3.

Sensor characteristics
Static sensor characteristics

Relationships between output and input signals of the sensor in conditions

of very slow changes of the input signal determine the static sensor
characteristics.

Some important sensor characteristics and properties include:

• transfer function, from which a sensitivity can be determined
• span (input full scale) and FSO (full scale output)
• calibration error
• hysteresis
• nonlinearity
• repeatability
• resolution and threshold

1
 Transfer function

An ideal relationship between stimulus (input) x and sensor output y is called

transfer function.
The simplest is a linear relationship given by equation

y = a + bx (3.1)
The slope b is called sensitivity and a (intercept) – the output at zero input.
Output signal is mostly of electrical nature, as voltage, current, resistance.
Other transfer functions are often approximated by:

logarithmic function y = a + b lnx (3.2)

exponential function y = a ekx (3.3)
power function y = ao + a1xc (3.4)

In many cases none of above approximatios fit sufficiently well and higher
order polynomials can be employed. For nonlinear transfer function the
sensitivity is defined as
S = dy/dx (3.5)
and depends on the input value x.

2
 Sensitivity

dy
Sx 
dx
a
y b x X

Measurement error Δx of
 y quantity X for a given Δy can
be small enough for a high
sensitivity.

x
 xa  xb

Over a limited range, within specified accuracy limits, the nonlinear transfer
function can be modeled by straight lines (piece-wise approximation). For these
linear approximations the sensitivity can be calculated by
S = Δy/Δx

3
 Span and full scale output (FSO)

An input full scale or span is determined by a dynamic range of stimuli which

may be converted by a sensor,without unacceptably high inaccuracy.
For a very broad range of input stimuli, it can be expressed in decibels,
defined by using the logarithmic scale. By using decibel scale the signal
amplitudes are represented by much smaller numbers.

For power a decibel is defined as ten times the log of the ratio of powers
1dB = 10 log(P/P0)

Similarly for the case of voltage (current, pressure)

one introduces
1dB = 20 log(V2/V1)

Full scale output (FSO) is the difference between

ouput signals for maximum and minimum stimuli
respectively. This must include deviations from
the ideal transfer function, specified by ± Δ.

4
Calibration error

Calibration error is determined by innacuracy

permitted by a manufacturer after calibration of
a sensor in the factory.

To determine the slope and intercept of the

function one applies two stimuli x1 and x2 and
the sensor responds with A1 and A2. The higher
signal is measured with error – Δ.
This results in the error in intercept (new
intercept a1, real a)

δa = a1 – a = Δ /(x2-x1)

and in the error of the slope

δb = Δ /(x2-x1)

5
 Hysteresis

y  - Max. difference at
output for specified
input

This is a deviation of the sensor output, when it is

approached from different directions.

6
 Nonlinearity

±% FSO

Best straight line

This error is specified for sensors, when the nonlinear transfer function is
approximated by a straight line. It is a maximum deviation of a real transfer
function from the straight line and can be specified in % of FSO.
The approximated line can be drawn as the so called „best straight line”
which is a line midway between two parallel lines envelpoing output values of
a real transfer function.
Another method is based on the least squares procedure.
7
 Repeatability

y Cycle 1
Cycle 2


Δ - max. difference
between output readings
for the same direction

This error is caused by sensor instability and can be expressed as the

maximum difference between output readings as determined by two
calibration cycles, given in % of FSO.

δr = Δ / FSO
8
 Rsolution and threshold

y
 resolution

threshold

Threshold is the smallest increment of stimulus which gives noticeable

change in output.
Resolution is the step change at output during continuous change of input.

9
 Dynamic characteristics

When the transducing system consists of linear elements dissipating and

accumulating energy, then the dependence between stimulus x and output signal
y can be written as equality of two differential equations

A0y + A1y(1) + A2y(2) + ... + Any(n) = k (B0x + B1x(1) + B2x(2) + ... Bmx(m)) (1)

y(1) – 1-st derivative vs. time

k – static sensitivity of a transducer
m≤n

Eq. (1) can be transformed by the Laplace integral transformation


F ( s)  L f (t )   e  st f (t )dt (2)
0

where s = σ + jω

10
 Integrating (2) by parts it is easy to show that

 df (t ) 
  sL f (t )  f (0)
(3)
L
 dt 

Transforming eq. (1) using Laplace transformation, with the help of property
(3) and with zero initial conditions one obtains the expression for operator
transmittance of the sensor

Y ( s) 1  B1 s  B2 s 2  ...Bm s m
K ( s)  k
X ( s) 1  A1 s  A2 s 2  ... An s n

In effect we transfer from differential to algebraic equations. The analysis of

an operator transmittance is particularly useful when the transducer is built
as a measurement chain.
The response y(t) one obtains applying reverse Laplace transformation.

11
 Excitation by a step-function

x(t) x(t) = 1(t)

0 for t < 0
1
1(t) =
1 for t  0

The response of a sensor system depends on its type.

It can be an inercial system, which consists of accumulation elements of one type
(accumulating kinetic or potencial energy) and dissipating elements.

12
 An example of inertial transducer

Resistance thermometer
immersed in the liquid
of elevated temperature
Electric analog

L{1(t)} = X(s) = 1/s

1 k 1
Y ( s)  K ( s)   Inertial element of the 1-st order
s 1  s s

13
 Inertial element of the 1-st order,
calculation of operator transmittance

 
f(s) e  st
f ( t ) dt   e  st k ( 1  e t /  ) dt
0 0
 
 1 
f(s)/ k  e  st
dt   e ( stt /  ) dt  1 / s e  st  e ( s 1 /  )t
0 0
0 s 1/ 0

1 1 1/ s
  
s s 1/ 1  s
k 1
f(s) 
1  s s

Therefore

Y( s ) k
K( s )  
1 / s 1  s

14
 Time response to the step function of inertial element of
the 1-st order

y(t )  L1Y (s)  k (1  et / )

y(t) 
k  τ – time constant, a measure of
sensor thermal inertia;

For electric analogue τ = RC

y(    k ( 1  1 / e )  0.6321 k what means for t = τ 63% of a steady value.

For t = 3τ one gets 95% of a steady value.

15
 Response to the step function of an inertial element
of higher order

y(t)

k
95%
90%

t95 - 95% response time

63%
Δt = t90 – t10 – rise time
τ = t63 – time constant

10%

0 t10  t90 t95 t

Insertion of a thermometer into an insulating sheath transforms it

into an inertial element of higher order.
16
 Response of an oscillating system to the step function

1
y(t)
1 – oscillations
K
2 – critical damping
3 – overdamping
3
2

0 t

Tranducer of the oscillation type consists of accumulating elements of both types

and dissipating elements.
Mechanical analogue is a damped spring oscillator (the spring accumulates
potential energy, the mass – kinetic energy, energy is dissipated by friction).
Electric analogue is an RLC circuit.
17
 RS
L
kS
u1(t)
u(t) C q = CU1(t) = kU1(t)
R
y m

F(t)

q  2q  02 q  02Cu(t )

Rs 1 F( t ) 1
y  y  y 02 
m mks m LC
R

Mechanic analogue 2L
Electric analogue

18
 Transmitance of the oscillating system
k 02
K ( s)  2
s  2s   02

The denominator of expression for transmittance can have:

1) Two real roots

s1, 2     2  02 overdamping

2) One real root

s   critical damping

3) Two complex roots

underdamping
s1, 2    j 02   2    j
(oscillations)

After inverse Laplace transformation one gets: 

y(t )  k 1  e t sin( t   ) 
19