Fourier Series For A Square Wave

The Fourier series is a powerful tool for representing periodic functions as a sum of sines and cosines. On the flip side, when applied to a square wave, it reveals the underlying harmonic structure of this fundamental waveform. This article explores the Fourier series representation of a square wave, delving into its mathematical derivation, properties, and applications.

Understanding Square Waves

A square wave is a non-sinusoidal periodic waveform in which the amplitude alternates at a regular rate between fixed minimum and maximum values. Even so, it's characterized by its duty cycle, which is the percentage of one period during which the wave is at its high value. A perfect square wave has a 50% duty cycle, meaning it spends half its period at the maximum value and the other half at the minimum value.

Square waves are prevalent in various fields:

• Digital electronics: As clock signals and logic level representations.
• Signal processing: As test signals and for generating other waveforms.
• Music synthesis: For creating distinct timbres.
• Power electronics: In inverters and switching power supplies.

The Power of Fourier Series

The Fourier series allows us to decompose a periodic function into a sum of simpler sinusoidal functions (sines and cosines) with different frequencies and amplitudes. This decomposition is incredibly useful because:

• It allows us to analyze the frequency content of a signal.
• It provides a way to synthesize complex waveforms from simpler components.
• It simplifies the analysis of linear time-invariant (LTI) systems.

For a periodic function f(t) with period T, the Fourier series representation is given by:

f(t) = a0/2 + ∑[an*cos(nωt) + bn*sin(nωt)]

where:

• a0 is the DC component (average value) of the signal.
• an and bn are the Fourier coefficients representing the amplitudes of the cosine and sine components, respectively.
• ω = 2π/T is the fundamental angular frequency.
• n is an integer representing the harmonic number (n=1 is the fundamental frequency, n=2 is the second harmonic, and so on).

Deriving the Fourier Series for a Square Wave

Let's consider a square wave f(t) with amplitude A and period T, defined as:

f(t) =  A,  0 < t < T/2
       -A,  T/2 < t < T

Our goal is to find the Fourier coefficients a0, an, and bn for this square wave.

Calculating a0 (DC Component)

The DC component a0 is the average value of the function over one period:

a0 = (2/T) ∫[0 to T] f(t) dt

Substituting the definition of the square wave:

a0 = (2/T) [∫[0 to T/2] A dt + ∫[T/2 to T] -A dt]
   = (2/T) [A(T/2) - A(T/2)]
   = 0

So, the DC component of a symmetrical square wave is zero.

Calculating an (Cosine Coefficients)

The cosine coefficients an are given by:

an = (2/T) ∫[0 to T] f(t) cos(nωt) dt

Substituting the definition of the square wave:

an = (2/T) [∫[0 to T/2] A cos(nωt) dt + ∫[T/2 to T] -A cos(nωt) dt]

Evaluating the integrals:

an = (2A/T) [ (sin(nωt) / nω) |[0 to T/2] - (sin(nωt) / nω) |[T/2 to T] ]
   = (2A/T) [ (sin(nωT/2) / nω) - (sin(0) / nω) - (sin(nωT) / nω) + (sin(nωT/2) / nω) ]

Since ω = 2π/T, we have ωT = 2π. Therefore:

an = (2A/T) [ (sin(nπ) / nω) - 0 - (sin(2nπ) / nω) + (sin(nπ) / nω) ]

We know that sin(nπ) = 0 and sin(2nπ) = 0 for all integer values of n. Therefore:

an = 0

All the cosine coefficients are zero. This is because the square wave is an odd function (symmetrical about the origin), and the cosine function is an even function. The integral of the product of an odd and an even function over a symmetrical interval is always zero.

Calculating bn (Sine Coefficients)

The sine coefficients bn are given by:

bn = (2/T) ∫[0 to T] f(t) sin(nωt) dt

Substituting the definition of the square wave:

bn = (2/T) [∫[0 to T/2] A sin(nωt) dt + ∫[T/2 to T] -A sin(nωt) dt]

Evaluating the integrals:

bn = (2A/T) [ (-cos(nωt) / nω) |[0 to T/2] - (-cos(nωt) / nω) |[T/2 to T] ]
   = (2A/T) [ (-cos(nωT/2) / nω) + (cos(0) / nω) + (cos(nωT) / nω) - (cos(nωT/2) / nω) ]

Again, since ω = 2π/T, we have ωT = 2π. Therefore:

bn = (2A/T) [ (-cos(nπ) / nω) + (1 / nω) + (cos(2nπ) / nω) - (cos(nπ) / nω) ]

We know that cos(2nπ) = 1 for all integer values of n. Also, cos(nπ) = (-1)^n. Therefore:

bn = (2A/T) [ (-(-1)^n / nω) + (1 / nω) + (1 / nω) - ((-1)^n / nω) ]
   = (2A / nωT) [ 2 - 2(-1)^n ]
   = (4A / nωT) [ 1 - (-1)^n ]

Now, consider two cases:

• If n is even: (-1)^n = 1, so bn = 0
• If n is odd: (-1)^n = -1, so bn = (4A / nωT) * 2 = 8A / nωT

Substituting ω = 2π/T:

bn = 8A / (n * (2π/T) * T) = 8A / (2nπ) = 4A / (nπ)  (for odd n)

The Fourier Series Representation

Combining the results, the Fourier series representation of a square wave with amplitude A and period T is:

f(t) = ∑[n=1,3,5,...] (4A / nπ) * sin(nωt)

This means the square wave can be represented as an infinite sum of sine waves with frequencies that are odd multiples of the fundamental frequency ω = 2π/T. The amplitudes of these sine waves are inversely proportional to the harmonic number n.

Key Properties and Observations

• Odd Harmonics Only: The Fourier series contains only odd harmonics (1st, 3rd, 5th, etc.). This is a characteristic of square waves with 50% duty cycle.
• Amplitude Decay: The amplitude of each harmonic decreases proportionally to 1/n. This means higher harmonics have smaller amplitudes.
• Gibbs Phenomenon: When approximating the square wave with a finite number of terms in the Fourier series, there will be overshoot and undershoot near the discontinuities (the sharp transitions). This is known as the Gibbs phenomenon. As you include more terms, the overshoot and undershoot become narrower but do not disappear entirely. The magnitude of the overshoot remains approximately 9% of the square wave's amplitude, regardless of the number of terms used.
• Symmetry: The square wave's symmetry around the t-axis leads to the absence of cosine terms in its Fourier series representation. This is because the square wave is an odd function.

Approximating the Square Wave

The Fourier series is an infinite sum, but in practice, we can only use a finite number of terms to approximate the square wave. The more terms we include, the better the approximation.

Want to learn more? We recommend words to do with music and which statement is incorrect concerning animal viruses for further reading.

Let's consider the first few terms of the Fourier series:

• 1st harmonic (n=1): (4A/π) * sin(ωt) (The fundamental frequency)
• 3rd harmonic (n=3): (4A/3π) * sin(3ωt)
• 5th harmonic (n=5): (4A/5π) * sin(5ωt)
• 7th harmonic (n=7): (4A/7π) * sin(7ωt)

By adding these terms together, we can get a closer and closer approximation of the square wave.

Applications

The Fourier series representation of a square wave has numerous applications:

• Signal Generation: Electronic circuits can be designed to synthesize square waves by summing sine waves with appropriate amplitudes and frequencies.
• Signal Analysis: Understanding the harmonic content of a square wave helps in analyzing how it interacts with different systems. As an example, knowing the frequency components allows engineers to predict how a square wave will be distorted when passed through a filter.
• Filter Design: Filters can be designed to selectively remove certain harmonics from a square wave, shaping it into a different waveform (e.g., a sine wave).
• Audio Synthesis: Square waves are used as a basic building block in some forms of audio synthesis, particularly in electronic music. By manipulating the amplitudes of the harmonics, different timbres can be created.
• Digital Circuits: The Fourier series is used to analyze the behavior of digital circuits driven by clock signals, which are often square waves.
• Power Electronics: Inverters that convert DC power to AC power often generate a square wave as an intermediate step. Understanding the harmonic content of the square wave allows engineers to design filters to remove unwanted harmonics and produce a cleaner sinusoidal output.

The Effect of Duty Cycle

The above derivation assumes a square wave with a 50% duty cycle. If the duty cycle is different, the Fourier series representation will be different. Specifically:

• The an coefficients will no longer be zero. The presence of cosine terms indicates that the waveform is no longer an odd function.
• The bn coefficients will also change. Their values will depend on the specific duty cycle.
• A DC component (a0) will likely be present if the waveform is not symmetrical about the t-axis.

The general formula for the Fourier series of a periodic signal x(t) with period T is:

x(t) = a0 +  ∑[n=1 to ∞] [an*cos(2πnt/T) + bn*sin(2πnt/T)]

where:

a0 = (1/T) ∫[0 to T] x(t) dt

an = (2/T) ∫[0 to T] x(t)*cos(2πnt/T) dt

bn = (2/T) ∫[0 to T] x(t)*sin(2πnt/T) dt

For a square wave with amplitude A, period T, and duty cycle D (where D is the fraction of the period the signal is at its high value), the signal can be defined as:

x(t) = A,  0 < t < DT
       0,  DT < t < T

Calculating the Fourier coefficients for this general square wave:

a0 = (1/T) ∫[0 to DT] A dt = (1/T) * A*DT = AD

an = (2/T) ∫[0 to DT] A*cos(2πnt/T) dt = (2A/T) * [T/(2πn)] * sin(2πnt/T) |[0 to DT] = (A/(πn)) * sin(2πnD)

bn = (2/T) ∫[0 to DT] A*sin(2πnt/T) dt = (2A/T) * [-T/(2πn)] * cos(2πnt/T) |[0 to DT] = (A/(πn)) * [1 - cos(2πnD)]

So, for a square wave with a general duty cycle D, the Fourier series is:

x(t) = AD + ∑[n=1 to ∞] {[(A/(πn)) * sin(2πnD)]*cos(2πnt/T) + [(A/(πn)) * (1 - cos(2πnD))]*sin(2πnt/T)}

If we let D = 0.5 (50% duty cycle), then:

a0 = A/2

an = (A/(πn)) * sin(nπ) = 0  (for all n)

bn = (A/(πn)) * [1 - cos(nπ)]

If n is even, cos(nπ) = 1, so bn = 0. If n is odd, cos(nπ) = -1, so bn = (2A/(πn)). Thus,

x(t) = A/2 + ∑[n=1,3,5,...] [(2A/(πn))*sin(2πnt/T)]

This differs from the symmetrical square wave representation we initially derived by a DC offset of A/2. To obtain the symmetrical square wave, we must center it around zero, subtracting A/2 from the equation. This gives:

x(t) - A/2 = ∑[n=1,3,5,...] [(2A/(πn))*sin(2πnt/T)]

For a square wave varying between +A/2 and -A/2, the above equation correctly describes it. To scale it up to vary between +A and -A, multiply the entire right-hand side by 2:

x(t) = ∑[n=1,3,5,...] [(4A/(πn))*sin(2πnt/T)]

Which matches our original result (with ω = 2π/T).

Limitations and Considerations

• Ideal Square Wave: The analysis assumes an ideal square wave with instantaneous transitions. In real-world scenarios, transitions are never instantaneous and will affect the harmonic content.
• Computational Complexity: Calculating a large number of terms in the Fourier series can be computationally expensive.
• Gibbs Phenomenon: The Gibbs phenomenon introduces artifacts when approximating the square wave, which can be problematic in certain applications.

Conclusion

The Fourier series provides a valuable tool for understanding the composition of a square wave. By decomposing the square wave into its constituent sine wave components, we gain insights into its frequency content, properties, and behavior in various systems. While approximations are necessary in practice, the Fourier series offers a powerful framework for analyzing and synthesizing this fundamental waveform, enabling us to manipulate and put to use it effectively in a wide range of applications. Understanding how duty cycle affects the Fourier series allows for a more complete and adaptable analysis of square waves in their various forms.