Introduction to Mathematical Modeling of

Signals and Systems

Mathematical Representation of Signals
Signals represent or encode information
In communications applications the information is almost
always encoded
In the probing of medical and other physical systems,
where signals occur naturally, the information is not purposefully encoded
In human speech we create a waveform as a function of
time when we force air across our vocal cords and through
our vocal tract

!"#$%&'()'*+"),%'*.+&/+0"/)+
-'1*0"(&+--1&+
2&'#"/)+".'%,3
/&,%/"$*/'",*
+3+%/&$%,3"-$4*,3
/),/".,&$+-"'.+&
/$#+5"!

11

Introduction to Mathematical Modeling of Signals and Systems

Signals, such as the above speech signal, are continuous

functions of time, and denoted as a continuous-time signal
The independent variable in this case is time, t, but could be
another variable of interest, e.g., position, depth, temperature,
pressure
The mathematical notation for the speech signal recorded by
the microphone might be %  " 
In order to process this signal by computer means, we may
sample this signal at regular interval & % , resulting in
% > ' @ = %  '& % 

(1.2)

The signal % > ' @ is known as a discrete-time signal, and & % is

the sampling period
Note that the independent variable of the sampled signal is
the integer sequence ' ^ } ! & " & ! } `
Discrete-time signals can only be evaluated at integer values

1-2

 The speech waveform is an example of a one-dimensional

signal, but we may have more that one dimension
An image, say a photograph, is an example of a two-dimensional signal, being a function of two spatial variables, e.g.
(  ! ) 
If the image is put into motion, as in a movie or video, we
now have a three-dimensional image, where the third independent variable is time,  ! ) " 
Note: movies and videos are shot in frames, so actually
time is discretized, e.g., " o '& % (often & e & % = '" fps)
To manipulate an image on a computer we need to sample the
image, and create a two-dimensional discrete-time signal
( > * ' @ = (  *' ! '' ) 

(1.3)

where m and n takes on integer values, and ' ! and ' ) represent the horizontal and vertical sampling periods respectively
Mathematical Representation of Systems
In mathematical modeling terms a system is a function that
transforms or maps the input signal/sequence, to a new output signal/sequence
)  "  = &+ ^ !  "  `
) > ' @ = &, ^ ! > ' @ `

(1.4)

where the subscripts c and d denote continuous and discrete

system operators
13

Introduction to Mathematical Modeling of Signals and Systems

Because we are at present viewing the system as a pure mathematical model, the notion of a system seems abstract and
distant
Consider the microphone as a system which converts sound
pressure from the vocal tract into an electrical signal
Once the speech waveform is in an electrical waveform format, we might want to form the square of the signal as a first
step in finding the energy of the signal, i.e.,
)" = >!"@

(1.5)

!"-61,&+&"-7-/+#

)" = >!"@

The squarer system also exists for discrete-time signals, and

in fact is easier to implement, since all we need to do is multiply each signal sample by itself
14

)>'@ = !>'@ = !>'@ !>'@

(1.6)

If we send ) > ' @ through a second system known as a digital

filter, we can form an estimate of the signal energy
This is a future topic for this course
Thinking About Systems
Engineers like to use block diagrams to visualize systems
Low level systems are often interconnected to form larger
systems or subsystems
Consider the squaring system
!"

& ^ ((( `

)"

T"$-","4+*+&$%"-7-/+#

!"

 ((( 

)"

The ideal sampling operation, described earlier as a means to

convert a continuous-time signal to a discrete-times signal is
represented in block diagram form as an ideal C-to-D converter
!"

80+,3
9:/':;
9'*.+&/+&

&%
15

! > ' @ = !  '& % 

!"-7-/+#"(,&,#+/+&"/),/
-(+%$2$+-"/)+"-,#(3+"-(,%$*4

Introduction to Mathematical Modeling of Signals and Systems

A more complex system, depicted as a collection of subsystem blocks, is a system that records and then plays back an
audio source using a compact disk (CD) storage medium

The optical disk reader shown above is actually a high-level

block, as it is composed of many lower-level subsystems,
e.g.,
Laser, on a sliding carriage, to illuminate the CD
An optical detector on the same sliding carriage
A servo control system positions the carriage to follow the
track over the disk
A servo speed control to maintain a constant linear velocity as 1/0 data is read from different portions of the disk
more ...
The Next Step
Basic signals, composed of linear combinations of trigonometric functions of time will be studied next
We also consider complex number representations as a means
to simplify the combining of more than one sinusoidal signal

16

Sinusoids
A general class of signals used for modeling the interaction of signals in systems, are based on the trigonometric functions sine and cosine
The general mathematical form of a single sinusoidal signal
is
x  t  = A cos  Z 0 t + I 

(2.1)

where A denotes the amplitude, Z 0 is the frequency in radians/s (radian frequency), and I is the phase in radians
The arguments of cos   and sin   are in radians
We will spend considerable time working with sinusoidal signals, and hopefully the various modeling applications presented in this course will make their usefulness clear
Example: x  t  = 10 cos > 2S  440 t 0.4S @

The pattern repeats every 1 e 440 = 0.00227 = 2.27ms

This time interval is known as the period of x  t 
21

Review of Sine and Cosine Functions

The text discusses how a tuning fork, used in tuning musical

instruments, produces a sound wave that closely resembles a
single sinusoid signal
In particular the pitch A above middle C has an oscillation
frequency of 440 hertz

Review of Sine and Cosine Functions

Trigonometric functions were first encountered in your K12
math courses
The typical scenario to explain sine and cosine functions is
depicted below

The right-triangle formed in the first quadrant has sides of

length x and y, and hypotenuse of length r
The angle T has cosine defined as x/r and sine defined as y/
r
The above graphic also shows how a point of distance r
and angle T in the first quadrant of the xy plane is related
22

Review of Sine and Cosine Functions

to the x and y coordinates of the point via sin( ) and cos( ),

e.g.,
 x y  =  r cos T r sin T 

(2.2)

Moving beyond the definitions and geometry interpretations,

we now consider the signal/waveform properties

The function plots are identical in shape, with the sine plot
shifted to the right relative to the cosine plot by S e 2
This is expected since a well known trig identity states that
sin T = cos  T S e 2 

(2.3)

We also observe that both waveforms repeat every 2S radians; read period = 2S
Additionally the amplitude of each ranges from -1 and 1
A few key function properties and trigonometric identities
23

Review of Sine and Cosine Functions

are given in the following tables

Table 2.1: Some sine and cosine properties

Property

Equation

Equivalence

sin T = cos  T S e 2  or
cos T = sin  T + S e 2 

Periodicity

cos  T 2Sk  = cos T , when k is an

integer; holds for sine also

Evenness of cosine cos  T  = cos T

Oddness of sine

sin  T  = sin T

Table 2.2: Some trigonometric identities

Number

Equation

sin T + cos T = 1

cos 2T = cos T sin T

sin 2T = 2 sin T cos T

sin  D r E  = sin D cos E r cos D sin E

cos  D r E  = cos D cos E 

 sin D sin E

1
2
cos T = ---  1 + cos 2T 
2

2
1
sin T = ---  1 cos 2T 
2

For more properties consult a math handbook

24

Review of Sine and Cosine Functions

The relationship between sine and cosine show up in calculus

too, in particular
d cos T
d sin T
-------------- = cos T and --------------- = sin T
dT
dT

(2.4)

This says that the slope at any point on the sine curve is the
cosine, and the slope at any point on the cosine curve is the
negative of the sine
Example: Prove Identity #6 Using Identities #1 and #2
If we add the left side of 1 to the right side of 2 we get
2

2cos T = 1 + cos 2T
1
2
or cos T = ---  1 + cos 2T 
2

(2.5)

Example: Find an expression for cos 8T in terms of cos 9T ,

cos 7T , and cos T using #5
Let D = 8T and E = T , then write out #5 under both sign
choices
cos  8T + T  = cos 8T cos T sin 8T sin T
cos  8T T  = cos 8T cos T + sin 8T sin T
+
cos 9T + cos 7T = 2 cos 8T cos T

(2.6)

cos 9T + cos 7T
cos 8T = ------------------------------------2 cos T

(2.7)

or

25

Review of Complex Numbers

Review of Complex Numbers

See Appendix A of the text for more information
A complex number is an ordered pair of real numbers1
denoted z =  x y 
The first number, x, is called the real part, while the second
number, y, is called the imaginary part
For algebraic manipulation purposes we write  x y 
= x + iy = x + jy where i = j = 1 ; electrical engineers typically use j since i is often used to denote current
Note:

1 u 1 = 1 j u j = 1

The rectangular form of a complex number is as defined

above,
z =  x y  = x + jy
The corresponding polar form is
z = re

jT

= r T = z e j arg z

1. Tom M. Apostle, Mathematical Analysis, second edition, Addison Wesley,

p. 15, 1974.
26

Review of Complex Numbers

We can plot a complex number as a vector  x y 

Example:: z = 2 + j5 , z = 4 j3 , z = 5 + j0 , z = 3 j3

27

Review of Complex Numbers

Example:: z = 2 45q , z = 3 150q , & z = 3 80 q

For complex numbers z 1 = x 1 + jy 1 and z 2 = x 2 + jy 2 we

define/calculate
z 1 + z 2 =  x 1 + x 2  + j  y 1 + y 2  (sum)
z 1 z 2 =  x 1 x 2  + j  y 1 y 2  (difference)
z 1 z 2 =  x 1 x 2 y 1 y 2  + j  x 1 y 2 + y 1 x 2  (product)
 x1 x2 + y1 y2  j  x1 y2 y1 x2 
z1
---- = ------------------------------------------------------------------------- (quotient)
z2
x2 + y 2
2

ECE 2610 Signals and Systems

28

Review of Complex Numbers

z1 =

x 1 + y 1 (magnitude)
1

z 1 = tan  y 1 e x 1  (angle)
z *1 = x 1 jy 1 (complex conjugate)
MATLAB is also consistent with all of the above, starting
with the fact that i and j are predefined to be 1

rectangular

polar

To convert from polar to rectangular we can use simple trigonometry to show that
x = r cos T
y = r sin T

(2.24)

Similarly we can show that rectangular to polar conversion is

r =

x +y
1

T = tan  y e x  note add r S outside Q1 & Q4

29

(2.25)

Review of Complex Numbers

Example: Rect to Polar and Polar to Rect

Consider z 1 = 2 + j5
In MATLAB we simply enter the numbers directly and then
need to use the functions abs() and angle() to convert
>> z1 = 2 + j*5
z1 = 2.0000e+00 + 5.0000e+00i
>> [abs(z1) angle(z1)]
ans = 5.3852e+00

1.1903e+00 % mag & phase in rad

Using say a TI-89 calculator is similar

Consider z 2 = 2 45q
In MATLAB we simply enter the numbers directly as a
complex exponential
>> z2 = 2*exp(j*45*pi/180)
z2 = 1.4142e+00 + 1.4142e+00i

210

Review of Complex Numbers

Using the TI-89 we can directly enter the polar form using
the angle notation or using a complex exponential

Example: Complex Arithmetic

Consider z 1 = 1 + j7 and z 2 = 4 j9
Find z 1 + z 2
>> z1 = 1+j*7;
>> z2 = -4-j*9;
>> z1+z2
ans = -3.0000e+00 - 2.0000e+00i

Using the TI-89 we obtain

211

Review of Complex Numbers

Find z 1 z 2
>> z1*z2
ans = 5.9000e+01 - 3.7000e+01i

Using the TI-89 we obtain

Find z 1 e z 2
>> z1/z2
ans =-6.9072e-01 - 1.9588e-01i

TI-89
Results

Eulers Formula: A special mathematical result, of special

importance to electrical engineers, is the fact that
e

212

jT

= cos T + j sin T

(2.26)

Sinusoidal Signals

Turning (2.26) around yields (inverse Euler formulas)

jT

jT

jT

jT

e e +e
sin T = e--------------------and cos T = ---------------------2j
2

(2.27)

It also follows that

z = x + jy = r cos T + jr sin T

(2.28)

Sinusoidal Signals
A general sinusoidal function of time is written as
x  t  = A cos  Z 0 t + I  = A cos  2Sf 0 t + I 

(2.29)

where in the second form Z 0 = 2Sf 0

Since cos T d 1 it follows that x  t  swings between r A , so
the amplitude of x  t  is A
The phase shift in radians is I , so if we are given a sine signal (instead of the cosine version), we see via the equivalence
property that
x  t  = A sin  Z 0 t + Ic  = A cos  Z 0 t + Ic S e 2 

(2.30)

which implies that I = Ic S e 2

Engineers often prefer the second form of (2.8) where f 0 is
the oscillation frequency in cycles/s.
Z 0 rad/s
1
------ ----------- = f 0  sec 
2S rad

213

Sinusoidal Signals

Example: x  t  = 20 cos > 2S  40 t 0.4S @

Clearly, A = 20 , f 0 = 40 cycles/s, and I = 0.4S rad
Maxima
Interval
(period)

1- = 0.025s
----40
= 25ms

Since this signal is periodic, the time interval between maxima, minima, and zero crossings, for example, are identical
Relation of Frequency to Period
A signal is periodic if we can write
x  t + T0  = x  t 

(2.31)

where the smallest T 0 satisfying (2.10) is the period

For a single sinusoid we can relate T 0 to f 0 by considering
x  t + T0  = x  t 
A cos  Z 0  t + T 0  + I  = A cos  Z 0 t + I 

(2.32)

cos  Z 0 t + I + Z 0 T 0  = cos  Z 0 t + I 
From the periodicity property of cosine, equality is maintained if cos  T r 2Sk  = cos  T  , so we need to have

214

Sinusoidal Signals

2S
Z 0 T 0 = 2S T 0 = -----Z0
1or  2Sf 0 T 0 T 0 = --f0

(2.33)

So we see that T 0 and f 0 are reciprocals, with the units of T 0

being time and the units of f 0 inverse time or cycles per second, as stated earlier
In honor of Heinrich Hertz, who first demonstrated the
existence of radio waves, cycles per second is replaced
with Hertz (Hz)

215

Sinusoidal Signals

Example: 5 cos  2Sf 0 t  with f 0 = 200 , 100, and 0 Hz

Period doubles as
frequency halves

A constant signal
as the oscillation
frequency is zero

The inverse relationship between time and frequency will be

explored through out this course

216

Sinusoidal Signals

Phase Shift and Time Shift

We know that the phase shift parameter in the sinusoid moves
the waveform left or right on the time axis
To formally understand why this is, we will first form an
understanding of time-shifting in general
Consider a triangularly shaped signal having piece wise continuous definition

st =

s t

2t,
0dtd1e2
1-- 4 2t , 1 e 2 d t d 2
3
0,
otherwise

1
---  4 2t 
3

2t

-1

(2.34)

1--2

Now we wish to consider the signal x 1  t  = s  t 2 

As a starting point we note that s  t  is active over just the
interval 0 d t d 2 , so with t o t 2 we have
0 d t 2 d 2 2 d t d 4

(2.35)

which means that x 1  t  is active over 2 d t d 4

The piece wise definition of x 1  t  can be obtained by direct
substitution of t 2 everywhere t appears in (2.34)
217

Sinusoidal Signals

x1  t  =

2  t 2 ,
0 d t 2 d 1 e 2
1-- 4 2  t 2  , 1 e 2 d  t 2  d 2
3
0,
otherwise
(2.36)
2t 4,
2dtd5e2
1-- 8 2t , 5 e 2 d t d 4
3
0,
otherwise

x1  t  = s  t 2 
1

1
---  8 2t 
3

2t 2

5--2

In summary we see that the original signal s  t  is moved to

the right by 2 s
Example: Plot s  t + 1 
With t o t + 1 we expect that the signal will shift to the left
by one second
st + 1

-1
218

1
--2

Sinusoidal Signals

The new equations are obtained as before

0 d  t + 1  d 2 1 d t d 1

(2.37)

so

st + 1 =

2  t + 1 ,
0 d t + 1 d 1 e 2
1-- 4 2  t + 1  , 1 e 2 d  t + 1  d 2
3
0,
otherwise
(2.38)
2t + 2,
1 d t d 1 e 2
1
---  2 2t , 1 e 2 d t d 1
3
0,
otherwise

Modeling time shifted signals shows up frequently

In general terms we say that
x1  t  = s  t t1 

(2.39)

is delayed in time relative to s  t  if t 1 ! 0 , and advanced in

time relative to s  t  if t 1  0
A cosine signal has positive peak located at t = 0
If this signal is delayed by t 1 the peak shifts to the right and
the corresponding phase shift is negative
Consider x 0  t  = A cos  Z 0 t 
219

Sinusoidal Signals

x 0  t t 1  = A cos > Z 0  t t 1  @
= A cos > Z 0 t Z 0 t 1 @

(2.40)

which implies that in terms of phase shift we have

I = Z0 t1
For a given phase shift we can turn the above analysis around
and solve for the time delay via
I
I
t 1 = ------ = ----------Z0
2Sf 0

(2.41)

Since T 0 = 1 e f 0 , we can also write the phase shift in terms

of the period
t1

I = 2Sf 0 t 1 = 2S ---- T 0

(2.42)

An important point to note here is that both cosine and sine

are mod 2S functions, meaning that phase is only unique
on a 2S interval, say ( S S] or (0 2S]
Example: Suppose t 1 = 10 ms and T 0 = 3 ms
Direct substitution into (2.21) results in
20
10
I = 2S ------ = ------ S = 6.6667S
3
3

(2.43)

We need to reduce this value modulo 2S to the interval

( S S] by adding (or subtracting as needed) multiples of 2S
The result is the reduced phase value

220

Sampling and Plotting Sinusoids

20
20 + 18
2
I = ------ S + 6S = ------------------ S = --- S = 0.6667S
3
3
3

(2.44)

Does this result make sense?

A time delay of 10 ms with a period of 3 ms means that we
have delayed the sinusoid three full periods plus 1 ms
A 1 ms delay is 1/3 of a period, with half of a period corresponding to S rad, so a delay of 1/3 period is a phase shift of
 2 e 3  S = 0.6667S ; agrees with the above analysis


Modulo the period

delay of 1 ms

Actual Delay of 10 ms





t (ms)




Blue = no delay
Red = 10 ms Delay

The value of phase shift that lies on the interval S  I d S is

known as the principle value

Sampling and Plotting Sinusoids

When plotting sinusoidal signals using computer tools, we
are also faced with the fact that only a discrete-time version
221

Sampling and Plotting Sinusoids

of
x  t  = A cos  2Sf 0 t + I 
may be generated and plotted
This fact holds true whether we are using MATLAB, C, Mathematica, Excel, or any other computational tool
When t o nT s we need to realize that sample spacing needs
to be small enough relative to the frequency f 0 such that
when plotted by connecting the dots (linear interpolation),
the waveform picture is not too distorted
In Chapter 4 we will discuss sampling theory, which will
tell us the maximum sample spacing (minimum sampling
rate which is 1 e T s ), such that the sequence
x > n @ = x  nT s  can be used to perfectly reconstruct x  t 
from x > n @
For now we are more concerned with having a good plot
appearance relative to the expected sinusoidal shape
A reasonable plot can be created with about 10 samples per
period, that is with T s | 1 e  10f 0  = T 0 e 10
We will now consider several MATLAB example plots
>>
>>
>>
>>
>>
>>
>>

t = 0:1/(5*3):1; x = 15*cos(2*pi*3*t-.5*pi);
subplot(311)
plot(t,x,'.-'); grid
xlabel('Time in seconds')
ylabel('Amplitude')
t = 0:1/(10*3):1; x = 15*cos(2*pi*3*t-.5*pi);
subplot(312)

222

Sampling and Plotting Sinusoids

>>
>>
>>
>>
>>
>>
>>
>>
>>

plot(t,x,'.-'); grid
xlabel('Time in seconds')
ylabel('Amplitude')
t = 0:1/(50*3):1; x = 15*cos(2*pi*3*t-.5*pi);
subplot(313)
plot(t,x,'.-'); grid
xlabel('Time in seconds')
ylabel('Amplitude')
print -depsc -tiff sampled_cosine.eps

f0 = 3 Hz, A = 15, I = -S/2

Amplitude

20

T
T s = -----05

10
0
5 Samples
10 per period

20

0.1

0.2

0.3

Amplitude

20

0.7

0.8

0.9

0.7

0.8

0.9

0.7

0.8

0.9

1T
T s = --------010

10
0
10
20

10 Samples
per period

0.1

0.2

0.3

20
Amplitude

0.4
0.5
0.6
Time in seconds

0.4
0.5
0.6
Time in seconds

T
T s = -----050

10
0
10 50 Samples
per period
20
0
0.1
0.2

223

0.3

0.4
0.5
0.6
Time in seconds

!"#$%&'()'$"*&*+,-%.(-*/(01-."2.

Complex Exponentials and Phasors

Modeling signals as pure sinusoids is not that common. We typically have more that one sinusoid present. Manipulating multiple
sinusoids is actually easier when we form a complex exponential
representation.
Complex Exponential Signals
Motivated by Eulers formula above, and the earlier definition of a cosine signal, we define the complex exponential
signal as
!  "  = #$

%  Z! " + I 

(2.44)

where !  "  = # and !  "  = "#$ ^ !  "  ` = Z ! " + I

Note that using Eulers formula
%Z " + I

!
!  "  = #$
= # %&'  Z ! " + I  + %# '()  Z ! " + I 

(2.45)

We see that the complex sinusoid has amplitude A, phase

shift I , and frequency Z ! rad/s
Note in particular that
*+ ^ !  "  ` = # %&'  Z ! " + I 
,- ^ !  "  ` = # '()  Z ! " + I 

3:3;

(2.46)

!"#$%&'()'$"*&*+,-%.(-*/(01-."2.

The result of (2.46) is what ultimately motivates us to consider the complex exponential signal
We can always write
&  "  = *+ ^ #$

%  Z! " + I 

` = # %&'  Z ! " + I 

(2.47)

The Rotating Phasor Interpretation

Complex numbers in polar form can be easily multiplied as

!/

'0 $

%T 0

%T /

!. = '/ $

= '/ '0 $

!0

%  T/ + T0 

(2.48)

For the case of

!  "  = #$
3:3<

%  Z! " + I 

(2.49)

!"#$%&'()'$"*&*+,-%.(-*/(01-."2.

we can write
%I

!  "  = #$ $
where ( = #$

%Z ! "

= ($

%Z ! "

(2.50)

%I

The complex amplitude ( is called the phasor, as it is the

gain and phase value applied to the time varying component
%Z "
$ ! to form !  " 
This is common terminology is electrical engineering circuit theory
%Z "

The time varying term $ ! has unit magnitude and rotates

counter clockwise in the complex plane at a rate of Z ! rad/s
( ) ! rotations/s)
The time duration for one rotation is the period * ! = / e ) !
The combination (product) of the fixed phasor ( and $
results in a rotating phasor

%Z ! "

For positive frequency Z ! the rotation is counter clockwise, and for negative frequency the rotation is clockwise
#$
%&'()(*"
+,"-."/01

2"34)(*"
+,"-."/01

T"

!"

!"#$#%&'()*$+",+

3:34

T"

!"

T  "  = Z! " + I

!"#$%&'()'$"*&*+,-%.(-*/(01-."2.

Example: !  "  = +23 > 0S" S e 1 @

Plot a series of snap shots of the rotating phasor when * + = 1/
8 (note * ! = / s)
!""""""""""""""""""""""""""""""""""""""""""""""""""""""
!#$#%&'()*#+(,-#+.'#/-0-'1*(0/#1#%-23-0&-#.+#'.*1*(0/#
!#)41%.'#%01)#%4.*%
!
!
!""""""""""""""""""""""""""""""""""""""""""""""""""""""
#
!#/-*#*4-#+.&3%#.+#+(/3'-#<(0=.<#>?#.'#&'-1*-#
!#(+#0.*#&'-1*-=@
+(/3'-A?B#
&,+A?BC#!#&,-1'#+(/3'-#<(0=.<#>?
#
$#D#?@:C#+:#D#?C#)4(#D#E)("FC
G#D#HC#!#&'-1*-#H#I-&*.'#),.*%
J%#D#?"HC
+.'#0#D#:KGE?
####%3L),.*AF8980M?B
####*#D#:K?"9::K?C
####),.*A&.%A9N)(N*B8%(0A9N)(N*B8O6KOB
####4.,=#.0
####P#D#$N-Q)ARNA9N)(N+:N0NJ%M)4(BBC
####),.*AS:8'-1,APBT8S:8(U1/APBT8OV(0-7(=*4O8?B
####!#(0%(=-#%)'(0*+#&'-1*-%#1#+.'U1**-=#%*'(0/
####*(*,-A%)'(0*+AOJ(U-#D#!?@F+#%O80NJ%BBC
####1Q(%A?@?NSE$#$#E$#$TBC#1Q(%#-231,C
####),.*A'-1,APB8(U1/APB8O'@O8O51'6-'W(P-O8?HB
####4.,=#.++
-0=

3:3=

!"#$%&'()'$"*&*+,-%.(-*/(01-."2.

Time = 0.0000 s

Time = 0.1250 s

0
1

$
2

% S e 1

0
1

$
2

Time = 0.2500 s
1

0
1

%S e 1

Time = 0.7500 s
1

3:3>

Time = 0.8750 s

Time = 0.6250 s

Time = 0.3750 s

Time = 0.5000 s

%!

01-."2(?//,+,"*

The inverse Euler formulas can be used to see that a cosine

signal is composed of positive and negative frequency exponentials
%  Z! " + I 

%  Z! " + I 

$
+$
# %&'  Z ! " + I  = # -----------------------------------------------------

0
/ %Z ! " / 4 %Z! "
= --- ($
+ --- ( $
0
0

(2.51)

/
/ 4
= --- !  "  + --- !  " 
0
0
= *+ ^ !  "  `

Phasor Addition
We often have to deal with multiple sinusoids. When the sinusoids are at the same frequency, we can derive a formula of the
form
-

#, %&'  Z! " + I,  = # %&'  Z! " + I 

(2.52)

,=/

At present we have only the trig identities to aid us, and this
approach becomes very messy for large N.
Phasor Addition Rule
We know that when complex numbers are added we must add
real and imaginary parts separately
Consider the sum
3:3@

01-."2(?//,+,"*

#, $

jI ,

( , { ( = #$

%I

(2.53)

,=/

,=/

The above is valid since the real and imaginary parts add
independently, that is

*+

,=/

(, =

*+ ^ (, `

(2.54)

,=/

and the same holds for the imaginary part

Secondly, a real sinusoid can always be written in terms of a
complex sinusoid via
# %&'  Z ! " + I  = *+ ^ #$

%  Z! " + I 

(2.55)

Proof:
-

# , %&'  Z ! " + I ,  =

,=/

+&55&6'78,&$79:;<4>

*+ ^ # , $

%  Z! " + I, 

,=/

%I , %Z ! "
= *+
#, $ $

, = /

%I

= *+ ^  #$ $
= *+ ^ #$

%Z ! "

%  Z! " + I 

= # %&'  Z ! " + I 
3:A6

01-."2(?//,+,"*

Example: Phasor Addition Rule in Action

Consider the sum
&  "  = &/  "  + &0  " 
= 156 %&'  .!S" + .6S e /7! 
8 + 950 %&'  .!S" + 7!S e /7! 

(2.56)

The frequency of the sinusoids is 15 Hz

Using phasor notation we can write that
& /  "  = *+ ^ 156$

%.6S e /7!

& 0  "  = *+ ^ 950$

%7!S e /7!

%.!S"

%.!S"

`
`

(2.57)

so in the phasor addition rule

( / = 156$

%.6S e /7!

 ( 0 = 950$

%7!S e /7!

(2.58)

We perform the complex addition and conversion back to

polar form using the TI-89

(
I8()8#":(")'

so
( = ( / + ( 0 = 15;.<16 + %;5<9/9/
= /!57679$
3:A5

%<05;<!0S e /7!

(2.59)

01-."2(?//,+,"*

Finally,
&  "  = /!57< %&'  .!S" + <05;<!0S e /7! 

(2.60)

We can check this by directly plotting the waveform in MATLAB

XX#*#D#:K?"AY:N?YBK:@9C
XX#Q?#D#F@YN&.%AZ:N)(N*MZYN)("?H:BC
XX#Q9#D#[@9N&.%AZ:N)(N*MH:N)("?H:BC
XX#Q#D#Q?MQ9C#!),.*#3%(0/#4.,=#10=#,(0-#%*\,-%
?@;A::

x1(t)

10

x2(t)

&"

x(t)

Amplitude

&/  " 
0

B??;CD7$'
5

&0  " 

10

0.06

0.04

0.02

0
0.02
Time in seconds

0.04

0.06

The measured amplitude, 10.822, is close to the expected

value
3:A3

01-."2(?//,+,"*

The location of the peak can be converted to phase via

//5<9
I = 0 S ---------------- = /5/!8#": = <.5!0q
<<5<9

(2.61)

Summary of Phasor Addition

When we need to form the sum of sinusoids at the same frequency, we obtain the final amplitude A and phase I via
( = ( / + ( 0 + } + ( - = #$
where ( , = # , $

%I ,

%I

(2.62)

and
-

&" =

#, %&'  Z! " + I, 
,=/

(2.63)

= # %&'  Z ! " + I 
Example:

&  "  = *+ .$

S
% 0S) ! " + ---

+ 6$

S
% 0S) ! " ---

+  . + %0 $

Find ( = ( / + ( 0 + ( .
From the given &  "  we observe that
S
% --0

( / = .$  ( 0 = 6$

3:AA

S
% --1

 ( . = . + %0

%0S) ! "

019.,B.("C(+1&(DE*,*8(F"2G

To perform the complex addition we will work step-by-step

To add complex numbers we convert to rectangular form
S
S
( / = . %&' --- + %. '() --- = %.
0
0
S
S
( 0 = 6 %&' --- + %6 '() --- = .56.66 %.56.66
1
1
( . = . + %0
Now,
( =  %.  +  .56.66 %.56.66  +  . + %0 
=  <56.66 + %/51<16 
For use in the phasor sum formula we likely need the answer
in polar form
( =

0
0
<56.66 + /51<16 "=") /51<16
---------------<56.66

= <5<;9< !500!1 = <5<;9<$

%!500!1

Physics of the Tuning Fork

The tuning fork signal generation example discussed earlier was
important because it is an example of a physical system that
when struck, produces nearly a pure sinusoidal signal.
By pure we mean a signal composed of a single frequency
sinusoid, no other sinusoids at other frequencies, say harmonics (multiples of ) ! ) are present
3:A;

019.,B.("C(+1&(DE*,*8(F"2G

Equations from Laws of Physicss

A 2-D model of the tuning fork is shown below
E6&7)(/"'7&8
)F"7)./(/378&,G

When struck the vibration of the metal tine moves air molecules to produce a sound wave
Hookes law from physics (springs, etc.) says that the force to
restore the tine back to its original & = ! position is the same
as the original deformation (striking force), except for a sign
change,
. = ,&

(2.64)

where k is the material stiffness constant

The acceleration produced by the restoring force (Newtons
second law) is
3:A<

019.,B.("C(+1&(DE*,*8(F"2G

1 &
. = /0 = / -------01"

(2.65)

To balance the two forces (sum is zero), we must have

0

1 &
/ -------0- = ,&  " 
1"

(2.66)

General Solution to the Differential Equation

To solve this equation we can actually guess the solution by
inserting a test function of the form &  "  = %&' Z ! "
0

1
1 &"
--------------- = -----  Z ! '() Z ! " 
0
1"
1"
0

= Z ! %&' Z ! "
We now plug this result into (2.66) to obtain
0

1 &
/ -------0- = ,&  " 
1"

(2.67)

/  Z ! %&' Z ! "  = , %&' Z ! "

which tells us that we must have
0

/Z ! = ,

(2.68)

,
Z ! = r ---/

(2.69)

so it must be that

3:A4

019.,B.("C(+1&(DE*,*8(F"2G

This tells us that the oscillation frequency of the tuning fork

is related to the ratio of the stiffness constant to the mass
Greater stiffness means a higher oscillation frequency
Greater mass means a lower oscillation frequency
In terms of a real sinusoid the sound wave, to within a phase
shift constant is of the form
,- " + I
&  "  = # %&' -- /

(2.70)

The sound produced by the 440Hz tuning fork was captured

using MATLAB on a PC with a sound card and microphone
The results were converted to double precision and saved
in a .mat file along with a time axis vector
XX#,.1=#*30(0/+.'6
XX#),.*A*8QB

Amplitude

1
0.5
0
0.5
1

Amplitude

H&&$7IJKJ;@:L7'

4
5
6
Time in seconds

10

0.5
0
0.5
1

3:A=

4.002 4.004 4.006 4.008 4.01 4.012 4.014 4.016 4.018

Time in seconds

4.02

019.,B.("C(+1&(DE*,*8(F"2G

How pure is the signal produced by the tuning fork?

In Chapter 3 of the text we begin a study of spectrum representation
The zoom of the captured signal looks like a single sinusoid,
but spectral analysis can be more revealing
Consider the use of MATLABs power spectral density function )%=AQ8G++*8+%1U)B (Detail comes later Chapter)
XX#)%=AQ89]?98H:::B
40

JJ@7MN78./O4$"/)457)./(/378&,G7P()0F

Power Spectrum Magnitude (dB)

20

AA@7MN7'"0&/O7F4,$&/(0

Q)F",7F4,$&/(0'

20

40

60

R."7)&7')4,)B.P7),4/'("/)
80

3:A>

500

1000

1500
2000
2500
Frequency (Hz)

3000

3500

4000

019.,B.("C(+1&(DE*,*8(F"2G

A time-frequency plot can be obtained using the MATLABs

spectrogram function (Detail comes later Chapter)
XX#%)-&*'./'1UAQ89]?:8Y:8ST8H:::B
EF"7JJ@7MN78./O4$"/)457,"$4(/'7'),&/3

EF"7F(3F",7F4,$&/(0'784O"7(/7)($"

Listening to Tones
To play the tuning fork sound on the PC speakers using Matlab we type
XX#%.30=AQ8H:::B

where the second argument sets the sampling frequency for

playback

3:A@

D,#&(7,8*-%.H(I"2&(D1-*(F"2#E%-.

Time Signals: More Than Formulas

The signal modeling of this chapter has focused on single
sinusoids
In practice real signals are far more complex, even a multiple
sinusoids model is only an approximation
Modeling still has great value in system design

3:;6