Cardiff School of Engineering

Autumn Semester 2020/21

EN2090 Engineering Mathematics 2

Fourier Analysis
Prof. Adrian Porch
Dr Yue Zhou

[email protected]
 1
Section 2: Complex Fourier Series

Aims of this section

• To introduce the complex exponential Fourier series representation
of a periodic waveform

Objectives of this section

After completing this section you should be able to
• Calculate the Fourier coefficients of a complex Fourier series for
any given periodic waveform
• Calculate the magnitude and phase of the Fourier coefficients and
plot their double-sided spectra
• Derive the multiplication theorem of two complex Fourier series

The complex Fourier series is another way of representing a real

periodic waveform (like the real Fourier series), but is preferred in
applications like signal processing since the maths is easier.
 2
2.1 Complex exponential (i.e. phasor) notation
In Appendix C we show that
e j = cos  + j sin 

i.e. the phasor e j sweeps out a unit circle

on the Argand diagram as the phase angle  is
varied. Some important cases are:

 = 0 or 2 or − 2  =  / 2 or − 3 / 2
e j 0 = e j 2 = e− j 2 = 1 e j / 2 = + j

 =  or −   = 3 / 2 or −  / 2
e j = e − j = −1 e − j / 2 = − j

For angles outside of the range from 0 to 2, we apply e j ( 2 n ) = e j

(n = integer). For further help with this, see phasor.mcd.
 3
2.2 Complex form for Fourier series
In Appendix C we also derive the very important identities
(
cos  = 12 e j + e − j , ) (
sin  = 21j e j − e − j )
A periodic waveform f (t ) of period T = 2 /  , can be thus be
written

f (t ) = 12 a0 +  (a
n =1
n cos nt + bn sin nt )


= 12 a0 + 12 
 n n
( a − jb ) e jnt
+ ( a n + jbn ) e − jnt

n =1

However, from Euler’s formulae a− n = + an and b− n = −bn .

Defining cn = 12 (an − jbn ) we thus obtain c−n = 12 (an + jbn ) = cn *, so
 
f (t ) = c0 + 
n =1
cn e jnt + 
n =1
c−n e − jnt (with c0 = 12 a0 )
 4
Letting the sum run from −  → +  gives the result (Appendix D)
 k +T


1
f (t ) =  f (t )e − jnt dt
jnt
→ cn e , where cn = (for any k)
n = −
T
k

This important result is the complex Fourier series representation

of a periodic waveform.
• The complex Fourier series is used extensively in signal processing
applications
• Generally, the coefficients are complex, i.e. cn = cn e jn
• The plot of cn against n is called the amplitude spectrum
• The plot of  n against n is called the phase spectrum
• These two spectra of a complex Fourier series are double-sided, i.e.
they consist of pairs of components at frequencies  n (unlike
the single-sided spectra of the real Fourier series representation)
 5
• The negative frequencies have no physical meaning but each is
linked to its real frequency partner and doesn’t occur in isolation
• The complex Fourier series and the real Fourier series are two
different ways of describing the same physical periodic waveform
• Once the complex coefficients cn = 12 (an − jbn ) have been
established, it is easy to calculated the coefficients (and hence
spectra) of the corresponding real Fourier series
• The single-sided spectra (n  0) of the real Fourier series are then
Coefficient spectra an = 2 Re(cn ), bn = −2 Im(cn )
Amplitude spectrum An = 2 cn = an2 + bn2
Time-averaged power spectrum (
Pn = 2 cn = 12 an2 + bn2
2
)
Phase spectrum  n = arg(cn )

The spectra of any waveform can be measured experimentally using a

spectrum analyser, which performs Fourier analysis on it.
 6
Worked example 2.1
Calculate the complex Fourier series and plot the amplitude and phase
spectra of the periodic waveforms: a cos t , b sin t , k cos 2 t .
(
a cos t = 12 a e jt + e − jt ) → c−1 = 12 a, c0 = 0, c1 = 12 a
b sin t = 21j b(e jt − e − jt ) → c−1 = j 12 b, c0 = 0, c1 = − j 12 b
(
k cos 2 t = 14 k e 2 jt + 2 + e −2 jt ) → c− 2 = 14 k , c−1 = 0, c0 = 12 k , c1 = 0, c2 = 14 k

The time domain waveforms and their spectra are plotted below:
 7
Worked example 2.2
Find the complex Fourier series of the pulse
train shown. Show that the results are
consistent with those of the real Fourier series.
+ / 2 + / 2
V0 V0  V0
d.c. term is c0 =
T 
− / 2
dt =
T
, for other terms cn =
T 
− / 2
e − jnt dt

 cn = −
V0
jnT
( 2V
)  n  V0 
e − jn / 2 − e + jn / 2 = 0 sin
nT  2  T
  sinc
 n 
 
 T 
The amplitude spectrum is shown (in units of V0 ) for  / T = 0.1.
This (and its phase spectrum) can be calculated using ex_2_2.mcd.
0.1  an = 2 Re(cn )
q
2V0   n 
c ( n ) 0.05 = sinc 
T  T 
bn = −2 Im(cn ) = 0
0
30 20 10 0
n
10 20 30 (as found in worked ex. 1.2)
 8
Worked example 2.3
Derive the complex Fourier
series of the waveform shown
and plot its spectra. Use the
complex coefficients to derive
the real Fourier coefficients.

For the dc term,

T /2 +T / 2 2 +T / 2
2V0 2V0  t 
 
1
c0 = f (t )e − j 0 t
dt = 2 t dt = 2   =0
T
−T / 2
T −T / 2
T  2  −T / 2
T /2 +T / 2
1 2V0
 f (t )e 
− jnt
For all other terms, cn = dt = 2 t e − jnt dt
T T −T / 2
−T / 2
 1 t  − jnt
 te
− j nt
Integrating by parts, dt =  + j e
 (n) n
2
 9
2V0  1 T  − jnT / 2 2V0  1 T  + jnT / 2
 cn = 2  +j  e − 2  −j  e
T  (n)2
2n T  (n)2
2n
But T = 2 , i.e. e − jnT / 2 = e − jn = (− 1)n , e + jnT / 2 = e + jn = (− 1)n
2V0  T  V0 V0
 cn = 2  j  (− 1) = j (− 1) → amplitude spectrum cn =
n n
T  n n n
V0 V0 V0 V0
 c0 = 0, c1 =  j , c 2 =  j , c3 =  j , c 4 =  j , etc.
 2 3 4
 0 = 0,  1 =   / 2,   2 =   / 2,  3 =   / 2,   4 =   / 2 , etc.
Therefore, the amplitude (in units of V0 ) and phase spectra are:
0.4 2

0.3 1

c( n) 0.2  ( n) 0

0.1 1

0 2
10 5 0 5 10 10 5 0 5 10
n n

The coefficients of the real Fourier series representation are:

an = 2 Re(cn ) = 0, bn = −2 Im(cn ) = −2V0 (− 1)n / n (i.e. odd)
 10
Worked example 2.4
Calculate the coefficients of the
complex Fourier series of the
half-rectified sine wave shown
and plot its spectrum. Also find
its real Fourier series.
T T /2
cn =
1
T  f (t )e − jnt
dt =
V0
2 jT  ( )
e − jnt e jt − e − jt dt (1)
0 0

jt − jnt − jt − jnt T / 2

V0 e e e e 
=−  1− n + 1+ n  with T = 2
2T  0
e j = −1, e − j = −1, e − jn = (− 1) (see section 2.2), hence
n
But
V0 1 + (− 1)n
→ cn =
2 1 − n 2
This formula for cn works for all n (including for the dc term n = 0 ,
where c0 = V0 /  ), apart from when n = 1. Returning to (1) above
 11
T /2 − 2 jt T / 2
 e 
 (1 − e )
V − 2 jt V
c1 = 0 dt = 0 t + 2 j 
2 jT
0
2 jT  0
j 2 − j0 V0
But e = 1, e = 1 (see section 2.2) → c1 = − j
4
To find c−1, use c−1 = c1* = + jV0 / 4 . Therefore we can now draw up
the following table and so plot the amplitude and phase spectra:
n −3 −2 −1 0 1 2 3
cn 0 − V0 / 3 jV0 / 4 V0 /  − jV0 / 4 − V0 / 3 0
cn 0 V0 / 3 V0 / 4 V0 /  V0 / 4 V0 / 3 0
n 0 + + / 2 0 −/ 2 + 0
0.4 4

0.3
2
c( n) 0.2  ( n)
0
0.1

0 2
6 4 2 0 2 4 6 6 4 2 0 2 4 6
n n
 12
To calculate the coefficients of the real Fourier series of the
waveform from those of the complex series, as always we use
an = 2 Re(cn ), bn = −2 Im(cn )
The following table covers the coefficients of the real Fourier series
up to n = 3 (remembering that n  0 for the real series):
n 0 1 2 3
an = 2 Re(cn ) 2V0 /  0 − 2V0 / 3 0
bn = −2 Im(cn ) 0 V0 / 2 0 0
2V0   1 1 1 
 f (t ) =  1 + sin t − cos 2 t − cos 4 t − cos 6 t −  
  4 3 15 35 
To examine the single-sided spectra for the real Fourier series (and
the double-sided spectra of the complex Fourier series) for this
particular worked example then please see ex_2_4.mcd.
 13
2.3 The multiplication theorem for two complex Fourier series
If periodic waveforms f (t ) and g (t ) have the same period T then
 
f (t ) = 
n = −
cn e jnt
, g (t ) = 
n = −
d n e jnt with  = 2 / T

The product of the two functions integrated over T is

T T   T


0
f (t )g (t )dt = 
0 m = −
cm e jmt 
n = −
d n e jnt dt = 
m n

cm d n e j(m+ n )t dt
0
when m + n = 0 (i.e. n = −m ), zero otherwise
T
But

0
e j(m+ n )t dt = T

T  

 
1

T 
0
f (t )g (t )dt =
m = −
cm d − m =
m = −
cm d m* multiplication theorem

A common application of the theorem is to find the electrical power

spectrum by applying it to the product v(t ) i(t ).
 14
2.4 Summary of Section 2

Complex Fourier series of a
periodic waveform:
f (t ) = 
n = −
cn e jnt ,  = 2 / T
k +T
1
• Fourier coeffs are found using cn =
T 
k
f (t )e − jnt dt = cn e jn

• cn generates the amplitude spectrum,  n the phase spectrum

0.1 The spectra are again discrete but
q
double-sided. The amplitude
c ( n ) 0.05
spectrum is the same as for the
real series, but halved and
0
30 20 10 0 10 20 30 reflected (e.g. pulse train shown).
n

•  n is the same as for the real series but double-sided (i.e. reflected)
• Coeffs related by cn = 12 (an − jbn ), c−n = 12 (an + jbn ) = cn * (n  0)
→ for the real Fourier series an = 2 Re(cn ), bn = −2 Im(cn )