Fundamentals of Statistics contains material of various lectures and courses of H. Lohninger on statistics, data analysis and chemometrics......click here for more.


Sensitivity

The sensitivity of a sensor or a measuring device is defined by the ratio of the change of the sensor signal to the change of the actual value of interest. The larger the change of the measured signal for a given change of the actual value the larger is the sensitivity. One may define the sensitivity as the slope of the characteristic of a sensor at a given point:
The characteristic of a linear sensor exhibiting a constant slope. The calibration points are indicated by red crosses.

If we look at the characteristics of various sensors, we immediately see that only very few measurement devices exhibit a linear characteristic. In most cases the characteristic is curved and can be approximated only by a more or less complicated function (a polynomial of higher order, for example). If the slope is not constant (i.e. the charactistic is curved) the sensitivity of the sensor changes in dependence of the input signal.

An extreme example of a sensor exhibiting a nonlinear characteristic is the NTC resistor which exhibits a hyperbolic characteristic. As a consequence measurements in the lower temperature range will be much more sensitive (by orders of magnitude) than measurements in the upper temperature range.
The highly nonlinear characteristic of an NTC resistor (NTC = negative temperature coefficient). The sensitivity changes with the temperature and is, for example, subtantially higher at temperature T1 than at T2.

As a highly nonlinear behavior causes considerable instrumental difficulties, one normally tries to linearize nonlinear sensors by adding correction networks which counteract the nonlinearity of the sensors.