Lecture 3 Slides
Lecture 3 Slides
Actuators for
Automation
(EE 5210)
Lecture : 3
Logeeshan V. (Ph.D.)
Senior Lecturer
57
Linear Regression
• If measurements of the input stimuli during calibration cannot be
made with high accuracy and large random errors are expected, the
minimal number of measurements will not yield a sufficient
accuracy.
58
Computation of Stimulus
• A goal of sensing is to determine a value of the input stimulus s
from the value of the sensor output signal S. This can be done by
two methods.
59
Computation from Linear Piecewise
Approximation
• Both triangles are similar, which
allows us to derive a linear
equation for computing the
unknown stimulus from the
measured value of .
60
Computation from Linear Piecewise
Approximation
• This equation is easy to program and compute by an inexpensive
microprocessor, which keeps in memory a look-up table containing
the knots.
61
Computation from Linear Piecewise
Approximation
• Figure shows a thermistor
temperature sensor with
a pull-up resistor R1
connected to a 12-bit
analog-to-digital (A/D)
converter.
62
Computation from Linear Piecewise
Approximation
• After manipulating, we arrive at the inverted transfer function
which enables us to compute analytically the input temperature
in :
63
Computation from Linear Piecewise
Approximation
Assume that a pull-up resistor is selected during design. Let
us select two calibrating temperatures in the middle of the
operating range as and The
thermistor sequentially is immersed into a liquidbath at these two
temperatures and the 12−bit A/D counts are registered
respectively as and Calculate
64
Computation from Linear Piecewise
Approximation
• Since the sensor (thermistor) is
calibrated and its specific transfer
function is known, during the
sensor use, can be used for
computing temperature from any
reasonable A/D count.
65
Computation from Linear Piecewise
Approximation
• Since the sensor (thermistor) is calibrated and its specific transfer
function is known, during the sensor use, can be used for
computing temperature from any reasonable A/D count.
66
Computation from Linear Piecewise
Approximation
Calculate the temperature using the equation given bellow and find
the difference?
67
Iterative Computation of Stimulus
(Newton Method)
• Since the transfer function can be rewritten as
, numerical iterative methods for finding roots of this
(such as Newton or secant methods [1, 2]3) can also be used for
calculating the unknown stimulus s without the knowledge of the
inverse transfer function.
69
Newton Method
• To illustrate the Newton method for a 3rd degree polynomial
70
Newton Method
71
Newton Method
• In other words, the method will fail where the transfer function
flattens.
73
Span (Full-Scale Full-Scale Input)
• A dynamic range of stimuli that may be converted by a sensor is
called a span or an input full scale (FS).
75
Full-Scale Output
• Full-scale output (FSO) is the
algebraic difference
between the electrical
output signals measured
with maximum input
stimulus and the lowest
input stimulus applied.
definition, decibels are
equal to ten times the log of
the ratio of powers:
76
Performance Terms
• Accuracy : Accuracy is the extent to which the value indicated by a measurement system
or element might be wrong.
For example, a thermometer may have an accuracy of .
• Accuracy is often expressed as a percentage of the full range output or full-scale deflection
(f.s.d).
77
Performance Terms
• Error : The term error is used for the difference between the result of the measurement
and the true value of the quantity being measured.
• Thus if the measured value is when the true value is , the error is .
78
Performance Terms
• Error : The term error is used for the difference between the result of the measurement
and the true value of the quantity being measured.
• Thus if the measured value is when the true value is , the error is .
79
Error Types
• Hysteresis error : The term hysteresis error is used for the difference in outputs given
from the same value of quantity being measured according to whether that value has
been reached by a continuously increasing change or a continuously decreasing change.
80
Error Types
• Non-linearity error : The term non-linearity error is used for the error that occurs as a
result of assuming a linear relationship between the input and output over the working
range.
81
Error Types
• Insertion error : The act of attempting to make the measurement has modified the actual
parameter being measured. This effect is called loading and the consequence as an
insertion error.
82
Performance Terms
• Range: The range of variable of system is the limits between which the input can vary.
83
Performance Terms
• Precision: The term precision is used to describe the degree of freedom of a measurement
system from random errors.
84
Performance Terms
• Repeatability: The term repeatability is used for the ability of a measurement system to
give the same value for repeated measurements of the same value of a variable.
85
Performance Terms
• Reproducibility: The term reproducibility is used to describe the ability of a system to give
the same output when used with a constant input with the system or elements of the
system being disconnected from its input and then reinstalled.
86
Performance Terms
88
Dynamic Characteristics
• Stability: The dynamic characteristics refer to the behaviour between the time that the
input value changes and the time that the value given by the system or element settles
down to the steady state value.
89
Dynamic Characteristics
• Response time: This is the time which elapses after the input to a system or element is
abruptly increased from zero to a constant value up to the point at which the system or
element gives an output corresponding to some specified percentage.
90
Dynamic Characteristics
• Rise time: This is the time taken for the output to rise to some specified percentage of the
steady-state output.
91
Dynamic Characteristics
• Settling time: This is the time taken for the output to settle to within some percentage.
92
Reliability
• The probability of a particular event occurring is defined as being.
• The probability of the coin landing standing on edge can be considered to be zero, since
the number of occurrences of such an event is zero.
• The agreed level of performance might be that the measurement system gives a particular
accuracy.
94
Reliability
• A failure rate of 0.4 per year means that in one year, if ten systems are observed, 4 will fail
to meet the required level of performance. If 100 systems are observed, 40 will fail to
meet the required level of performance.
95
Reliability
• With a measurement system consisting of a number of elements, failure occurs when just
one of the elements fails to reach the required performance.
96
Requirements
• The main requirement of a measurement system is fitness for purpose.
• This means that if, for example, a length of a product has to be measured to a certain
accuracy that the measurement system is able to be used to carry out such a
measurement to that accuracy.
∓1𝑚𝑚 𝑎𝑐𝑐𝑢𝑟𝑎𝑐𝑦 97
Example
• A special sensor is designed to sense the percent vapor present in a liquid-vapor mixture. If
during a static calibration, the sensor indicates 80 units when in contact with 100% liquid, 0
units with 100% vapor, and 40 units with a 50:50% mixture. Determine the static sensitivity
of the sensor.
98
Example
• A microwave power sensor was calibrated using a faulty calibration source. The calibration
source was enabled as 1mw, but the actual power output of the calibration source was only
0.8 mW. This incorrectly calibrated power sensor was used to measure the output power of
an RF signal generator. The reading shown by the power meter was 10 dBm. What is the
actual output power of the RF signal generator in dBm.
99
Example
• A manufacturer constructed a pressure sensor. Pressure sensor was connected to a known
pressure source and the excitation for the sensors was provided by an ideal power supply.
Five reading at a pressure point was measured. The manufacture averaged the output of
the sensor at each of the measurement pressures. The result of these measurements is
given in the table next slide.
a). Fit the data to several polynomials of different order. What order of polynomial would you
recommend for fitting this data? Explain why you picked this order.
b). Using that order of polynomial what is the equation of the best fit polynomial?
c). What is the standard deviation of the error for your best fit polynomial?
d). The manufacturer would like to use a linear fit for this sensor between 10 KPa and
20Kpa.What is the best fit line for this region?
e). The manufacturer would like to use a linear fit for this sensor between 100KpA and
120Kpa.What would you recommend for the equation of the line in this region?How would
you determine the best fit line in this region given you only have 1 measured point in this
range? 100
Pressure (Kpa) Output Voltage (V)
12.3 0.06642
14.1 0.06861
15.8 0.06874
17.9 0.07303
22.1 0.07638
45.6 0.1517
78.9 0.4756
95.9 0.7588
109.6 1.110
122.4 1.491
139.3 2.166
156.7 2.930
178.9 4.464
Example
• A temperature sensor is calibrated at 5 different temperatures as shown in the table below.
Using this data, derive the best linear calibration curve for this sensor, sketch the calibration
and show the mathematical method for finding the best line. Report the line slope and
intercept. From your result, report the non linearity error from this sensor.
102
Example
• A platinum resistance sensor is to be used to measure temperatures between 0 and 200
degree Celsius. During the calibration process, the engineer obtained the following
readings.
103
Example
104