics
Introduced in: rtd-acquire 0.1.0a1
Different acquisition devices use different terminology for similar conditions.
rtd-acquire maps supported observations to stable DiagnosticCode values so
application logic does not have to key off vendor-specific strings.
Examples include:
- sensor/open/short/input-fault conditions;
- high and low resistance thresholds;
- overrange and underrange;
- reference and input-voltage faults;
- ADC/PGA electrical conditions;
- CRC, register, memory, and SPI integrity errors; and
- generic configuration or hardware faults when the native evidence does not justify a more specific claim.
Severity
Every diagnostic has a warning or fault severity.
A warning can coexist with a usable resistance. A fault means the measurement cannot carry a trustworthy resistance.
Do not infer more than the hardware reports
Normalization is evidence-driven. For example, physically disconnecting a wire
may be known to the person running a test, but the driver should not report a
specific sensor_circuit_open code unless the device's observable native state
supports that interpretation.
The acquisition layer should preserve useful evidence without turning a coarse vendor status into a more specific physical diagnosis than the hardware actually established.