The logic of validation
To validate is to demonstrate that the sensor meets expectations under its conditions of use. This requires a stimulus whose uncertainty is clearly lower than the tolerance to be verified and an output meter that is likewise better than the sensor.
Preparation
- Stable power supply: many sensors are ratiometric, meaning their output is proportional to their supply, and any variation in the supply translates into error.
- Controlled and recorded environmental conditions, because temperature affects the output.
- Stabilization time before starting.
- Proper connection: short cables, shielding and an output load that does not distort the signal.
Measurement sequence
- Choose points distributed across the full range, including the extremes.
- Apply the stimulus in ascending and then descending order.
- Wait for stabilization at each point and record several readings.
- Repeat the cycle several times to evaluate repeatability.
What is calculated
- Accuracy error: difference between the output obtained and the output expected at each point.
- Nonlinearity: deviation from an ideal line, whose definition must be stated, for example end-point or best-fit.
- Hysteresis: difference between the ascending and descending response at the same point.
- Repeatability: spread between identical cycles.
- Noise and resolution: variation in the output with a constant stimulus.
- Time response: with a step, the time the output takes to approach the final value.
How to interpret the result
A constant systematic error can be corrected with zero and gain adjustment; one that changes with sweep direction indicates hysteresis, and high spread indicates noise or an unstable setup. It is advisable to separate the contribution of the sensor from that of the conditioning and the instrument.
If the stimulus involves hazardous voltages, pressures or energies, the work must be performed by qualified personnel in accordance with the applicable regulations.