Chapter 4.

Fourier Series II

Kwangsoo Kim

Hanbat National University

Department of Electronics Engineering
Fourier series

Properties of Fourier Series

System Analysis

Fourier Series Transformations

Properties of Fourier Series 2 / 23 .

The Dirichlet conditions

• Any single-valued periodic function 𝑥(𝑡) that satisfies the

Dirichlet conditions can be expanded into a Fourier series
• The Dirichlet conditions are
1. 𝑥(𝑡) has at most a finite number of discontinuities in one
period;
2. 𝑥(𝑡) has at most a finite number of maxima and minima in one
period;
3. 𝑥(𝑡) is bounded.

Properties of Fourier Series 3 / 23 .

Some properties of Fourier series I

1. The Fourier series converges to the value of 𝑥(𝑡) at every point

of continuity where 𝑥(𝑡) has a right-hand and a left-hand
derivative, whether these derivatives are the same or different.
𝑥(𝑡)

⋯ ⋯
𝑡

2. If 𝑥(𝑡) has a discontinuity at a point, the Fourier series

converges to the mean of the limits approached by 𝑥(𝑡) from
the right and from the left; that is,
∞
𝑥(𝑡− +
𝑎 ) + 𝑥(𝑡𝑎 )
∑ 𝐶𝑘 𝑒𝑗𝑘𝜔0 𝑡𝑎 =
𝑘=−∞
2

Properties of Fourier Series 4 / 23

Some properties of Fourier series II
3. Almost any continuous function 𝑥(𝑡) of period 𝑇0 can be
uniformly approximated by a truncated Fourier series with any
preassigned degree of accuracy, where
𝑁
𝑥𝑁 (𝑡) = ∑ 𝐶𝑘 𝑒𝑗𝑘𝜔0 𝑡
𝑘=−𝑁

The higher harmonics are added, the less the error

𝑥(𝑡) − 𝑥𝑁 (𝑡) becomes.
4. The Fourier coefficients are the “optimal” in the sense that
they minimize the mean-square error, defined as
1 2
mean-square error = ∫ |𝑥(𝑡) − 𝑥𝑁 (𝑡)| 𝑑𝑡.
𝑇0 𝑇0

There is no other choice of coefficients in the harmonic series

will produce a smaller mean-square error.
Properties of Fourier Series 5 / 23
Some properties of Fourier series III

5. A sum of trigonometric functions of 𝜔0 𝑡 that is periodic is its

own Fourier series.

𝑥(𝑡) = 10 + 3 cos 𝜔0 𝑡 + 5 cos(2𝜔0 𝑡 + 30∘ ) + 4 sin 3𝜔0 𝑡

6. The Fourier series of a periodic sum of periodic functions is

equal to the sum of the Fourier series for the functions.

Properties of Fourier Series 6 / 23 .

The Gibbs phenomenon
• The ripples in the waveform of the series become narrower as
the number of terms used becomes larger. However, the
amplitudes of the ripple nearest each discontinuity do not
approach zero, but instead, approach approximated 9 percent
of the height of the discontinuity.

Properties of Fourier Series 7 / 23 .

Fourier series

Properties of Fourier Series

System Analysis

Fourier Series Transformations

System Analysis 8 / 23 .
Basic concepts I
• Here, we consider the analysis of stable LTI systems with
periodic inputs.
• The LTI system “steady-state” response to the “periodic”
input signal

|𝑋| cos(𝜔1 𝑡 + 𝜙) → |𝑋||𝐻(𝑗𝜔1 )| cos(𝜔1 𝑡 + 𝜙+∠𝐻(𝑗𝜔1 ))

• If the input signal is periodic,

∞
𝑥(𝑡) = 𝐶0𝑥 + ∑ 2|𝐶𝑘𝑥 | cos(𝑘𝜔0 𝑡 + 𝜃𝑘𝑥 )
𝑘=1
∞
→ 𝑦𝑠𝑠 (𝑡) = 𝐶0𝑦 + ∑ 2|𝐶𝑘𝑦 | cos(𝑘𝜔0 𝑡 + 𝜃𝑘𝑦 )
𝑘=1

where
𝐶𝑘𝑦 = |𝐶𝑘𝑦 |∠𝜃𝑘𝑦 = 𝐻(𝑗𝑘𝜔0 )𝐶𝑘𝑥

System Analysis 9 / 23
Basic concepts II

• Note that the variation of the system sinusoidal response

(𝐶𝑘𝑦 ) is a function of frequency.
‵
𝐶𝑘𝑦 = 𝐻(𝑗𝑘𝜔0 )𝐶𝑘𝑥

This variation is called the system frequency response.

• The analysis in this section does not give a good indication of
a plot of the steady-state response as a function of time;
instead, it yields the frequency spectrum of the output signal.

System Analysis 10 / 23 .
Example: LTI system response for a square-wave input

• Consider an LTI system

𝑥(𝑡) 𝑦(𝑡)
ℎ(𝑡)

where
1
ℎ(𝑡) = 𝑒−𝑡 𝑢(𝑡) ⟺ 𝐻(𝑠) =
𝑠+1
• The input signal is the square wave with the fundamental
period to be 2𝜋.
𝑥(𝑡)
4

−𝜋 0 𝜋 2𝜋 3𝜋 𝑡

System Analysis 11 / 23 .
 • The Fourier series of 𝑥(𝑡) is given by
∞ ∞
4 −𝑗𝜋/2 𝑗𝑘𝜔0 𝑡
𝑥(𝑡) = 𝐶0𝑥 + ∑ 𝐶𝑘𝑥 𝑒𝑗𝑘𝜔0 𝑡 = 2 + ∑ 𝑒 𝑒
𝑘=−∞ 𝑘=−∞
𝜋𝑘
𝑘≠0 𝑘≠0

• The system frequency response is

1 1
𝐻(𝑗𝑘𝜔0 )∣ = =√ ∠ tan−1 (−𝑘)
𝜔0 =1
1 + 𝑗𝑘 1 + 𝑘2

• The Fourier coefficient of the output, 𝐶𝑘𝑦 is

1 4 𝜋
𝐶𝑘𝑦 = 𝐻(𝑗𝑘𝜔0 )𝐶𝑘𝑥 = √ [ ] ∠ − − tan−1 (𝑘)
1 + 𝑘2 𝜋𝑘 2

System Analysis 12 / 23 .
 • First 4 nonzero Fourier coefficients

𝑘 𝐶𝑘𝑥 𝐶𝑘𝑦 |𝐶𝑘𝑥 | |𝐶𝑘𝑦 | |𝐶𝑘𝑦 |/|𝐶𝑘𝑥 |

0 2 2 2 2 1
4 ∘ 4
1 𝜋 ∠ − 90
√
𝜋 2
∠ − 135∘ 1.273 0.900 0.707
4 ∘ 4
2 3𝜋 ∠ − 90
√
3𝜋 10
∠ − 161.6∘ 0.424 0.134 0.316
4 ∘ 4
√ ∠ − 168.7∘
3 5𝜋 ∠ − 90 5𝜋 26
0.255 0.050 0.196

• The system attenuates the high harmonics relative to the low

harmonics. (See the last column in the above table) ⇝ a
low-pass system.
• The frequency-response concept will be extended to aperiodic
signals in Chapter 5 and 6.

System Analysis 13 / 23 .
Fourier series

Properties of Fourier Series

System Analysis

Fourier Series Transformations

Fourier Series Transformations 14 / 23 .

A constant offset in the amplitude
• A triangular wave
𝑥(𝑡)
𝑋0

−𝑇0 𝑇0 𝑡
∞
𝑋0 −2𝑋0 𝑗𝑘𝜔0 𝑡
𝑥(𝑡) = + ∑ 𝑒
2 𝑘=−∞
(𝜋𝑘)2
𝑘odd
• The shifted down triangular waveform
𝑦(𝑡) 𝑋0
2
−𝑇0 𝑇0
𝑋
𝑡
− 20

∞
𝑋0 −2𝑋0 𝑗𝑘𝜔0 𝑡
𝑦(𝑡) = 𝑥(𝑡) − = ∑ 𝑒
2 𝑘=−∞
(𝜋𝑘)2
𝑘odd
Fourier Series Transformations 15 / 23 .
Amplitude transformation effect on the Fourier coefficients
• The general amplitude transformation, 𝑦(𝑡) = 𝐴𝑥(𝑡) + 𝐵
∞
𝑦(𝑡) = 𝐶0𝑦 + ∑ 𝐶𝑘𝑦 𝑒𝑗𝑘𝜔0 𝑡
𝑘=−∞
𝑘≠0

∞
⎡ ⎤
= 𝐴 ⎢𝐶0𝑥 + ∑ 𝐶𝑘𝑥 𝑒𝑗𝑘𝜔0 𝑡 ⎥ + 𝐵
⎢ ⎥
𝑘=−∞
⎣ 𝑘≠0 ⎦
∞
= (𝐴𝐶0𝑥 + 𝐵) + ∑ 𝐴𝐶𝑘𝑥 𝑒𝑗𝑘𝜔0 𝑡
𝑘=−∞
𝑘≠0

• Therefore,
𝐶0𝑦 = 𝐴𝐶0𝑥 + 𝐵
and
𝐶𝑘𝑦 = 𝐴𝐶𝑘𝑥 , 𝑘≠0
Fourier Series Transformations 16 / 23 .
Example of amplitude transformation effect
• We know that the Fourier series of the sawtooth signal 𝑥(𝑡):
𝑥(𝑡)

𝑋0

−𝑇0 0 𝑇0 2𝑇0 𝑡
∞
𝑋0 𝑋
𝑥(𝑡) = + ∑ 0 𝑒𝑗𝜋/2 𝑒𝑗𝑘𝜔0 𝑡
2 𝑘=−∞
2𝜋𝑘
𝑘≠0

• Then, what is the Fourier series of the following signal?

𝑦(𝑡)

1
−𝑇0 0 𝑇0 2𝑇0 𝑡
−3
Fourier Series Transformations 17 / 23 .
 • Find the relation between 𝑥(𝑡) and 𝑦(𝑡)

4
𝑦(𝑡) = 𝐴𝑥(𝑡) + 𝐵 = − 𝑥(𝑡) + 1
𝑋0
• Then,

4 𝑋
𝐶0𝑦 = 𝐴𝐶0𝑥 + 𝐵 = (− ) 0 + 1 = −1
𝑋0 2

and
4 𝑋 2 −𝑗𝜋/2
𝐶𝑘𝑦 = 𝐴𝐶𝑘𝑥 = (− ) 0 𝑒𝑗𝜋/2 = 𝑒 , 𝑘 ≠ 0.
𝑋0 2𝜋𝑘 𝜋𝑘

Fourier Series Transformations 18 / 23 .

Time transformation effect on the Fourier coefficients I

• The general time transformation is

𝑦(𝑡) = 𝑥(𝑎𝑡 + 𝑏)

• When 𝑎 = −1, 𝑏 = 0 (time reversal),

∞ ∞
𝑦(𝑡) = ∑ 𝐶𝑘𝑦 𝑒𝑗𝑘𝜔0 𝑡 = 𝑥(−𝑡) = ∑ 𝐶𝑘𝑥 𝑒−𝑗𝑘𝜔0 𝑡
𝑘=−∞ 𝑘=−∞

Let 𝑘 = −𝑚, then

∞ ∞
𝑦(𝑡) = ∑ 𝐶−𝑚𝑥 𝑒𝑗𝑚𝜔0 𝑡 = ∑ 𝐶𝑘𝑥
∗ 𝑗𝑘𝜔0 𝑡
𝑒
𝑚=−∞ 𝑘=−∞

Hence, the only angles of the Fourier coefficients are affected.

Fourier Series Transformations 19 / 23

Time transformation effect on the Fourier coefficients II

• When 𝑎 = 1, 𝑏 = −𝑡0 (time shift),

∞
𝑦(𝑡) = 𝑥(𝑡 − 𝑡0 ) = ∑ 𝐶𝑘𝑥 𝑒𝑗𝑘𝜔0 (𝑡−𝑡0 )
𝑘=−∞
∞
= ∑ (𝐶𝑘𝑥 𝑒−𝑗𝑘𝜔0 𝑡0 )𝑒𝑗𝑘𝜔0 𝑡
𝑘=−∞

Hence, the effect of a time shift is to change the angles of the

Fourier coefficients.
• The general case is referred to Problem 4.28

Fourier Series Transformations 20 / 23 .

Key equations of Chapter Four I

Exponential form of Fourier series

∞
𝑥(𝑡) = ∑ 𝐶𝑘 𝑒𝑗𝑘𝜔0 𝑡 , 𝐶𝑘 = 𝐶−𝑘
∗

𝑘=−∞
Combined trigonometric form of Fourier series
∞
𝑥(𝑡) = 𝐶0 + ∑ 2|𝐶𝑘 | cos(𝑘𝜔0 𝑡 + 𝜃𝑘 )
𝑘=1
Trigonometric form of Fourier series
∞
𝑥(𝑡) = 𝐴0 + ∑[𝐴𝑘 cos 𝑘𝜔0 𝑡 + 𝐵𝑘 sin 𝑘𝜔0 𝑡]
𝑘=1
Relation of different forms of Fourier series
2𝐶𝑘 = 𝐴𝑘 − 𝑗𝐵𝑘 ; 𝐶𝑘 = |𝐶𝑘 |𝑒𝑘𝜃𝑘 ; 𝐶0 = 𝐴0
1
Fourier series coefficients 𝐶𝑘 = ∫ 𝑥(𝑡)𝑒−𝑗𝑘𝜔0 𝑡 𝑑𝑡
𝑇0 𝑇0

Fourier Series Transformations 21 / 23

Key equations of Chapter Four II

sin 𝑥
sinc function sinc𝑥 =
𝑥
Steady-state output expressed as Fourier series .
∞ ∞
𝑗𝑘𝜔0 𝑡
𝑥(𝑡) = ∑ 𝐶𝑘 𝑒 → 𝑦𝑠𝑠 (𝑡) = ∑ 𝐻(𝑗𝑘𝜔0 )𝐶𝑘 𝑒𝑗𝑘𝜔0 𝑡
𝑘=−∞ 𝑘=−∞

Fourier Series Transformations 22 / 23 .

Homeworks

1. Derive the Fourier series for all the signals in Table 4.3
2. Plot the frequency spectra for all the signals in Table 4.3
3. Problems
• 4.13
• 4.17
• 4.18
• 4.23 (Except (b))
• 4.27
• 4.34
• 4.36

Fourier Series Transformations 23 / 23 .