Modeling

• Introduction
• General model
• Static transfer
• Dynamic transfer
What are the important parameters of a
measurement system?
 Force sensor sensitivity

Force sensor: strain gauge on a steel bar R a

 K  F
Strain gauge R F Y
Force F a – proportionality factor
Y – Young’s modulus
a/Y – Sensitivity

Resistance R

120 
Temperature T
12 N Force F
 Force sensor
• Influence of the force (measured Resistance R
quantity) on the sensor output:
R  K 
a
F 125 
R Y
120 
F

• Influence of the temperature

12 N Force F
R Resistance changes with the
 T T temperature:
R T,offset added offset on the left‐hand side

Y
 Y T, Y  0.26103 /C (steel)
Y T
Material’s stiffness (Young’s modulus of
the steel bar) changes with the
temperature:
modified slope
xi – interfering input
– adds an offset

xm – modifying input
– changes the slope (sensitivity)

y – measured quantity
 General model

xd Fd Fd, Fi – transfer functions

xi Fi

xd – desired quantity

xi – interfering input
 General model

xd Fd mFd Fd, Fi – transfer functions

mFd, mFi – parameters of Fd, Fi
xm y
xi Fi mFi

y y(xd, xi, xm)

xd– desired quantity xm
– modifying input xi mFd y(xd)
influencing
– interfering input inputs
mFi
y(xi)

x
 Examples
• Temperature and a strain gauge

• Temperature and a photoresistor

• Temperature and humidity and a pressure sensor (changes in

the membrane properties)
 Example
• Knowing the modifying and interfering inputs can help us
compensate their influence
• Example: strain gauge
Rj = R in the absence of deformation
Rj
R j
R  K
Rj sensitivity
F
U0
R
R U out  U 0  1 KU 0    S  
R 4R 4
Influence of T
V
R j
Uout   R T U out,T   R T
U0 1
 KU 0  
Rj 4 4
Interfering input
 Example
• Knowing the modifying and interfering inputs can help us
compensate their influence
• Example: strain gauge
Influence of U0
Rj R
R
U out  U 0  U 0 
4R
U0 S  1 KU0 Modifying input
4
R R

V
Uout
 Static transfer characteristics
• Acquisition of static transfer • Components of the static
characteristics: transfer characteristics:
‐ All the inputs are kept Range
constant, except for one which Sensitivity
is varied stepwise (static Offset
calibration) Drift
‐ Measure the output after all Linearity
the transients have Hysteresis
disappeared Repeatability
‐ This is called static calibration Resolution
xd Fd mF d
Threshold
Stability
xm y
xi Fi mFi
 Measurement range
• Input range (input span): the interval within which the input
values can vary
Range  xmax  xmin

• Output range (output span)

• Measurement range – Output range = 10mA

sometimes also called full
scale (FS)

Input range = 100 bar

 Sensitivity
x F y

Calibration curve k
S’
S
y y

non‐linear
linear

x xo x’o x

dy
Sensitivity S
dx x x 0
 Offset
x F y

• Value of the output ( y0 ) for input

x=0
y  S  x  y0

• Error in terms of the input value:

y0
xoffset 
S
 Example: pressure sensor

Sensitivity: S = (14mA‐4mA)/100 bar = 0.1 mA/bar

Offset: y0 = 4mA (40 bar)
Example: specification sheet for a pressure
sensor

Full scale

Pressure range FS 0‐200 kPa

Offset Voff ±2 mV
Sensitivity ΔV/ΔP 0.2 mV/kPa
 Drift
• Slow (in terms of time) variations of the sensitivity or offset
‐ Example: instrument warming up
‐ Repeated
measurements give Condition 2
successively lower or y
higher results than
previous measurements Initial measurement
S(C2)
‐ Can be checked by
Condition 1
repeatedly performing
zero readings S(C1)

C2: the perturbation changes the sensitivity (modifying)

C1: the perturbation changes the offset (interfering)
 Example of drift
• Influence of the temperature on a pressure sensor

Offset drift: self‐heating of the strain gauge (resistor)

Sensitivity drift: heating of the membrane resulting in the
 Linearity
• Describes in what measure the sensitivity independent of the
measurand (input value) 1Sx (% of the read value)
Reference line
y
(least squares fit)

 y max
2Sxmax (% of the FS)

x
Full scale
y  yo  S  x  (1Sx or 2Sxmax )

Measuring Systems whichever is bigger
 Reference line – least squares fit
• Assume
ye  ax b
(x3, y 3)
• Minimize total distance ye  ax b

N
(x1, y1)
D   yi  axi  b 
2

i1

D D
 0, 0 (x2, y2)
a b

N N N N N N N

 i i N  i i N
1 1 1
x y  x y i i x 2

N i1 i1
y  x x y
i i i
a i1 i1 i1
2
b  y  ax  i1 i1
2
1 N  1 N 
N N

 xi  N   xi 
2
 x  N   xi 
2
i
i1  i1  i1  i1 
Example of linearity
Actual response

Idealised response

xmax
Linearity (%) 
FS
 Hysteresis
• Response depends on history xmax
Hysteresis (%) 
‐ Magnetic polarisation FS
‐ Piezoelectric polarisation
‐ Friction y

2ymax x

2xmax

FS
Example: pressure sensor datasheet

FS

Pressure range FS 0‐200 kPa

Offset Voff ±1 mV
Sensitivity ΔV/ΔP 0.2 mV/kPa
Linearity ±0.5% FS
Hysteresis ±0.5% FS
Example: pressure sensor datasheet

FS

Pressure range FS 0‐200 kPa

Offset Voff ±1 mV
Sensitivity ΔV/ΔP 0.2 mV/kPa
Temperature effect on FS T_FS ±2% FS ErrT,_FS = ±2·200/100
(0 to 50oC, Tref=25oC) = 4kPa
Temperature effect on Offset T_OFF ±1 mV ErrT,_OFF = ±1/0.2
(0 to 50oC, Tref=25oC) =5kPa
 Repeatability
• Distribution of successive
measurements of y under the same
conditions xmax
Repeatability (%) 
FS
y

Same conditions, measured at different times

x
 Stability
• The ability to maintain a response y for a constant x and during
a given time period

Pressure range FS 0‐200 kPa

Offset Voff ±1 mV
Sensitivity ΔV/ΔP 0.2 mV/kPa
Offset Stability ±0.5% FS

ErrStab = ± 0.5·200/100 = ±1kPa

 Maximal and probable error
• Maximal error
Errormax   xi
i

• Probable error

Errorprobable    i  
x 2

65% probability the actual error is within [‐,+]

98% probability the actual error is within [‐2,+2]
99% probability the actual error is within[‐3,+3]
 Resolution and threshold
• Resolution – the smallest detectable change of the input value
y

x resolution
• Threshold – resolution at the origin (input = 0)

Resolution at
x the origin
Transfer characteristics of conditioning circuits

• A measurement system is not just the sensor but the entire

measurement chain
• In order to determine the global transfer characteristics, one
must take into account also the transfer characteristics of
conditioning circuits
• Achieving the highest sensitivity of the circuit is a common
design goal

Noise
Sensor Conditioning reduction Acquisition
and signal
processing
 Example: voltage divider
Rs
U out U 0
Rs  R
Rs  Rs
U out  U out U 0 
Rs  Rs  R
Rs  Rs 1
U 0
Rs  R 1 Rs
Rs  R
For Rs  R  Rs
Rs  Rs  Rs 
U out  U out U 0 1  Linear with Rs
Rs  R  Rs  R 

Rs  Rs  Rs  Rs U0R

U out U 0 1  U 0  Rs
Rs  R  Rs  R  Rs  R Rs  R 2

Rs
Max sensitivity for R = R s. In that case U out U 0
4R
 Example: Wheatstone bridge
1 R
U out  U 0  U0
2 2R R
R
 U0
R
4(R  )
2
R
 U0
4R

‐No power supply noise for a balanced bridge

‐The effect of temperature can be compensated by choosing
resistors with the same temperature coefficient as the sensor
 Force sensor
• Based on a strain sensor attached to a test object

extension l / l  0

compression l / l  0
gauge
Deformation of the Resistance change Voltage drop
Force F
test object of the gauge in the circuit

l  F R l U R
K  I
l A Y R l U R
Wheatstone bridge – sensitivity optimisation

R
U out  U0
2R
R
U out  U0
4R

R R
U out  U U out  U0
R 0 2R