General Sensor Characteristics
• Sensor concept
• Error
• Static calibration
• Uncertainty
1/10/06 BAE 5413 1
Sensor concept
– Transducer – Device which converts a sensed physical manifestation
of a measurand to a signal representation of the measurand.
Normally an energy conversion occurs. The process may be active or
passive.
– Sensor – Device which provides a signal related to a measurand.
This device may contain one or more transducers.
– Instrument – A device that processes and displays or logs a
measurand. This device may contain sensors and other signal
processing and display or logging devices.
1/10/06 BAE 5413 2
Error and Calibration
• Definitions
– Measurand - The object of a measurement
– Measurement – A particular determination of the measurand
– Inaccuracy - Difference between true value and a
measurement.
– Imprecision / Random Error - Difference between a
measurement and mean value of a measurand
• Normally specified as ±ks limits where s = std. dev., k =
coverage factor (normally =2 unless a reason otherwise)
– Bias - Difference between true value and mean value of a
measurement. Bias is normally largely a function of
adjustable parameters of the instrument
– Calibration - Determination of the relationship between the
output of an instrument and the measureand and use of that
information to eliminate bias. (AKA “static calibration”)
1/10/06 BAE 5413 3
Error and Calibration
• Inaccuracy (Computational definition)
– Inaccuracy is the sum of Bias and Imprecision.
Einaccuracy = B + Eimprecision
– Calibration - Sets B = 0
• Uncertainty – A specification of the magnitude of the bounds
of error remaining after calibration
• Sensitivity (static)
– Ratio of change in output to change in input
for: ∂q0
q0 = k static qi = k static
∂qi
– (dimensions?) Note that static sensitivity has units.
1/10/06 BAE 5413 4
Error and Calibration
• Independent nonlinearity - deviation of the input to output
relationship from a best fit straight line in the dimension of the
output.
– Generally specified as
• ± % Full Scale
• ± % Reading
– May be used to specify inaccuracy (common)
• Hysteresis - A response characteristic where the relationship
between input and output (calibration) changes with the sign of
the time rate of change of the output. Also known as deadband
• Drift - Rate of change of bias
– Specified with respect to time, temperature or other
o
modifying input. Eg. C/hr
1/10/06 BAE 5413 5
Error and Calibration
• Threshold - Minimum deviation from zero input detectable by
an instrument
• Resolution - Minimum deviation in input causing a
measurable change in output
• Span - Range of the measurand that an instrument is designed
to measure
• Dynamic range - ratio of largest to smallest dynamic input
that an instrument is designed to measure
– May be specified in db = 20 log (largest/smallest)
1/10/06 BAE 5413 6
Static Calibration
• Method
– 1. Identify all variables of importance including modifying
and interferring inputs
– 2. Fix all input variables except the measurand.
– 3. Vary the measurand while recording the instrument
output.
• The levels of the input must be known and if measured should
be traceable to a standard.
– 4. Repeat 2 and 3 above for each measurand
– 5. Document the input/output relationships for each
measurand
– 6. Use repeated measurements of the same true value of the
measurand to determine imprecision (may be required over
the range)
– 7. Adjust if possible the instrument to remove bias or specify
the calibration curves for this use
– 8. Specify the remaining inaccuracy of the instrument
1/10/06 BAE 5413 7
Characteristics of uncertainty
• Composed of:
– A. Factors described by statistical methods (random error)
– B. Factors described by other methods (scientific judgment)
• See CIPM or NIST for definitive calculation methods
– NIST document can be found at:
/home/mstone/5413/downloads/Lab1/[Link]
• A. and B. combined is known as “combined standard
uncertainty”
• Both A and B may be composed of many elements.
– See Fraden, Table 2.2
1/10/06 BAE 5413 8
Law of propagation of uncertainty
• “Where uncertainty is composed of additive factors,
the combined standard uncertainty should be
computed as a root sum of squares”
• uc = u12 + u22 + u32 + L + un2
• Expanded uncertainty: U = kuc k = 2
• Specification of uncertainty: Y = y ± U where Y is the
measurand and y is a measurement.
1/10/06 BAE 5413 9
Calculating uncertainty as a function of multiple ui
• If uc can be expressed a function of multiple inputs:
(
uc = f u1 , u 2 , u3 ,Lun )
• Then uc can be expressed as a function of the
uncertainties of each of the inputs:
2 2 2 2
∂f ∂f ∂f ∂f
uc = ∆u1 + ∆u 2 + ∆u3
+ L + ∆u n
∂u1 ∂u2 ∂u3 ∂un
1/10/06 BAE 5413 10