A measurement system takes an input and

transforms it into an output

e output can be obse ed or recorded, such as
the movement of a pointer or the magnitude of a
digital display
e shape and form of a signal : waveform.
T he waveform contains information about the
magnitude and amplitude, which indicate the
size of the input quantity, and the frequency,
which indicates the way the signal changes in
time.
A signal is the physical information about a
measured variable. It can be;
◦ transmitted between a process and the measurement
system,
◦ between the stages of a measurement system, or
◦ the output from a measurement system.
e form of a signal can be analog, discrete time,
or digital
Analog signal is continuous in time.
It provides a ready representation of their time-
dependent behavior.
e magnitude of the signal is continuous
Aatdiscrete time signal usually
repeated nite time inte als.results from the sampling of a continuous variable
Information
in time. about the magnitude of the signal is available only at discrete points
Some assumption must be made about
during the times when it is not available. the behavior of the measured variable
One approach is to assume the
and hold method (Figure 2.3b). signal is constant between samples, a sample
A digital signal exists at discrete values in time, like a discrete time signal.
Magnitude of a digital signal is discrete, determined by a process known as quantization at each discrete point in time.
Figure
ce ain2.4a showsvalues.
discrete digital and analog forms of the same signal where the magnitude of the digital signal can have only
e waveform that would result from assuming that the signal is constant between sampled points in time is shown in Figure
2.4b.
Sampling of andevice
a solid-state analogthatsignal
conveto sproduce a digital
an analog voltagesignal
signalcantobea bina
accomplished
numberbysystem
usingrepresentation.
an analog-to-digital (A/D) conve er,
Signals may be characterized as either static or
dynamic.
A static signal does not va with time (for example
diameter of a shaft).
Or, outdoor temperature does not change signi cantly
in a matter of minutes, this input signal might be
considered static when compared to our time period of
interest.
A mathematical representation of a static signal is
given by a constant, as indicated in Table 2.1.
Awaveforms,
dynamic signal is de ned as a time-dependent
y(t), may be classi ed as shown in Table 2.1.) signal, y(t). (dynamic signal
Awave,
deterministic signal varies in time in a predictable manner,
a step function, or a ramp function, as shown in Figure 2.5. such as a sine
Arepeats
signalatis regular
steady periodic
inte als if
in the variation
time(ideal of the
pendulum, magnitude
and the of the signal
temperature
variations in the
operating conditions). cylinder of an internal combustion engine under steady
Awaveform
simple periodic
contains waveform
multiple contains
frequencies only one
and is frequency.
represented A complex
as a periodic
superposition of multiple simple periodic waveforms.
Aperiodic is the term used to describe deterministic
repeat at regular inte als, such as a step function. signals that do not
Nondeterministic signal has no distinct pattern.
Nondeterministic signals are generally described by their
statistical characteristics
Consider the analog signal shown in Figure 2.6a, which is
continuous over the recorded time period from t1 to t2.
e average or mean value of this signal is found by
the rms value of any continuous analog variable y(t) over the time,
t2-t1, is expressed as
(mean value)

Root-Mean-
Square Value
It provides a measure of the static po ion of a signal over the time t2
- t1 (it is sometimes called the DC component or DC o set of the
signal).
e mean value does not provide any indication of the amount of
variation in the dynamic po ion of the signal. Root mean square gives
an indication for the uctuations of the signal.
A complex signal can be approximated as an in nite series of sine
and cosine functions.
e method of expressing such a complex signal as a series of sines
and cosines is called Fourier analysis.
e study of Fourier series is a branch of Fourier analysis. Fourier
series were introduced by Joseph Fourier (1768–1830) for the
purpose of solving the heat equation in a metal plate.
Today many modern software tools such as Matlab, Labview has
built in functions for Fourier analysis.
 e rst four Fourier series
approximations for a square
wave.
 Example : a sawtooth wave

Plot of a periodic identity function—a sawtooth wave.

Animated plot of the rst ve successive pa ial Fourier series.

Simple harmonic motion can ser ve as a
mathematical model of a variety of motions,
such as the oscillation of a spring.
Also, other motions can be approximated by
simple harmonic motion, including the motion of
a simple pendulum.
Period : e time required for the mass to nish
one complete cycle of the motion is called the
period. Generally represented by the symbol T.
Frequency : the number of complete cycles of
the motion per unit time. is frequency, f, is
measured in cycles per second (Hz; 1 cycle/s).
e term  is also a frequency, but instead of
having units of cycles per second it has units of
radians per second.
Complex waveforms present di culties in the
selection of a measurement system and in the
interpretation of an output signal.
For example what is the fundamental natural
frequency of red signal?, how many frequencies
does the signal convey? etc.
Let y(t) be periodic signal with period
ω =2π.f =2π/T
en y can be expressed as a sum of sine and
cosine functions in nω, n=0,1,2,3,…
n=0 DC
n=1 fundamental
n=2 second harmonic
n=3 third harmonic
Ao=2/T∫0T y(t)dt
Fourier analysis allows essentially all mathematical functions of practical interest to
be represented by an in nite series of sines and cosines
e coe cients of a trigonometric series representing a function having an
arbitra period T are given by the Euler formulas:
Example 4.1
Solution
Ao=0 no DC
e fundamental frequency ω=80
A1=0 (no cos(80t)
Since cos(a+b)=cos(a)cos(b)-sin(a)sin(b)
then :
cos(160t-π/4)=0.707cos(160t)+0.707sin(160t)
Hence the coe cients of the second harmonics are:
A2=50*0.707=35.35
B2=50*0.707=35.35
Example 4.2
Find the Fourier coe cients of the periodic
signal
y(t)= A 0< ωt <π
=-A π< ωt <2π
Solution:
T=2π/ω; T
Ao=1/T∫0 y(t)dt=1/T [∫0 T/2 Adt- ∫T/2 T Adt]
=0
An=2/T[∫0T/2 Acos(nωt)dt- ∫T/2T Acos(nωt)dt
=2A/nωT[sin(nπ)-sin(0)-sin(2nπ)+sin(nπ)]
=0
Bn=2/T[∫0T/2 Asin(nωt)dt- ∫T/2T Asin(nωt)dt
=-2A/nωT[cos(nπ)-cos(0)-cos(2nπ)+cos(nπ)]
For n even Bn=0
For n odd Bn=4A/nπ
e function of square wave

e spectrum of the square wave

4A/π

(4/3)A/π
(4/5)A/π
1st 2nd 3rd
If two waves of equal amplitude and nearly equal
frequency are added, the resulting wave
undergoes slow beats where the amplitude rises
and falls with a cyclic frequency=half their
frequency di erence.
y = A sin(2πfot) + A sin(2π(fo+∆f)t)
= 2 A cos(2π∆ft/2) sin(2π(2fo+∆f)t/2)
Beat frequency of two waves
Helium-neon laser light has a frequency of 473.8
THz a helium-neon laser is re ected from a
moving target. is creates a Doppler shift in the
beam which increases its frequency by 3MHz. If
the re ected wave is added to the unshifted
beam of equal intensity what is the resulting
beat frequency.
Solution 1.5MHz
Generally, y(t) is measured and recorded, then it will
be stored in the form of a discrete time or digital
signal.
Techniques for approximation to the Fourier
transform integral are used in practice : Discrete
Fourier Transform (DFT)
A fast algorithm for computing the DFT, known as
the Fast Fourier Transform(FFT),was developed
In many software applications a built-in function is
available (Matlab - t command, Labview etc.)
Natural frequencies
characteristics. of systems present dif f
e rent
For example a single degree of freedom mass
oscillator and two degrees of freedom can explain these. spring
Texcitation
he se systems will have dif f
e rent response to
(with one natural and two natural frequencies). an
Consequently
ways. these systems will resonate in dif f
e rent
FFT
these analysis and plots
characteristics bycan be
using a useful
the timetool to represent
response of the
systems (such as amplitudes and frequency contents).
Normally a computer will read signal at time
inte als of ∆t.
e computer will record a total of N points
(samples) over the time period N∆t.
y(∆t), y(2 ∆t),…,y(N∆t)
Or y(tr), r=1,2,3,..,N tr=r ∆t.
To pe orm a Fourier series analysis of a discrete
time series we conve integral into sum:
t is replaced by tr and T is replaced by N∆t
 e harmonic coe cients of a discretely sampled
wave form are
2 N 2 
A n   y ( tr ) cos( nr ) n  0 ,1,..., N
N r1 N 2
2 N 2  N
B n  N  y ( tr ) sin( N nr ) n  1,.., 2  1
r 1


ese are called discrete Fourier

transform (DFT) of y(tr)
And the discrete Fourier series is

A N /2 2  N /2 1

2 
y ( tr )  2   An cos( N nr )   B n sin( N nr )
o
n 1 n 1

 e fundamental cyclic frequency is ∆f
2rn  2  n r t  2 ( n f )
N  N t 
t r

• e fundamental cyclic frequency is

∆f
f 
1
N t
 and the discrete Fourier series is
A o
 N /2 1
y ( t )   n1  A n cos( 2n ft )  B n sin( 2n ft ) 


2 


A N / 2 
cos  2 N f 
t
2  2 
Which is the same as Eq 4.18 and equivalent to Eq 4.16 (see textbook)
Two sampling parameters are to be controlled
(1) Sample rate, f t 1
s 
(2) Number of points, N 

e highest frequency resolved in the signal is

determined by the Nyquist frequency:
f
f Nyq  s
2
Signals with frequencies lower than Nyquist frequency
are accurately sampled
Sampled signal taking on a di erent frequency is
called “aliasing”.
Sampling with 3/2 fsignal(=fs)
Sampling with fsignal=(1/2)fs
Sampling with fsignal > (1/2)fs
e lowest resolved frequency in the signal is
f lowest 1
 
1 f
 s  f
T N t N
 ree signal generators are used to produce sine
waves and connected to a loudspeaker.

Generator A is set to frequency 500Hz,

generator B to 1000Hz and
generator C to 1500Hz.
A microphone is used to measure the sound level. Each
generator is adjusted to 50mV in amplitude.
en all three signal generators are run simultaneously,
and the microphone signal is sampled by computer.
Find the Nyquist frequency if , the sampling
frequency=9000Hz, and the lowest frequency for DFT,
nd the number of points in the DFT.
Solution:
e minimum
frequencies: f Sampling Frequency to resolve all
s_min>2*1500Hz=3000Hz. In this case the
fNyq=1500Hz, ∆f=500Hz, and N=fs/∆f=6pts.
Since
f4500Hz the sampling frequency is set as 9000Hz,
Nyq=fs/2=4500Hz (signals with frequency content up to
are resolved)
e lowest frequency =500Hz hence
flowest=500Hz=∆f=fs/N
Hence N=fs/flowest=9000/500=18 samples
e sampling
T=18*0.11ms=2ms. time T=N∆t where ∆t=1/f s=.11ms hence
https://www.youtube.com/watch?v=-nxp52bNZWw
E ect of Moving Average for signals
Moving average technique smoothens the signal.
It eliminates the high frequencies involved in the
signal
e main trends in the signal can be more clearly
seen after applying moving average method and
by eliminating the noise
In other words it works similar to low pass ltering
(demonstrating the dominant low frequency in the
signal)
close all force
T=0.9;f1=20;f2=200;f3=340;noise=0.0;
deltaT=0.001;t = 0:deltaT:T;
x = 1*sin(2*pi*f1*t) + 0.0*sin(2*pi*f2*t)+ 0.0*sin(2*pi*f3*t);
%x=1*(square(2*pi*f1*t));
y = x + noise*randn(size(t));
subplot(3,1,1)
plot(y(1:0.2*length(t)));
title('Noisy time domain signal')
Y = t(y); Pyy = abs(Y);
Pyy1 = mag2db(abs(Y))
f = 1/(deltaT*length(t))*(0:f3+20);
subplot(3,1,2)
plot(f,Pyy(1:f3+21))
title('Power spectral density')
xlabel('Frequency (Hz)')
subplot(3,1,3)
plot(f,Pyy1(1:f3+21))
title('Power spectral density dB')
xlabel('Frequency (Hz)')