| A good sensor obeys the following rules: | | | | If the output signal slowly changes independent of |
| 1. the sensor should be sensitive to the measured | | | | the measured property, this is defined as drift. |
| property | | | | Long term drift usually indicates a slow degradation |
| 2. the sensor should be insensitive to any other | | | | of sensor properties over a long period of time. |
| property | | | | Noise is a random deviation of the signal that varies |
| 3. the sensor should not influence the measured | | | | in time. |
| property | | | | Hysteresis is an error caused by the fact that the |
| In the ideal situation, the output signal of a sensor is | | | | sensor not instantly follows the change of the |
| exactly proportional to the value of the measured | | | | property being measured, and therefore involves the |
| property. The gain is then defined as the ratio | | | | history of the measured property. |
| between output signal and measured property. For | | | | If the sensor has a digital output, the signal is |
| example, if a sensor measures temperature and has | | | | discrete and is essentially an approximation of the |
| a voltage output, the gain is a constant with the unit | | | | measured property. The approximation error is also |
| [V/K]. | | | | called digitization error. |
| If the sensor is not ideal, several types of deviations | | | | If the signal is monitored digitally, limitation of the |
| can be observed: | | | | sampling frequency also causes a dynamic error. |
| The gain may in practice differ from the value | | | | The sensor may to some extent be sensitive for |
| specified. This is called a gain error. | | | | other properties than the property being measured. |
| Since the range of the output signal is always limited, | | | | For example, most sensors are influenced by the |
| the output signal will eventually clip when the | | | | temperature of their environment. |
| measured property exceeds the limits. The full scale | | | | All these deviations can be classified as systematic |
| range defines the outmost values of the measured | | | | errors or random errors. Systematic errors can |
| property where the sensor errors are within the | | | | sometimes be compensated for by means of some |
| specified range. | | | | kind of calibration strategy. Noise is a random error |
| If the output signal is not zero when the measured | | | | that can be reduced by signal processing, such as |
| property is zero, the sensor has an offset or bias. | | | | filtering, usually at the expense of the dynamic |
| This is defined as the output of the sensor at zero | | | | behaviour of the sensor. |
| input. | | | | Resolution |
| If the gain is not constant, this is called nonlinearity. | | | | The resolution of a sensor is the smallest change it |
| Usually this is defined by the amount the output | | | | can detect in the quantity that it is measuring. Often |
| differs from ideal behaviour over the full range of the | | | | in a digital display, the least significant digit will |
| sensor, often noted as a percentage of the full | | | | fluctuate, indicating that changes of that magnitude |
| range. | | | | are only just resolved. The resolution is related to the |
| If the deviation is caused by a rapid change of the | | | | precision with which the measurement is made. For |
| measured property over time, there is a dynamic | | | | example, a scanning probe (a fine tip near a surface |
| error. Often, this behaviour is described with a bode | | | | collects an electron tunnelling current) can resolve |
| plot showing gain error and phase shift as function of | | | | atoms and molecules. |
| the frequency of a periodic input signal. | | | | |