Bringing the Benefits of Real-Time Data Collection to the World
Sutron Corporation, 22400 Davis Drive, Sterling, Virginia 20164-4444
46
ERROR DEFINITIONS AND EXAMPLES
Linearity:
Linearity error is the deviation of the output from a straight line.
Many transducers for measuring physical phenomena have outputs which
do not vary linearly with the phenomena being measured. Sometimes the
deviation from linear is slight and is accepted as part of the error of the
device, in other cases manufacturers attempt to "linearize" the output.
Usually when this done through an analog means, there will be a residual
non-linearity, i.e., the non-linearity is not completely removed. The
following non-linearity graph illustrates why a two point calibration (zero
and full scale) is many times insufficient and ensures that the maximum
error is achieved. A best fit straight line decreases the maximum error by
ensuring that the errors fall evenly on both sides.
Example of Linearity Error
Input
O
u
tp
u
t
-20
0
20
40
60
80
100
0
50
100
Non-linear
Output
Ideal Output
Best Fit
Temperature coefficient --
This is made up of two components, the temperature
coefficient of the offset and the temperature coefficient of the slope. These can be referred to as thermal
shifts, temperature dependence, and other names. The offset is sometimes referred to as zero. The
slope is sometimes referred to as span, sensitivity, or scale factor. The following graph illustrates why the
temperature coefficient is an important specification for remote equipment. In an indoor environment a
competitor's barometer is about a 0.5 mB instrument. Over the operating temperature range of Sutron
equipment, it degrades to a +3 mB instrument.
