ENG349 Sinusoidal Waveforms

A sinusoidal waveform is the starting or end point for many of the power conversion
operations that will be studied in this unit. An important design objective will be the
ability of the switching circuit to generate acceptable sinusoidal waveforms.
Consequently, the main aims of this lesson will be
1.
2.

to investigate how to convert a pulse waveshape into an acceptable sinusoidal

waveshape, and
to propose a suitable metric that measures the acceptability of the approximate
sinusoid to the real thing.

The Sinusoidal Waveform

A sine wave signal has the general form

x t

X M cos

(1)

where XM is the amplitude of the sine wave, is the angular frequency, and is the
phase shift. The design specifications are usually given in terms of frequency in Hertz
(cycles/sec), which is related to angular frequency by
f

(2)

The period of the waveform measures the time taken for the waveform to repeat itself; it
is related to frequency by

1
f

(3)

The average value of a sinusoidal waveform is zero, that is

T

X av

1
x t dt
T 0

(4)

The root-mean-squared (rms) value is

X rms

1
T

x 2 t dt
0

XM
2

(5)

ENG349 Sinusoidal Waveforms

Note: the rms value is useful because it is related to the power levels in a signal. For
v2 t
example, the instantaneous power delivered to a resistor is p t
. The
R
average power dissipation is then given by
1
T

Pav

v2 t
dt
R
0

2
Vrms
R

This relationship is true irrespective of the shape of the voltage waveform. If the shape is
sinusoidal, then the rms value for the voltage is given by eq.(5).

Phasor Analysis of a Circuit

When the input signal to a linear circuit is a sinusoid, and if the circuit has reached its
steady-state, then all voltages and currents throughout the circuit will be sinusoids of the
same frequency as the input signal.
The circuit equations for sinusoidal signals can be very messy to deal with. Fortunately, it
is possible to represent sinusoids by phasors, whereby the circuit equations reduce to a
set of algebraic equations. The only downside is that phasors are complex-valued
variables.
A sinusoidal analysis of a circuit can be carried out using the following steps:
1.

Convert the excitation signal(s) into equivalent phasor(s). For example, vin(t)
VM 0 . All the unknown
= VM cos( t) is represented by the phasor Vin
voltages and currents around the circuit will be represented by their equivalent
phasors. Note that each phasor will have two unknowns, a magnitude and a
phase. Also, notice that the frequency of the original signal does not appear in
the phasor representation.

2.

Convert each circuit element to its equivalent impedance. For resistors,

inductors and capacitors these are:
R

L
1

j C

(6)
j

1
C

ENG349 Sinusoidal Waveforms

For example, the voltage-current relationship for an inductpr is

VL

Z L IL

L IL

Note that the frequency is needed in order to calculate the impedances of the
energy storage elements. However, once these impedance values have been
determined, then frequency will not be used again until the final conversion
back into sinusoidal functions in Step 5.
3.

Use KVL and KCL to write down the set of circuit equations.

4.

Solve the set of equations for the unknown phasor magnitudes and phases.

5.

Convert the final phasor quantities back into their equivalent sinusoidal forms.
IM
For example, if I
then i(t) = IM cos ( t + ).

Example
Consider the following R-L circuit shown in fig.1. If vin (t )
an expression for the output voltage vO(t).

5 cos 300 t V , then find

0.5H
+
vin(t)

10

vo(t)
-

Figure 1
Step 1: The input phasor is:

Vin

5 0

Step 2: The impedances of the two elements are:

ZR
R
10
ZL
j L
j 300 0.5

j 150

150 90

The circuit relabeled in terms of phasors and impedances is shown in figure 2.

ENG349 Sinusoidal Waveforms

j150
+

5 /_ 0o

Vo

10
-

Figure 2
Steps 3 and 4:
The voltage drop across the output resistor is
Vo

10 I

(7)

From KVL around the loop:

5 0

j 150 I

(8)

10 I

From (8):

5 0
10

(9)

j 150

Finally, substituting eq.(9) into (7):

Vo

10

5 0

10
j 150
50 0
150.33 86.19
0.3326
86.19

Step 5: Converting the output phasor into the time domain:

vo t

0.3326 cos 300 t

86.19

ENG349 Sinusoidal Waveforms

Transfer Functions
It is useful to obtain the ratio of an output phasor to an input phasor as a general function
of frequency. This function is called the transfer function of the circuit. For example,
consider the circuit from the previous example, but with the input source left undefined
and the frequency ( ) kept as an independent variable.

Vo

10 I

Vin

j 0.5 I
Vo
Vin

(10)

10 I
10
10

j 0.5

The transfer function is often written in terms of its magnitude (or gain) and phase:
10

H
102

0.5

(11)

and
H

atan
atan

0.5
10

20

These functions are plotted in figure 3.

(12)

ENG349 Sinusoidal Waveforms

1 vmagh

2 ph_vh

900m

Plot1
mag_H

700m
500m
300m
100m
1

Plot2
ph_H in degrees

0
-20.0
-40.0
-60.0
-80.0
100m

10
frequency in hertz

100

2
1k

Figure 3
A useful single metric for the shape of a magnitude response is the cut-off frequency, fC.
1
This is defined as the frequency at which the gain drops by
from its low frequency
2
value. In the above example, we can re-write eq.(11) as

10

H
102

0.5
1

0.5
10

(13)

1
2

20
1
2

1
C

ENG349 Sinusoidal Waveforms

where

= 20 rad/sec is the cut-off angular frequency; the cut-off frequency in Hertz is

fC

20
2

3.183 Hz

If we return to the original circuit, and leave the resistor and inductor as variables, then
we can obtain an expression for the cut-off frequency that can be used as a design
formula this formula will allow us to select suitable values for inductor and resistor to
achieve a required cut-off frequency.

Vo
Vin

R
j

1
j

L
R

L
R

(14)

1
2

1
C

where
C

R
L

(15)

Reconstructing the output signal

If the input signal is the sinusoid vin t
function, the output signal will be
vout t

VM cos

VM cos t V , then from the transfer

(16)

From the graphs given in figure 3, it is clear that high frequency sinusoids will be much
reduced in amplitude compared to lower frequency sinusoids. Consequently, this circuit
behaves as a low pass filter it passes low frequency signals but blocks or filters out high
frequency signals.

Example
An input signal

vin t

5 cos 2 t

1cos 2 10 t V

is applied to the circuit shown in figure 1. This signal and the output signal are shown in
figure 4. The 10 Hz sinusoid is much less noticeable in the output waveform.

ENG349 Sinusoidal Waveforms

1 vin

10.0

2 vo

30.0

-10.0

20.0

vo in volts

Plot1
vin in volts

10.0

2
-20.0

-30.0

-10.0

4.40

4.80

5.20
time in seconds

5.60

6.00

Figure 4

Fourier Series
In this unit we will deal mainly with pulsed waveforms, which are clearly not sinusoidal.
However, in most cases the pulsed waveforms will be periodic, in which case they can be
represented as a series of sine waves called a Fourier series. The frequencies of these
sine waves will be the fundamental frequency (which is the frequency of the original
pulsed waveform) and its harmonics (which are integral multiples of the fundamental
frequency). Of major concern to us will be to find ways of keeping the fundamental
component while eliminating all the harmonics. If we can do this, then we can generate a
pure sine wave from the original switching waveform.
In general, a periodic signal x(t), with period T can be expanded as
x t

X av

X n cos n

n 1

X av
n

2 nt
X n cos
T
1

(17)
n

ENG349 Sinusoidal Waveforms

Xav is the average value of the steady-state waveform, which we looked at in the previous
2
lesson; o
is the fundamental angular frequency, while { n o; n = 2, } are
T
the harmonic frequencies.
The rms value of the signal is given by

X rms

1
2

X av2

X n2

(18)

n 1

The rms expression forms the basis for a measure of the sinusoidal-ness of a periodic
waveform. The percent total harmonic distortion (%THD) is defined by

X n2
%THD

100

n 2

(19)

X 12

Notice that %THD does not include the average component (Xav) because this component
has no effect on the shape of the waveform. If a periodic waveform is nearly sinusoidal
in shape, then its harmonic components {Xn; n = 2, } will be small, and so the %THD
will be small.
It is possible to calculate the coefficients { Xn; n = 1, . } for any given periodic
waveform. For example, the Fourier series expansions for two square waveforms, with
and without an average value (see fig.5), are:
xsqA t

xsq 0 t

A
2

4A

2A

cos

cos

1
cos 3
3

1
cos 3
3

1
cos 5
5

1
cos 5
5

(20)

(21)

ENG349 Sinusoidal Waveforms

10

t
T

t
-A
T

Figure 5

To calculate the Fourier coefficients refer to the relevant formulae are given in Mohan,
pp.39-40. For complex waveforms we will leave the task of calculating Fourier
coefficients to the circuit simulator.

ENG349 Sinusoidal Waveforms

11

Power and Power Factor

A situation that often arises in power systems is where a sinusoidal voltage is applied
across a non-linear element, causing a periodic, but non-sinusoidal current to flow. If the
voltage is given by

vS t

VM cos

then the current can be represented by the Fourier series

iS t

I av

I Mn cos n t

n 1

The average power delivered to the element is given by

T

Pav

1
vS t iS t dt
T 0
VM I M 1
cos
2

Notice that the average power depends only on the fundamental component of the
current. [Note: this simple result, and the earlier result given by eq.(18), are
consequences of the orthogonality of the harmonic set of sinusoids, whereby
T

cos n t cos m t dt

for all m n ]

The term cos 1 is the power factor. In our example we assumed the voltage has zero
phase, and the fundamental component of the current has the phase 1 . In more general
terms the power angle is the phase of the voltage minus the phase of the fundamental of
the current.

Constructing Sine Waves from Square Waves

We can construct a pure sine wave from an arbitrary periodic wave x(t) by eliminating
(filtering out) all the unwanted harmonics. The preceding work on transfer functions
gives us an idea for how to do this we design a transfer function such that the
magnitude for frequencies about o is much greater than the magnitude at the harmonic
frequencies.

ENG349 Sinusoidal Waveforms

12

Example
Consider a square wave with peak values of 5V and period of 20ms. We could attempt
to construct a sine wave by passing this square wave signal through the low-pass filter
circuit shown in figure 1. Recall that the magnitude response for this circuit is

(22)

1
C

where the cut-off frequency is given by

R
L

(23)

In order to achieve the best possible reduction in harmonic components, it is actually

better to have the fundamental frequency outside the pass-band (i.e. o > C). For
frequencies well outside the passband, the gain function reduces to

1
2

for

1
C

(24)

From this expression, it can be seen that the gain at n

For convenience, we will choose
for the inductor becomes:
L

10

o.

will be

1
times the gain at
n

o.

If R is fixed at 10 , then the design equation

10
2 50

0.0318 H

When this circuit is simulated, the input and output waveforms are as shown in figure 6.

ENG349 Sinusoidal Waveforms

1 vin

30.0

10.0

20.0

13

2 vo

10.0

vin in volts

Plot1
vo in volts

-10.0

-20.0
2

-10.0

-30.0

10.0m

30.0m

50.0m
time in seconds

70.0m

90.0m

Figure 6

The output waveform looks closer to a sine wave that the input square wave, however the
result is not great. We can improve on this result by using a filter that has a sharper cutoff. One approach is to turn our original first-order filter into a second-order filter by
introducing a second energy-storage element. A possible circuit configuration is shown
in figure 7.
L
+
vin

vo
-

Figure 7

ENG349 Sinusoidal Waveforms

14

This circuit has a magnitude response given by

(25)

1
C

where
L
and
C

2 C R2

(26)
1
LC

For frequencies well above

C,

the gain simplifies to

for

We can now see that the gain at n

(27)

will be

1
times the gain at
n2

o.

Example
Revisit the case where fo = 50 Hz, and R = 10 . The required component values are:
C = 225 F
L = 45 mH
This design is tested in the simulator. The frequency response is given in figure 8. Note
that this response confirms that the cut-off frequency is 50Hz. The gain for the first-order
design is also shown in the figure the sharper response of the second-order filter is clear
to see.

ENG349 Sinusoidal Waveforms

1 vmago1

2 vmago2

15
3 ph_vo2

Plot1
vmago1, vmago2 in volts

900m
700m
500m
300m
100m

1
2

Plot2
ph_vo2 in degrees

0
-40.0
-80.0
-120
-160
100m

10
frequency in hertz

100

3
1k

Figure 8

The design is then used to filter the square wave from the previous example. The results
of a transient analysis are shown in fig.9.

ENG349 Sinusoidal Waveforms

21.0

36.0

-5.00

11.0

26.0

-15.0

-25.0

1.00

-9.00

vo2 in volts

5.00

vo1 in volts

Plot1
vin in volts

1 vin

2 vo1

16
3 vo2

16.0
2

6.00

3
-35.0

-19.0

-4.00

30.0m

40.0m

50.0m
time in seconds

60.0m

70.0m

Figure 9
The output of the first-order system is also given in figure 9 for comparison. It is clear
that the second-order filter given a more sinusoidal-looking waveform.
The simulation program can also carry out a Fourier analysis of the waveforms. The
results of this analysis give %THD figures of:

Square wave input

First-order output
Second-order output

%THD
42.9
16.3
5.4

The %THD figures support the case that the second-order circuit generates a better sine
wave.

ENG349 Sinusoidal Waveforms

17

Pulse-Width Modulation
An alternative way of generating better quality sine waves is to use a high frequency
pulse generator to produce a pulse train with variable width pulses, where the width of
each pulse reflects the amplitude of the required sine wave. An example of this type of
waveform is shown in figure 10, along with the filtered waveforms from our first and
second order filters. This approach to sine wave generation will be studied when we look
at dc ac inverters later in this unit.

1.20

7.40

2.60

-800m

5.40

600m

-2.80

vo2 in volts

4.60

vin in volts

Plot1
vo1 in volts

1 vin

2 vo1

3 vo2

3.40

2
-1.40

-4.80

1.40

-3.40

-6.80

-600m

50.0m

60.0m

70.0m
time in seconds

80.0m

90.0m

Figure 10

The full Fourier results for the second-order filtered output are given below. Notice that
%THD = 0.663%, which is a clear improvement over all earlier results. The column
labeled Norm.Mag lists the values of the Fourier coefficients normalized to the value of
the fundamental component X1. Notice that the second harmonic is over 10000 times
smaller than the fundamental component.

ENG349 Sinusoidal Waveforms

18

Fourier analysis for v(2):

No. Harmonics: 10, THD: 0.662952 %, Gridsize: 200, Interpolation Degree: 1
Harm
-----0
1
2
3
4
5
6
7
8
9

Freq
-----0
50
100
150
200
250
300
350
400
450

Magnitude
-----------------0.0002034
0.636636
5.05518e-005
2.86793e-005
2.43965e-005
1.49108e-005
0.00033233
9.88996e-006
0.00420696
9.0717e-006

Phase
--------0
-89.972
-52.719
11.9237
-73.683
-130.8
-76.433
-112.63
-79.81
-56.609

Norm. Mag
-------------0
1
7.94047e-005
4.50483e-005
3.8321e-005
2.34213e-005
0.00052201
1.55347e-005
0.00660811
1.42494e-005

Norm. Phase
--------------0
0
37.2527
101.895
16.2886
-40.827
13.539
-22.663
10.1612
33.3623

Revision History
Revision File Name
Description
ENG349 Lesson 3
Original document
ENG349lesson3_2008s1
Added Appendix1
ENG349TopicNotesSinusoidalWaveforms2009s1

Author
Greg Crebbin
Martina Calais
Martina Calais

ENG349 Sinusoidal Waveforms

19

Appendix 1
Average power in a circuit with an undistorted voltage and distorted current
T

Pav

1
v s (t )is (t )dt
T 0

With
v s (t ) VM cos( t )
m

i s (t )

I av

I Mn cos(n t

n 1

T
1 VM cos( t ) I av VM cos( t ) I M 1 cos( t 1 ) VM cos( t ) I M 2 cos(2 t 2 ) ...
Pav
dt
T 0 ... VM cos( t ) I Mm cos(m t m )
Applying the trigonometric identity
1
cos x cos y
(cos( x y ) cos( x y ))
2
And
cos( x) cos( x)
We obtain for the product
1
VM cos( t ) I M 1 cos( t 1 )
VM I M 1 (cos 1 cos(2 t 1 ))
2
Integrating the above over one period T and dividing by T results in
T
1 1
1
VM I M 1 (cos 1 cos(2 t 1 ))
VM I M 1 (cos 1 )
T 02
2
Similarly we obtain for the product
1
VM cos( t ) I M 2 cos(2 t
VM I M 2 (cos( t 1 ) cos(3 t
2)
2 ))
2
Integrating the above over one period T and dividing by T results in
T
1 1
VM I M 2 (cos( t 1 ) cos(3 t 2 )) 0
T 02
Similarly the other products of the sum can be evaluated, resulting in the average power
only depending on the fundamental component of the current:
1
Pav
VM I M 1 cos( 1 )
2