EECE 301
Signals & Systems
Prof. Mark Fowler
Note Set #7
C-T Signals: Three Forms of Fourier Series
1/14
Fourier Series Motivation
Fourier Series allows us to write virtually any real-world PERIODIC
signal as a sum of sinusoids with appropriate amplitudes and phases.
So we can think of building a periodic signal from sinusoidal building blocks.
Later we will extend that idea to also build many non-periodic signals from
sinusoidal building blocks!
Thus, it is very common for engineers to think about virtually any signal as being
made up of sinusoidal components.
Q: Why all this attention to sinusoids?
A: Recall from Circuits sinusoidal analysis of RLC circuits:
Fundamental Result: Sinusoid In Sinusoid Out
(Same Frequency, Different Amplitude & Phase)
A cos 0t , t
R
L
B cos 0t , t
2/14
This sinusoid in, sinusoid out result holds for Constant-Coefficient, Linear
Differential Equations as well as any LTI system. Well only motivate this
result for this Diff. Eq.:
y (t ) a1 y (t ) a0 y (t ) x (t )
If the input x(t) is a sinusoid A cos 0t , t
then the solution y(t) must be such that it and its derivatives can
be combined to give the input sinusoid.
So suppose the solution is y (t ) B cos 0t , t
o2 B cos(ot ) a1o B sin(ot ) a0 B cos(ot ) A cos(ot )
By slogging through lots of algebra and trig identities we can show
this can be met with a proper choice of B and .
But it makes sense that to add up to a sinusoid wed need all the
terms on the left to be sinusoids of some sort!!!
So we have reason to believe this:
Fundamental Result: Sinusoid In Sinusoid Out
(Same Frequency, Different Amplitude & Phase)
3/14
Now if our input is the linear combination of sinusoids:
x (t ) A1 cos 1t 1 A2 cos 2t 2 A3 cos 3t 3 , t
By linearity (i.e., superposition) we know that we can simply handle each
term separately and we know that each input sinusoid term gives an
output sinusoid term:
y (t ) B1 cos 1t 1 B2 cos 2t 2 B3 cos 3t 3 , t
So breaking a signal into sinusoidal parts makes the job of solving a
Diff. Eq. EASIER!! (This was Fouriers big idea!!)
But. What kind of signals can we use this trick on?
Or in other words
What kinds of signals can we build by adding together sinusoids??!!!
4/14
So Lets Explore What We Can Build with Sinusoids!
Let 0 be some given fundamental frequency
Q: What can I build from building blocks that looks like:
Ak cos( ko k ) ?
Only frequencies that are integer multiples of o
Ex.: o = 30 rad/sec then consider 0, 30 60, 90,
We can explore this by choosing a few different cases of values for the Ak
and k
On the next slide we limit ourselves to looking at three cases where we limit
ourselves to having only three terms
For this example let 0 = 2 rad/sec and look at a sum for k = 1, 2, 3:
x (t ) A1 cos 2 t 1 A2 cos 2 2 t 2 A3 cos 3 2 t 3
5/14
1 = 0
2 = /4
3 = /2
1
x(t)
A1 = 1.0
A2 = 0.5
A3 = 0.5
0
-1
Equal
1 = 0
2 = /4
3 = /2
0.5
1.5
2
time (sec)
2.5
3.5
0.5
1.5
2
time (sec)
2.5
3.5
0.5
1.5
2
time (sec)
2.5
3.5
2
1
x(t)
A1 = 0.1
A2 = 1.0
A3 = 0.5
-2
0
0
-1
Equal
-2
0
2
1 = 0
2 = /7
3 = /14
1
x(t)
A1 = 0.1
A2 = 1.0
A3 = 0.5
0
-1
Note:
1. All are periodic with period of 1s
2. All are centered vertically @ 0
-2
0
In one period: Area Above = Area Below
6/14
Why do these all have period of 1 s???
x (t ) A1 cos 2 t 1 A2 cos 2 2 t 2 A3 cos 3 2 t 3
Repeats every 1 s
Repeats every 1/2 s
so it also repeats
every 1 s
Repeats every 1/3 s
so it also repeats
every 1 s
This motivates the following general statement:
A sum of sinusoids with frequencies that are integer multiples of some
lowest fundamental frequency o will give a periodic signal with
period T = 2/o seconds.
So we can now think about adding together any number of
harmonically-related sinusoids even infinitely many!
x (t ) Ak cos kot k , t
k 1
i.e., all frequencies are an integer multiple of fund. freq. o
7/14
Why are these all centered vertically @ 0???
x (t ) A1 cos 2 t 1 A2 cos 2 2 t 2 A3 cos 3 2 t 3
Centered @ 0
Centered @ 0
Centered @ 0
This motivates the following general statement:
Unless we have a constant term added, a sum of sinusoids (with
frequencies at o, 2o, 3o, ) will be centered vertically at 0
So we can now add a constant term
x (t ) A0 Ak cos kot k , t
k 1
Note: for k = 0 we have A0cos(0ot) = A0 so we can think of the
constant term as a cosine with frequency = 0 and phase = 0
8/14
Fourier Series A Way to Build a Periodic Signal
x (t ) A0 Ak cos kot k , t
k 1
This signal has Period T = 2/o
Big Idea: We can think of (virtually) any real-world periodic signal as being
made up of (possibly infinitely) many sinusoids whose frequencies are all an
integer multiple of a fundamental frequency o.
(We wont prove that here but it can be proven and the proof is in the book)
Once we set o all we have to do is specify all the amplitudes (Ak) and phases
(k) and we get some periodic signal with period T = 2/o.
But if we are GIVEN a periodic signal how do we determine the correct:
Fundamental Frequency o (rad/sec)
Easy: o = 2/T
Amplitudes (Ak)
Need to Learn How!!
Phases (k)
9/14
Three Forms of Fourier Series
x (t ) A0 Ak cos kot k
Amplitude & Phase
Form
k 1
The equation above is just one of three (totally equivalent!) different forms of
the Fourier Series.
Each one contains the same information but presents it differently.
Which form you use in a particular setting depends.
Partly on your preference
Partly on what you are trying to do
Both of these come
with experience
We can easily find the other two by applying trig identities to the terms in the
above form.
10/14
Convert to Complex Exponential Form
x (t ) A0 A1 cos 1ot 1 A2 cos 2ot 2
Eulers Formula
cos( ) 12 e j e j
Amplitude
&
Phase
Form
x (t ) A0 A21 e j1 e j1ot A21 e j1 e j ( 1)ot A22 e j2 e j 2ot A22 e j2 e j ( 2)ot
c0
c
c
c1
c2
1
2
x (t )
jk0t
c
e
k
Complex Exponential
Form
11/14
Convert to Sine-Cosine Form
x (t ) A0 A1 cos 1ot 1 A2 cos 2ot 2
Trig Identity
cos( )
cos( ) cos( B ) sin( )sin( )
Amplitude
&
Phase
Form
x (t ) A0 A1 cos 1 cos 1ot A1 sin 1 sin 1ot
a0
a1
b1
A2 cos 2 cos 2ot A2 sin 2 sin 2ot
a
b
2
2
x (t ) a0 ak cos( k0t ) bk sin( k0t )
k 1
Sine-Cosine
Form
12/14
Three (Equivalent) Forms of FS and Their Relationships
Best for thinking about real-world ideas
Trig Form: Amplitude & Phase
x(t ) A0 Ak cos(k0t k )
k 1
c0 A0
A0 a0
a0 c0
Ak ak2 bk2
bk
ak
k tan 1
ak Ak cos( k )
bk Ak sin( k )
A0 c0
Ak 2 ck
k 1, 2, 3,
k ck
Best for doing math
(ck are like phasors!!)
ck 12 Ak e jk
k 1, 2, 3,
ck 12 Ak e k
Exponential Form
c0 a0
ck
1
2
x(t )
ak jbk
Trig Form: Sine-Cosine
x (t ) a0 ak cos(k0t ) bk sin( k0t )
jk0t
c
e
k
k 1, 2, 3,
ck 12 ak jbk
Best for some
special scenarios
a0 c0
ak 2 Reck , k 1, 2, 3,
bk 2 Imck , k 1, 2, 3,
k 1
13/14
x (t ) cos(t ) 0.5cos(4t / 3) 0.25cos(8t / 2)
which is already in Amp-Phase Form of the Fourier Series with 0 = 1 :
Example: Consider
A1 = 1
A4 = 0.5
A8 = 0.25
1 = 0
4 = /3
8 = /2
(all other Ak are 0)
Using the conversion results on the previous slide we can re-write this in
Complex Exponential Form of the FS as:
c1 = 0.5
c4 = 0.25e j/3
c8 = 0.125e j/2
c-1 = 0.5
c-4 = 0.25e-j/3
c-8 = 0.125e -j/2
(all other ck are 0)
x (t ) 0.5e jt 0.5e jt 0.25e j /3e j 4 t 0.25e j /3e j 4 t 0.5e j /2 e j 8t 0.5e j /2 e j 8t
Using the conversion results on the previous slide we can re-write this in
Sine-Cosine Form of the FS as:
a1 = 1
a4 = 0.25
a8 = 0
b1 = 0
b4 = 0.43
b8 = 0.25
(all other ak, bk are 0)
x (t ) cos(t ) 0.25cos(4t ) 0.43sin(4t ) 0.25sin(8t )
14/14
