SENSORS PERFORMANCE

CHARACTERISTICS

Setyawan P. Sakti Dept. of Physics Brawijaya University

Input and Output

Input:

stimulus or measurand (temperature pressure, light

intensity, etc.)
Ouput: electrical signal (voltage, currentfrequency, phase, etc.)
Variations: output can sometimes be displacement
(thermometers, magnetostrictive and piezoelectric
sensors). Some sensors combine sensing and actuation

Setyawan P. Sakti Dept. of Physics Brawijaya University

Transfer function

Relation between input and output

Other names:
Input

output characteristic function

transfer characteristic function
Response characteristic

Setyawan P. Sakti Dept. of Physics Brawijaya University

Type of Transfer function

Linear or nonlinear
Single valued or not
One dimensional or multi dimensional
Single input, single output
Multiple inputs, single output

In most cases:
Difficult to describe mathematically (given graphically)
Often must be defined from calibration data
Often only defined on a portion of the range of the device

Setyawan P. Sakti Dept. of Physics Brawijaya University

Transfer function approximation

T1 to T2 - approximately linear
Most useful range
Typically a small portion of the range
Often taken as linear

S = f(x)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Data taken from Transfer function

Other data from transfer function

saturation
sensitivity
full scale range (input and output)
hysteresis
deadband
etc.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Transfer function (cont.)

Other types of transfer functions

Response with respect to a given quantity

Impulse response, Step response, Linier response, etc

Performance characteristics (reliability curves, etc.)

Viewed as the relation between any two characteristics

Setyawan P. Sakti Dept. of Physics Brawijaya University

Impedance and impedance matching

Black box theory of two port devices

Input impedance: ratio of the rated voltage and the resulting
current through the input port of the device with the output
port open (no load)
Output impedance: ratio of the rated output voltage and
short circuit current of the port (i.e. current when the output is
shorted)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensor Impedance

Only output impedance of the sensor is relevant

As it is mainly in electrical circuit, hence electrical impedance
is importance especially for further processing of signals

Setyawan P. Sakti Dept. of Physics Brawijaya University

Importance of Impedance

Impedance affects sensor (system) performance

Example: 500 W sensor (output impedance) connected to
a processor
Processor input impedance is infinite (Fig. b)
Processor input impedance is 500 W (Fig. c)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Strain Gauge Impedance Example

Case: impedance is 500 W at zero strain, 750 W at

measured strain
a.
b.

sensor output: 2.5V (at zero strain), 3V at measured strain

sensor output: 1.666V to 1.875V

Result:

Case b:

the sensor is loaded

Sensitivity is reduced (smaller output change for the same strain input)

Case a:

Loading effect not exists (infinite impedance of the processor)

Better sensitivity than case b

Setyawan P. Sakti Dept. of Physics Brawijaya University

Impedance matching

To avoid loading effect

Often need a matching circuit : from high to low or from
low to high impedancesmo
Sometimes can be done directly (C-MOS
devices/processors have very high input impedances)
Voltage output: impedance is high need high impedance
at processor
Current output: impedance is low - need low impedance at
processor

Setyawan P. Sakti Dept. of Physics Brawijaya University

Why Impedance Matching ?

Achieve maximum signal transmission with less (without)

reflection
In addition to the previous:
Conjugate

matching (Zin=Zout*) - maximum power transfer

Zin=R+jX,

No

Zout*=R-jX.

reflection matching (Zin=Zout) - no reflection from load

Important

at high frequencies (transmission lines)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Range and Span

Range: lowest and highest values of the stimulus

Span: the arithmetic difference between the highest and
lowest values of the stimulus that can be sensed within
acceptable errors
Input full scale (IFS) = span
Output full scale (OFS): difference between the upper and
lower ranges of the output of the sensor corresponding to
the span of the sensor
Dynamic range: ratio between the upper and lower limits
and is usually expressed in db

Setyawan P. Sakti Dept. of Physics Brawijaya University

Range and Span (Cont)

Example: a sensors is designed for: -30 C to +80 C to

output 2.5V to 1.2V
Range: -30C and +80 C
Span: 80- (-30)=110 C
Input full scale = 110 C
Output full scale = 2.5V-1.2V=1.3V
Dynamic range=20log(140/30)=13.38db

Setyawan P. Sakti Dept. of Physics Brawijaya University

Range and Span (cont.)

Range, span, full scale and dynamic range may be

applied to actuators in the same way
Span and full scale may also be given in db when the
scale is large.
In actuators, there are other properties that come into
play:
Maximum

force, torque, displacement

Acceleration
Time response, delays, etc.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Accuracy, errors, repeatability

Errors: deviation from ideal

Sources:
materials

used
construction tolerances
ageing
operational errors
calibration errors
matching (impedance) or loading errors
noise
many others
Setyawan P. Sakti Dept. of Physics Brawijaya University

Accuracy, errors (cont.)

Error:
a.

As a difference: e = V V0 (V0 is the actual value, V is that

measured value (the stimulus in the case of sensors or output in
actuators).
b. As a percentage of full scale (span for example) e =
t/(tmax-tmin)*100 where tmax and tmin are the maximum and
minimum values the device is designed to operate at.
c. In terms of the output signal expected.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example: errors
Example: A thermistor is used to measure temperature
between 30 and +80 C and produce an output voltage
between 2.8V and 1.5V. Because of errors, the accuracy in
sensing is 0.5C.

VO 2.8

Setyawan P. Sakti Dept. of Physics Brawijaya University

1.5- 2.8
80-( -30)

Example (cont)
a. In terms of the input as 0.5C
b. Percentage of input: e = 0.5/(80+30)*100 = 0.454%
c. In terms of output. From the transfer function: e= 0.059V.

Setyawan P. Sakti Dept. of Physics Brawijaya University

More on errors
Static errors: not time dependent
Dynamic errors: time dependent
Random errors: Different errors in a parameter
or at different operating times
Systemic errors: errors are constant at all times
and conditions

Setyawan P. Sakti Dept. of Physics Brawijaya University

Error limits - linear TF

Linear transfer functions

Error

equal along the transfer function

Error increases or decreases along TF
Error limits - two lines that delimit the output

Setyawan P. Sakti Dept. of Physics Brawijaya University

Error limits - nonlinear TF

Nonlinear transfer functions

Error

change along the

transfer function
Maximum error from ideal
Average error
Limiting curves follow ideal
transfer function

Setyawan P. Sakti Dept. of Physics Brawijaya University

Error limits - nonlinear TF

Calibration curve may be

used when available
Lower

errors
Maximum error from
calibration curve
Average error
Limiting curves follow the
actual transfer function
(calibration)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Repeatability

Also called reproducibility: failure of the sensor or

actuator to represent the same value (i.e. stimulus or input)
under identical conditions when measured at different
times.
usually

associated with calibration

viewed as an error.
given as the maximum difference between two readings taken
at different times under identical input conditions.
error given as percentage of input full scale.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensitivity

Sensitivity of a sensor is defined as the change in output

for a given change in input, usually a unit change in input.
Sensitivity represents the slope of the transfer function.
Same for actuators

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensitivity

Sensitivity of a sensor is defined as the change in output

for a given change in input, usually a unit change in input.
Sensitivity represents the slope of the transfer function.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensitivity (cont.)

Can be highly nonlinear along the transfer function

Measured in units of output quantity per units of input
quantity (W/C, N/V, V/C, etc.)
For a linear transfer function, sensitivity is the slope or
transfer function.
Example:
d aT + b = 1
dR = a
W

dR
dT
C

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensitivity analysis (cont.)

A difficult task
there

is noise
a combined function of sensitivities of various
components, including that of the transduction
sections.
device may be rather complex with multiple
transduction steps, each one with its own sensitivity,
sources of noise and other parameters
some properties may be known but many may not
be known or may only be approximate.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Sensitivity analysis (cont.)

An important task
provides

information on the output range of signals one can

expect,
provides information on the noise and errors to expect.
may provide clues as to how the effects of noise and errors
may be minimized
Provides clues on the proper choice of sensors, their connections
and other steps that may be taken to improve performance
(amplifiers, feedback, etc.).

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example - additive errors

Fiber optic pressure sensor

Pressure

changes the length of the fiber

This changes the phase of the output
Three transduction steps

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example-1 - no errors present

Individual sensitivities

Overall sensitivity

dy1
dy2
dy3
s1 = , s 2 = ,
s3 =
dx1
dx2
dx3
dy1 dy2 dy3
S = s 1s2 s3 =
dx1 dx2 dx3

But, x2=y1 (output of transducer 1 is the input to

transducer 2) and x3=y2
dy3

S = s 1s2 s3 =

Setyawan P. Sakti Dept. of Physics Brawijaya University

dx1

Example -1 - errors present

First output is y1=y01 + y1. y01 = Output without error

2nd output

y2 = s2 y10 + y1 + y2 = y20 + s 2y1 + y2

3rd output

y3 = s3 y20 + s2y1 + y2 + y3 = y30 + s2s3y1 + s3y2 + y3

Last 3 terms - additive errors

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example -2 - differential sensors

Output proportional to difference between the outputs of

the sensors
Output is zero when T1=T2
Common mode signals cancel (noise)
Errors cancel (mostly)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example -2 - (cont.)

dy1
dy2
s1 = ,
s2 =
dx1
dx2
y = y1 - y2 = s1x1 - s 2yx
d y1 - y2
s=
d x1 - x2

Setyawan P. Sakti Dept. of Physics Brawijaya University

Example -3 - sensors in series

Output is in series
Input in parallel (all sensors at
same temperature)
Outputs add up
Noise multiplied by product of
sensitivities

y = y1 + y2 + y3 + ... + yn = (s1 + s2 + s3 + ... + s n)x = nsx

S = ns
Setyawan P. Sakti Dept. of Physics Brawijaya University

Hysteresis

Hysteresis (literally lag)- the

deviation of the sensors
output at any given point
when approached from two
different directions
Caused by electrical or
mechanical systems
Magnetization
Thermal properties
Loose linkages

Setyawan P. Sakti Dept. of Physics Brawijaya University

Hysteresis - Example

If temperature is measured, at a rated temperature of

50C, the output might be 4.95V when temperature
increases but 5.05V when temperature decreases.
This is an error of 0.5% (for an output full scale of 10V in
this idealized example).
Hysteresis is also present in actuators and, in the case of
motion, more common than in sensors.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Nonlinearity

A property of the sensor (nonlinear transfer function) or:

Introduced by errors
Nonlinearity errors influence accuracy.
Nonlinearity is defined as the maximum deviation from the
ideal linear transfer function.
The latter is not usually known or useful
Nonlinearity must be deduced from the actual transfer
function or from the calibration curve
A few methods to do so:

Setyawan P. Sakti Dept. of Physics Brawijaya University

Nonlinearity (cont.)

a. by use of the range of the sensor

Pass

a straight line between the range points (line

1)
Calculate the maximum deviation of the actual
curve from this straight line
Good when linearities are small and the span is
small (thermocouples, thermistors, etc.)
Gives an overall figure for nonlinearity

Setyawan P. Sakti Dept. of Physics Brawijaya University

Nonlinearity (cont.)

b. by use of two points defining a portion of the span of

the sensor/actuator.
Pass

a straight line between the two points

Extend the straight line to cover the whole span
Calculate the maximum deviation of the actual curve from this
straight line
Good when a device is used in a small part of its span (i.e. a
thermometer used to measure human body temperatures
Improves linearity figure in the range of interest

Setyawan P. Sakti Dept. of Physics Brawijaya University

Nonlinearity (cont.)

c. use a linear best fit(least squares) through

the points of the curve
Take

n points on the actual curve, xi,yi, i=1,2,n.

Assume the best fit is a line y=ax+b (line 2)
Calculate a and b

Setyawan P. Sakti Dept. of Physics Brawijaya University

Nonlinearity (cont.)

d. use the tangent to the curve at some point on the curve

Take a point in the middle of the range of interest

Draw the tangent and extend to the range of the curve (line 3)
Calculate the nonlinearity as previously
Only useful if nonlinearity is small and the span used very small

Setyawan P. Sakti Dept. of Physics Brawijaya University

Saturation

Saturation a property of sensors or actuators when they no

longer respond to the input.
Usually at or near the ends of their span and indicates that
the output is no longer a function of the input or, more likely
is a very nonlinear function of the input.
Should be avoided - sensitivity is small or nonexistent

Setyawan P. Sakti Dept. of Physics Brawijaya University

Saturation Condition
saturated

saturated

Setyawan P. Sakti Dept. of Physics Brawijaya University

Frequency response

Frequency response: The ability of the device to respond to

a harmonic (sinusoidal) input
A plot of magnitude (power, displacement, etc.) as a function
of frequency
Indicates the range of the stimulus in which the device is
usable (sensors and actuators)
Provides important design parameters
Sometimes the phase is also given (the pair of plots is the
Bode diagram of the device)

Setyawan P. Sakti Dept. of Physics Brawijaya University

Frequency response (cont)

Important design parameters

Bandwidth (B-A, in Hz)
Flat frequency range (D-C in Hz)
Cutoff frequencies (points A and B in Hz)
Resonant frequencies

Setyawan P. Sakti Dept. of Physics Brawijaya University

Frequency response (cont.)

Bandwidth: the distance in Hz between the half power

points
Half-power

points: eh=0.707e, ph=0.5p

Flat response range: maximum distance in Hz over which

the response is flat (based on some allowable error)
Resonant frequency: the frequency (or frequencies) at
which the curve peaks or dips

Setyawan P. Sakti Dept. of Physics Brawijaya University

Half power points

Also called 3db points

Power is 3db down at these points:
10*log0.5=-3db

or
20*log (sqrt(2)/2)=-3db

These points are arbitrary but are now standard.

It is usually assumed that the device is useless beyond the
half power points

Setyawan P. Sakti Dept. of Physics Brawijaya University

Frequency response (example.)

Bandwidth: 16.5kHz-70Hz=16.43 kHz

Flat frequency range: 10kHz-120Hz=9880 Hz
Cutoff frequencies: 70 Hz and 16.5 kHz
Resonance: 12 kHz

Setyawan P. Sakti Dept. of Physics Brawijaya University

Response time

response time (or delay time), indicates the time needed for
the output to reach steady state (or a given percentage of
steady state) for a step change in input.
Typically the response time will be given as the time needed
to reach 90% of steady state output upon exposure to a unit
step change in input.
The response time of the device is due to the inertia of the
device (both mechanical and electrical).

Setyawan P. Sakti Dept. of Physics Brawijaya University

Response time (cont.)

Example: in a temperature sensor:

the time needed for the sensors body to reach the temperature it is
trying to measure (thermal time constant) or
The electrical time constants inherent in the device due to
capacitances and inductances
In most cases due to both

Example: in an actuator:
Due to mass of the actuator and whatever it is actuating
Due to electrical time constants
Due to momentum

Setyawan P. Sakti Dept. of Physics Brawijaya University

Response time (cont.)

Fast response time is usually desirable (not always)

Slow response times tend to average readings
Large mechanical systems have slow response times
Smaller sensors and actuators will almost always respond
faster
We shall meet sensors in which response time is slowed down
on purpose

Setyawan P. Sakti Dept. of Physics Brawijaya University

Calibration

Calibration: the experimental determination of the transfer

function of a sensor or actuator.
Typically, needed when the transfer function is not known or,
When the device must be operated at tolerances below
those specified by the manufacturer.
Example, use a thermistor with a 5% tolerance on a full scale
from 0 to 100C to measure temperature with accuracy of,
say, 0.5C.
The only way this can be done is by first establishing the
transfer function of the sensor.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Calibration Method

a. known transfer function:

Determine the slope and crossing point (line function) from two known stimuli (say two
temperatures) if the transfer function is linear
Measure the output
Calculate the slope and crossing point in V=aT+b
If the function is more complex, need more points: V = aT + bT2 + cT3 + d
4 measurements to calculate a,b,c,d
Must choose points judiciously - if linear, use points close to the range. If not, use equally
spaced points or points around the locations of highest curvature

b. Unknown transfer function:

Measure the output Ri at as many input values Ti as is practical

Use the entire span
Calculate a best linear fit (least squares for example)
If the curve is not linear use a polynomial fit
May use piecewise linear segments if the number of points is large.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Calibration (cont.)

Calibration is sometimes an operational requirement

(thermocouples, pressure sensors)
Calibration data is usually supplied by the manufacturer
Calibration procedures must be included with the design
documents
Errors due to calibration must be evaluated and specified

Setyawan P. Sakti Dept. of Physics Brawijaya University

Resolution

Resolution: the minimum increment in stimulus to which it can

respond. It is the magnitude of the input change which results
in the smallest discernible output.
Example: a digital voltmeter with resolution of 0.1V is used
to measure the output of a sensor. The change in input
(temperature, pressure, etc.) that will provide a change of
0.1V on the voltmeter is the resolution of the
sensor/voltmeter system.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Resolution (cont.)

Resolution is determined by the whole system, not only by the

sensor
The resolution of the sensor may be better than that of the
system.
The sensor itself must interact with a processor, the limiting
factor on resolution may be the sensor or the processor.
Resolution may be specified in the units of the stimulus (0.5C
for a temperature sensor, 1 mT for a magnetic field sensor,
0.1mm for a proximity sensor, etc) or may be specified as a
percentage of span (0.1% for example).

Setyawan P. Sakti Dept. of Physics Brawijaya University

Resolution (cont.)

In digital systems, resolution may be specified in bits (1 bit or

6 bit resolution)
In analog systems (those that do not digitize the output) the
output is continuous and resolution may be said to be
infinitesimal (for the sensor or actuator alone).
Resolution of an actuator is the minimum increment in its
output which it can provide.
Example: a stepper motor may have 180 steps per
revolution. Its resolution is 2.
A graduated analog voltmeter may be said to have a
resolution equal to one graduation (say 0.01V). ( higher
resolution may be implied by the user who can easily
interpolated between two graduations.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Other parameters

Reliability: a statistical measure of quality of a device

which indicates the ability of the device to perform its
stated function, under normal operating conditions without
failure for a stated period of time or number of cycles.
Given in hours, years or in MTBF
Usually provided by the manufacturer
Based on accelerated lifetime testing

Setyawan P. Sakti Dept. of Physics Brawijaya University

Other parameters

Deadband: the lack of response or insensitivity of a device

over a specific range of the input.
In this range which may be small, the output remains
constant.
A device should not operate in this range unless this
insensitivity is acceptable.
Example, an actuator which is not responding to inputs
around zero may be acceptable but one which freezes
over a normal range may not be.

Setyawan P. Sakti Dept. of Physics Brawijaya University

Other parameters

Excitation: The electrical supply required for operation of a sensor or

actuator.
It may specify the range of voltages under which the device should
operate (say 2 to 12V), range of current, power dissipation, maximum
excitation as a function of temperature and sometimes frequency.
Part of the data sheet for the device
Together with other specifications it defines the normal operating
conditions of the sensor.
Failure to follow rated values may result in erroneous outputs or
premature failure of the device.

Setyawan P. Sakti Dept. of Physics Brawijaya University