C H A R

Harmonic Oscillators

3.1 INTRODUCTION

Consider a system in static or dynamic stable equilibrium. When such a system is displaced
slightly from its equilibrium position, the resulting oscillatory motion is called harmonic mo-
tion. Such motions occur frequently in nature and are investigated, both from a practical as well
as a theoretical point of view, in physics and engineering. A few examples of such motions are
elastic springs, bending beams, pendula, vibrating strings, resonance of air cavities, and the mo-
tion of charges in certain electrical circuits and cavities.
To start, we shall study the motion of a linear harmonic oscillator (motion resulting from
a small displacement of a system from its equilibrium) in one dimension. Unavoidable inclu-
sion of friction in such motion leads to the investigation of a damped harmonic oscillator. To
maitain oscillatory motion in the presence of friction, some external force must be applied. Such
an oscillating system is called a forced or driven oscillator.
When the displacement of the system from equilibrium is large, the system is no longer
linear. Such oscillating systems are called nonlinear. We divide our study into two parts. This
chapter is mainly devoted to the study of a linear system including damped and forced harmonic
oscillators. The study of nonlinear oscillations, the electrical equivalent of mechanical oscilla-
tors and multidimensional oscillators, will be investigated in Chapter 4. It may be pointed out
that, in general, oscillations of systems occurring in nature are nonlinear. But their approxima-
tion to linear systems allows us to use strong analytical techniques developed for this purpose.

3.2 LINEAR AND NONLINEAR OSCILLATIONS

Consider a particle of mass m moving in an arbitrary conservative force field for which the
potential energy V(x) of the particle as a function of its displacement is represented by a

58
Sec. 3.2 Linear and Nonlinear Oscillations 59

heavy curve, as shown in Fig. 3.1. For a conservative force field, the total energy E of the par-
ticle is
E = K + V = constant (3.1)
If x is the velocity of the particle,

E = \ mx2 + V(x) (3.2)

which when solved for x yields

dx
x = dt [E - V(x)] (3.3)
m
If E = Eo, as shown in Fig. 3.1, then Eo — V(x) = 0 and x = 0; that is, the particle stays
at rest in a stable equilibrium at x = x0. Let us consider the case in which the particle energy Ex
is slightly greater than Eo. For x <xl and x > x2, x will be imaginary; hence the particle cannot
exist in these regions. Thus a particle with energy Ex is constrained to move in a potential well
i or valley) between JC, and x2. The particle moving to the right is reflected back when it reaches
.Y-I, and when traveling to the left it is reflected again at xv The points X\ and x2 are called turn-
ing points, and the velocity of the particle at these points is zero. These points are obtained by
solving E1 — V(x) = 0. In between these points, the velocity of m changes continuously de-
pending on the value of V(x). Hence a particle in a potential well moves back and forth and os-
cillates between x1 and x2 when its energy is greater than Eo.
The position x(t) of a particle moving in potential well can be found by integrating
Eq. (3.3); that is,
dx
(3.4)

V(x)

Parabolic
\ \ /
\ \ /
\\ /
\\ / . _
\\
\\
\\
1/
1 /
/X -----—_'
Y 1/
m
\ y
— 1^

Figure 3.1 A particle of mass m and energy E is moving in an arbitrary poten-

tial energy function V(x) shown by the solid heavy curve. The dotted curve is the
parabolic potential approximation of the arbitrary potential.
 Harmonic Oscillators Chap. 3
60

while the time period T of one complete oscillation is given by

. rXl dx (3.5)
2
-I, Vfi - V(x)
Equations (3.4) and (3.5) cannot be solved unless we know the form of the potential func-
tion V(x). The motion of the particle can be limited to the region in the neighborhood of x0, and
for small displacements, such as these, it is possible to approximate the arbitrary potential func-
tion V(x) by a parabolic potential shown by the dotted curve in Fig. 3.1. As we shall show later,
this potential may be written as V(x) = \k(x — x0)2, where k is a constant, thereby enabling us
to solve Eqs. (3.4) and (3.5).
Suppose a particle is oscillating about a point of stable equilibrium x0, where the minimum
potential is V(x0) at x = xQ. Let us expand the potential function V(x) in a Taylor series about the
point x0.
dV d2V
(x - x0)2

(* - xo)3 + XT (3.6)
6\ dx3 24 V dxA = x0

We limit our discussion to small displacements in symmetrical potentials. The term X(x0) is a
constant term and can be dropped without affecting the results. Also, since x0 is a point of min-
imum, for stable equilibrium in a symmetrical potential, the odd terms must be zero. [Note that
if the expression resulting from the expansion of F(x) were used the even terms would be zero.]
Therefore,
dV (3.7a)
= 0, = 0
dx3

( >0 (3.7b)
while
\ dx2
Define
(x - x0) = x' (3.8)

= k (3.9)

= +e
6\ dx4 (3.10)
Then the potential function may be written as
V(x') = \kx'2
(3.11)
Sec. 3.2 Linear and Nonlinear Oscillations 61

Let us assume that the origin is located at the equilibrium point so that x0 = 0 and x' — x, and
?\ neglecting the higher-order terms in Eq. (3.11), we get

V(x) = \kx22 + \ex

\exA (3.12)

Furthermore, since the motion of the particle is in a conservative force field, using the definition

dV
F(x) = -
dx
jjid substituting for V(x) from Eq. (3.12), we may write

F(x) = — kx — ex3 (3.13)

Linear Oscillations

In the first approximation, we can neglect all terms except the first in Eqs. (3.12) and (3.13)
-o that

V(x) = \kx2 (3.14)

F(x) = -kx (3.15)

•- here

dF
k = 2 (3.16)
dx dx

>:nee (J2V/ck2)0is always positive, k will be positive also. Hence a force F(x) = —kx is always
r.rected toward the center and proportional to x. Such a force is called a linear restoring force.
The potential corresponding to such a force is parabolic as given by Eq. (3.14) and shown by
.-.e dotted curves in Figs. 3.1 and 3.2 for different values of k. The corresponding linear forces
lit shown by the dotted lines.
Physical systems involving springs, pendula, and elastic deformation are described by
Eqs. (3.14) and (3.15) and are said to obey Hooke's law. This is true only if the displacements
^-e small and we remain within elastic limits, as shown in Fig. 3.2(a). Moreover, the results ob-
-^ined are still approximate. We shall spend most of the time discussing linear oscillations re-
citing from approximate linear systems, k has been given several names, but is usually called
ir.e spring constant or stiffness constant, k is defined as the force per unit length with units of
-ewtons per meter (N/m). Ilk is called the compliance of the spring.

Nonlinear Oscillations

'.: the displacement of the system from stable equilibrium is not small (or if we definitely want
-^gher-order improvements in the linear approximation), we cannot drop the second term in
Eqs. (3.12) and (3.13). Thus, according to Eq. (3.13), the force is no longer linear because of
r e presence of a x3 term, while the potential is no longer parabolic because of the presence of
 62 Harmonic Oscillators Chap. 3

Hard F(x)
F(x), 0)'
N \

\
xQ Soft
(£<0)
Linear Elastic

Soft Hard
(e>0)

V(x)t
Actual Parabolic
\\ /
\\ /
\\
\\ j /

VJ
\\ J

(a) (c)

Figure 3.2 The plots of F(x) versus x and V(x) versus x for a variety of systems.
The magnitude and sign of e determines whether the system is hard or soft. For
soft systems, e < 0, while for hard systems, e > 0.

a x4 term. Different forms of forces and potentials are illustrated in Fig. 3.2 for systems with
large displacements (hence no longer linear).
Let us further consider Eq. (3.13) for a nonlinear system; that is,
F(x) = — kx — ex3 (3.13)
We must remember that e is a very small quantity as compared to k, but its magnitude and sign
affect the linear term - kx, hence the resulting force F(x). If e < 0, the magnitude of the force
F(x) will be less than the linear force kx alone and the system is said to be soft. On the other
hand, if e > 0, the magnitude of the force F(x) is greater than the linear force kx alone and the
system is said to be hard. The forces and potentials of such systems are shown in Fig. 3.2.

3.3 LINEAR HARMONIC OSCILLATOR

Consider the prototype of a linear or simple harmonic oscillator shown in Fig. 3.3. It consists of
a mass m tied to a spring having a force constant k. The spring-mass system oscillates in one di-
mension along the X-axis on a horizontal frictionless surface. The system obeys Hooke's law;
Sec. 3.3 Linear Harmonic Oscillators 63

k
I F = 0
'V/V/VX/V/N/X/V/V

x = 0

—W^AA/Wv
x>0

x<0

x = -A x= 0 x= + A
v= 0 Figure 3.3 Prototype of a linear har-
« = + "m a « = ~ «m monic oscillator showing the maximum
=° and minimum values of x, v, a and F.
F=0

r.ence the system is linear. Measuring the displacement x from the equilibrium position, the po-
.ential energy V(x) is
V(x) = \kx2 (3.17)
- hile the restoring force F(x) is
F(x) = -la (3.18)
2 2
From Newton's second law, F(x) = m(d x/dt ); therefore,
Sx
k (3.19)

= 0 (3.20)
dt2

ihere (3.21)
m
JJ is a constant and is called the/ree natural angular frequency (or free oscillation frequency)
rthe system.
Our aim is to solve Eq. (3.20) for x(t). Before we do this, it must be pointed out that an
equation of this form is frequently encountered both in physics and engineering; hence its so-
.ution must be thoroughly investigated. Equation (3.20) is a second-order, linear, homogeneous
64 Harmonic Oscillators Chap. 3

differential equation. The highest derivative that occurs in a differential equation is called its
order, while a differential equation is linear if it does not contain terms higher than the first de-
gree in the dependent variable [x is the dependent variable in Eq. (3.20)] and its derivative. Also,
Eq. (3.20) is homogeneous because it does not contain terms other than the dependent variable
and its derivatives. Thus the most general form of a differential equation of the nth order, linear
and inhomogeneous, is

(T'lx
C, = b(t) (3.22)
" df d^1 dt
If bit) = 0, the equation is homogeneous. The coefficients Cn, Cn __ b . . . , Ci are constants that
may or may not be independent of time, but we assume them to be independent of time.
We shall be dealing with second-order differential equations. We summarize next some
properties of such equations, which will be helpful.

1. The general solution of any second-order differential equation depends on only two
arbitrary constants. Suppose we choose Cj and C2 to be the arbitrary constants; then

x = x(t; C,, C2)

C\ and C2 are arbitrary because any values of Cx and C 2 will satisfy a second-order
differential equation.
2. If x{(t) is any solution of a linear homogeneous differential equation, then Cx{{t) is
also a solution, where C is an arbitrary constant.
3. If x,(f) and x2(t) are solutions of a linear homogeneous differential equation, then
X](f) + x2(t) or any other linear combination Clxl(t) + C2x2(t) is also a solution.

Let us now go back to Eq. (3.20) and try to find its solutions. To start, we may write it as

X + (O^yK = 0

Now, multiplying both sides by 2x,

2xx = — 2a»gxx

and integrating, we get

x2 = -a>2ox2+ C

where C is a constant. When x = A, x = 0; hence C = ai^A2. Thus

p = (J(A2 - x2) (3.23)

which, after separating the variables, may be written as

dx ,
dt
 Sec. 3.3 Linear Harmonic Oscillators 65

Integrating this equation, we get

,-1
sin - = wot +

:- here </> is a constant, called the initial phase ox phase constant. We may write this equation as

x = A sm(co0t + 4>) (3.24)

Thus the solution of Eq. (3.20), which is a second-order differential equation, is given by
Eq. (3.24) and contains two arbitrary constants A and <j> to be determined from the initial con-
iitions. Equation (3.24) is a solution of a linear oscillator or harmonic oscillator. The graph of
x versus t is shown in Fig. 3.4. x is called the displacement; the maximum displacement is called
•he amplitude of the oscillator and is equal to A. The quantity to0 is called the angular frequency
jjid is given by Eq. (3.21). Also, co0 - 2vv0, where v0 is called the frequency of the oscillator.
The time period To of the oscillator is the time required to complete one oscillation. Thus, in
::me t = To in Eq. (3.24), <Dot increases by 2TT, that is,

(o0T0 = 2ir

_ 277 . /m
= 2TTA - (3.25a)
to0 k

1
?r vn = — = (3.25b)
LTT

The expressions for velocity and acceleration may be obtained by differentiating Eq. (3.24);
:hat is,

v = x = a)0A cos((o0t + (3.26a)

a = x = —col sin(co0t +</») to20x (3.26b)

The plots of x, v, and a versus t are shown in Fig. 3.4.
The solution given by Eq. (3.24) may be written in a different form as follows:

x — A sin(<Ug? + <p)

= A sin (o0t cos (^ + A cos o>0f sin (f>

Substituting

A cos <f> = B (3.27a)

and A sin (b — C (3.27b)
we get

x = B sin co0t + C cos co0t (3.28)

66 Harmonic Oscillators Chap. 3

Figure 3.4

Below is the graph of the displacements N:=100 n : = 0..N n

x and xO versus time t for two oscillators tn
'~~K)
having the same frequency coO. The A:=.O5 coO :=l.f
phase difference between the two
functions is § = JI/2 or 90 degrees, as T:=— T = 4.002 v : = - v = 0.25
coO T
illustrated.
max(x) =0.05
x n :=A-sWooO-tn-j-<|>)

The maximum values of the vn :=coO-A-cosfcoO-tn-t-<|))

max(v) =0.078
displacements (amplitudes) of x and xO, 2
max(a) =0.123
velocity v, and acceleration a are as a n :=-A-o)0

shown. Also, the phase angles may be x0 :=Asin(coOt max(xO) =0.05

calculated as
Simple Harmonic Motion
0.05
x 0 0 - x0V20-0)0 =-1.57 -rad
0.025
- x00)-20-co0 =89.954'deg
x0 0
T\1
n
-X-
-0.025

Below is the graph of -0.05

- 1 0 1 2 3 4 5 6 7 8 9 10
displacement x, velocity v, and t
n
acceleration a versus time t. Time

Simple Harmonic Motion

(a) If we change the values of §
to 0, n/4, jt, 3n/2, and 2rc, how
do these graphs change?

(b) What are the phase

relations between x and v
and between x and a?
a,

(c) If the phase of x changes, 5

how will it affect the phase of v -0.2
and a?
n
Time
 Sec. 3.3 Linear Harmonic Oscillators 67

which is another form of the solution for a second-order differential equation. Squaring and
adding Eqs. (3.27a) and (3.27b), we get

A = Vfi 2 + C2 (3.29a)
and dividing Eq. (3.27b) by (3.27a), we get

<t> = tan" 1 ! - (3.29b)

Equations (3.27) and (3.29) give the relations between the constants A, (f>, and B, C. We may
differentiate Eqs. (3.28) to get expressions for velocity and acceleration.
Equation (3.20) or any other second-order differential equation with constant coefficients
may be solved by a trial solution of an exponential form, as explained next. Let the trial solu-
tion be

x = ex< (3.30)

where A is a constant to be determined. Substituting Eq. (3.30) into (3.20),

AV + coy = 0
A2 + col = 0 (3.31)
This is called the characteristic, initial, or auxiliary equation. Thus

~k
A = ±ito0 = ±i\l — (3.32)

where / = V — 1. Thus, with two roots A! = +iw 0 a n d A2 = —ift>0>tne general solution is, using
A+ and A_ as two constants,

x = A + e +i<"°' + A _ e """»' (3.33)

All three solutions, Eqs. (3.24), (3.28), and (3.33), are equivalent, and any one can be derived
from the other two. Each contains two constants. Equation (3.33) may be reduced to the other
two by using the Euler formulas

e±w = cos 6 ± i sin 0 (3.34)

It may be well to remember that in a solution given by Eq. (3.33), the constants A+ and
A _ are complex quantities and sometime inconvenient to use. Still another way of writing this
general solution is (with A and <f> as two constants)

x = A

= A cos(co0t iA si (3.35)
where both the real and the imaginary parts of this equation are solutions of a general differen-
tial equation.
68 Harmonic Oscillators Chap. 3

Energy of the Simple Harmonic Oscillator

For a simple harmonic oscillator, the displacement is

x = A sin(wof + (/>)

while the velocity is

x = cos(o)0f + 4>)
at
and the maximum value of the velocity v0 is

[7
A
=\ m
Hence the kinetic energy K of the oscillator is

K = \mx2 = ^m

= KQ cos2(co0t + $) (3.36)

where ^ 0 is the maximum kinetic energy given by

Ko = \ma>lAz = \kA2 (3.37)

The potential energy of the system is equal to the work done by the applied force Fa =
—F = -(-foe) = kx in displacing the system from x = 0 to x = x. Thus

V(x) = W = \ Fadx= \ kxdx = \kx (3.38)

Substituting for x:
V(x) = ~kA2 sin 2 (w 0 f + <f>)

V(x) = Vo sin2(o)0f + 4>) (3.39)

where Vo is the maximum potential energy when x = A; that is,

Vo = kkA2 (3.40)

Thus the total energy E, which is always constant whenever there is a conservative force held, is

E = K +V = l \kx2 (3.41)

This equation can be solved for x(t) and provides more information about the problem under
consideration, as we shall show next. From Eq. (3.41),
1/2
2E k ,
(3.42a)
x = :£| x'
m m
Sec. 3.3 Linear Harmonic Oscillators 69

dx
= \ - dt (3.42b)
or V(2E/k) - x2
we get the solution for x to be

x = A sin(wof + fa) (3.43a)

x = A cos(co0t + 4>2) (3.43b)

or
where fa and fa are constants, while the amplitude A is given by

2E
A = (3.44)
k
This relation tells us that x can vary between +A and ~A, that is, between

+ 2E andJ IE
VT ~VT
This has to be true because only then will x be real as given by Eqs. (3.42). The value of x then
must lie between two limits that are determined by the energy E and the spring constant k.
Tofindthe average values of V and K over one complete time period, we use the follow-
ing general expression for the average value of quantity j\t):

(3.45)

That is,
.T fT

Vdt V0sin2((o0t+ <$>)dt

_ Jo lo
dt

= \kA2 (3.46)

and similarly = \K0 = \kA2 (3.47)

That is, (V) ={K) = \ (3.48)

If, instead of time averages, we calculate space averages over one complete time period, we get
(see Problem 3.1)

<V>.._ = \kA2, < £ L _ = kkA2 (3.49)

and <£)Space = Wspace + <#)space = ( A i (3.50)

 70 Harmonic Oscillators Chap. 3

3.4 DAMPED HARMONIC OSCILLATOR

Theoretically, a linear or a simple harmonic oscillator once set into motion will continue oscil-
lating forever. Such oscillations are called free oscillations. In practice, however, in any physi-
cal situation there are dissipative or damping forces, and the oscillating system will lose energy
with time. Thus the oscillating system is damped and eventually comes to rest. The differential
equation for a linear oscillator given by Eq. (3.20) must be modified to include the effect of
damping.
Once again we consider a mass m tied to a spring, as shown in Fig. 3.5, as a prototype and
restrict its motion to one dimension. As the mass moves in a fluid, air or liquid, the frictional
force is the viscous force that produces the damping. As long as the speed of the mass is small
so as not to cause turbulence, the frictional force or damping force Fd may be assumed to be
proportional to the velocity. That is,
Frf = —bv= -bx (3.51)
where b must be a positive constant. The net force Fne, due to forces acting on mass m as shown
in Fig. 3.5 is
(3.52)
Using Newton's second law and substituting Fnet = nix in Eq. (3.52), we get
nix + bx + he = 0 (3.53)

which is a second-order differential equation for a damped harmonic oscillator. To solve this
equation, we divide both sides by m and substitute

7 = (3.54a)
2m

and (3.54b)
m
to obtain

x + 2yi + col = 0 (3.55)

As before, let us try an exponential solution of the form

x = ex\ x = Xex>, x =

-kx

hhhhhhhh
.-bx

Figure 3.5 Forces acting on a proto-

type of a damped harmonic oscillator.
Sec. 3.4 Damped Harmonic Oscillator 71

and substituting in Eq.(3.55), we get

ex\X2 o
Since eXt # 0, we must have

(A2 + 2yA + WQ) = 0 (3.56)

This auxiliary equation has the roots

A, = - y + V y 2 + co20 (3.57a)

and A 2 = — y — vy2 — col (3.57b)

Thus the general solution of Eq. (3.55) is, with A, and A2 as arbitrary constants,

x(t) = Axe^' + A2e^

or x(t) = (3.58)

The following three cases of this solution are of special interest and will be discussed in some
detail.

Case (a) Underdamped co20 >y2 X{ and A2 are

(oscillatory) imaginary roots
Case (b) Critically damped co20 = y2 A[ and A2 are real
(not oscillatory) and equal roots
Case (c) Overdamped col<y2 A[ and A2 are
(not oscillatory) real roots
>
Case (a) Underdamped Oscillations, col y2'- F ° r this case, it is convenient to make
a substitution:

(3.59)
m 4m2
Thus the exponentials inside the parentheses in Eq. (3.58) are imaginary, and we may write this
equation as

x(t) = (3.60)

which is a solution of an underdamped oscillator. Using the relation e ± 'e = cos 6 ± i sin 6, we
may write Eq. (3.60) as

x(t) = e'yt[i(Al - A2) sin coxt + (A, + A2) cos coxt\

Substituting i(Ax - A2) = B and (A, + A2) = C, we obtain an alternative solution:

x(t) = e'^lBsina)^ + Ccos^f] (3.61)

72 Harmonic Oscillators Chap. 3

This may still be written in a slightly different form by making the following substitutions in
Eq. (3.61).

A = VB2 + C2 and tan <j> =

Thus we obtain

(3.62)

Of the three solutions given by Eqs. (3.60), (3.61), and (3.62), we shall concentrate on
Eq. (3.62). It may be pointed out that the constants Ax and A2 in Eq. (3.60) are complex quanti-
ties, while B and C in Eq. (3.61) and A and <f> in Eq. (3.62) are all real quantities.
The solution given by Eq. (3.62) indicates that for a damped oscillator the motion is os-
cillatory, but the amplitude of the oscillations decays exponentially, as shown in Fig. 3.6. The
natural angular frequency, w^ or the frequency of the damped oscillator is always less than the
free oscillation frequency to0. The natural frequency o>, is not a frequency in the true sense of
the word because the oscillator never passes through the same point twice with the same veloc-
ity; that is, the motion is not periodic. But if y is very small, then o>, — a>0 (as shown later), and
we can call w, the "frequency." If y is small, we can expand Eq. (3.59) (using the binomial ex-
pansion) as
1/2
—
= (,2z -
(co = co0 1 -
0
tor

(3.63a)

= to0 (3.63b)
2(Or,

If 7 < o)0,

(3.63c)

According to Eq. (3.62) the case for 4> = 0 is shown in Fig. 3.6. Equation (3.62) states (and this
is demonstrated in Fig. 3.6) that the maximum amplitude of the oscillations decreases expo-
nentially with time because of the factor e ~ * and lies between the two curves given by

Ah) = ±Ae~ (3.64)

where Ae (t) is the envelope that limits the displacement of the oscillations. For comparison the
x0 graph represents the oscillations for a free oscillator, that is, for y = 0. The graphs with dif-
ferent dampings are shown in Fig. 3.6. Plots x4 and x5 represent the envelope of the damped mo-
tion (y ^ 0) that (due to the presence of the cosine term) touches the envelope at cos o)xt = ± 1 ,
that is, at times toxtn = nir or tn = nvlo)y, where n is an integer. The period of the damped os-
cillation is Tl = 2 7r/«!. Since o)i < a)0, that is, the damped frequency is smaller than the free
Sec. 3.4 Damped Harmonic Oscillator 73

) Figure 3.6

For undamped and underdamped oscillators, coO

N:=100 n:=0..N t :=- A:=10 coO:=1
11
Below is a graph of x versus t for 5

ihe following degrees of freedom:

70:=0 coOl := coOl =1
\0 Undamped, y=0
7I :=.O5 co) :=^coO - 7I ml =0.999
\ 1 Lightly damped, y not 0

72:=.2 •-72 ©2=0.98

\2 Moderately damped, y not 0

73 :=.7 co3 1=^0)0 -73 co3 =0.714

\3 Heavily damped, y not 0
n
xO : = A-cos(co01-t xl :=A-e -cosf(ol-t
\4 and x5 are two envelopes for
plot xl showing that the amplitude
7 tn
of the oscillations decays x 2 :=A-e -cos(co2-t) x3 :=A-e 7" 'n-cos(co3-t
n \ nj n \ n
exponentially.
-lit
x4 : = A-e x5 : = -A-e

i a) What equations (other than

:osine and sine) can be used to
obtain the same graph? xO

i b) How do the graphs of the

:our functions differ in their
:ime periods, amplitude, and
frequencies? Explain.

20

Underdamped Oscillators
74 Harmonic Oscillators Chap. 3

frequency, the period T, of the damped oscillation is longer than the free period To. This is clear
from the comparison of the co values of the plots in Fig. 3.6. It is worth noting that the points of
the curve that touch the envelope are TXI2 (= TT/O);) apart, but the maxima and the minima of the
curve, even though separated by TXI2, do not coincide with the points of maxima and minima of
the undamped motion curve.
Figure 3.6 shows the plots of x(t) versus t for different degrees of damping. The ratio y/w,
determines the essential feature of these plots. If y/tox < 1, the amplitude envelope Ae(t) given
by Eq. (3.64) changes very slowly with time, while the cosine term in x(t) makes several zero
crossings. Such a system is said to be lightly damped. On the other hand, if y/a)x > 1, the sys-
tem is said to be heavily damped because Ae(t) will decrease very rapidly and goes to zero, while
the cosine term makes only a few zero crossing. In either case, the ratio of the two successive
maxima is given by

(3.65)

where tx = tm is the time when the first maximum occurs and t2 = tm + 7\ is the time when the
next maximum occurs, Tx being the time period of the damped oscillation. The quantity exp
(yTx) is called the decrement of motion, while its logarithm, yTx, is called the logarithmic decre-
ment, 8; that is,

_b_ 2TT b IT
(3.66)
2m 10, m
Case (b) Critically Damped, co20 = y2: For this case, the two roots A; and A2 given by
Eqs. (3.57) are equal, that is,

Ai = A2 = - y

and the general solution given by Eq. (3.58) takes the form

x(t) = = Bxe

where (Ax + A2) = Bx = constant. This is not a general solution because it contains only one
constant. We can show that in such cases, if e ~ V is a solution,

x = tev (3.67)

is also a solution. Substituting in the differential equation

x + 2yi + &4x = 0

we get (col ~ y2)e~y' = 0

Sec. 3.4 Damped Harmonic Oscillator 75

Since o>0 = y, the equation is satisfied, and te v is also a solution. Thus, for a critically damped
case, the general solution is a linear combination of e ^ and te ~ v; that is,

x(t) = B2t)e (3.68)

where Bl and B2 are constants to be determined by the initial conditions.

Figure 3.7 represents three cases of interest (for a critically damped oscillator) resulting
from the solution given by Eq. (3.68). If we differentiate x with respect to t and equate to zero,
we get the positions of maxima in the plot of x versus t. Thus dxldt = 0 gives

2m
, (3.69)
B7 b

If

2m 2m
and
~b ~b
the curves for x versus t have maxima at t = 0 and t > 0, as in Fig. 3.7.

B± > 2m
If
b
the curve does not have a maximum for t > 0, as demonstrated in Fig. 3.7.
Critical damping plays a very important role in the design of such instruments as gal-
vanometers, hydraulic springs, and pointer reading meters. It is desired that the system attain an
equilibrium position rapidly and smoothly in the presence of frictional damping.

Case (c) Overdamped, col < J2'- If the damping increases such that y2 > afc, then the
iwo roots A, and A2 are real. If we represent

the general solution given by Eq. (3.58) takes the form

x(t) = e (3.70)

Note that o>2 is n o longer a frequency because the motion is no longer oscillatory. The exponents
are real, and both terms on the right decay exponentially, one faster than the other. As shown in
Fig. 3.8, for the case x(0) =£ 0, x(0) =£ 0, the displacement goes to zero asymptotically, but not
as rapidly as in the case of a critically damped system. For the case when i(0) > 0 or < 0,
Fig. 3.8, shows how x(t) varies with time. For x(0) > 0, x(t) reaches maximum for t > 0. For
x (0) < 0, but small, x(t) has no maximum for t > 0. For x(0) < 0, but sufficiently large, x(t) has
a maximum for t > 0, as shown.
76 Harmonic Oscillators Chap. 3

Figure 3.7

For a critically damped oscillator (not oscillatory) coO2 2

Plots are for the same values of B1 and B2 but

N:=50 n:=0..N _n
different values of y. For the present situation,
B2/B1 = 0.5, while7 = b/2m. The graphs for
three different values of -y and b are below. Bl :=6 B2:=3

72:=.34 73:=.7
(a) What is the effect of increasing the
value of y. Explain "Tit
xl„:= Bl + B2-t Ve

(b) What causes the changes in the values

-»2-t
ofBl and B2? Explain x2n:= B1 + B2-1 Ve

(c) How will a change in the value of b -,3-t

and m change the values of B and y? x3n;= B1 + B2-1 Ve

(d) What is the significance of the following

maximum values in the three cases?
Critically Damped Oscillator

min(xl) =6 max(xl)= 13.48 U 11-25

n

x2
min(x2) =0.016 max(x2) =6.406 7.5
x3

min(x3) =2.034-10,-6 max(x3) =6 3.75

0 5 10 15 20
t
n
Time

(e) How does the graph of a critically damped oscillator differ from that of an undamped
oscillator in terms of frequency of oscillations and amplitude?
Sec. 3.4 Damped Harmonic Oscillator 77

Figure 3.8
2
For an overdamped oscillator coO <y At t=0 x*0 -0
not oscillatory) n
N:=50I n : = 0..N C
n
10
Below is the graph of x versus t
for three different initial conditions. b: = 6 k :=4 M :=1
b
i a) How do different values of the 2-M
coefficients B, C, and D affect
different plots? coO: 0)0 = 2
M

ib) What causes the change in the

value of co and how does it affect
:he graph?
2
« , : = ( Y --0)0 f co=2 .236 (CO<Y)

Bl := 10 Cl :=1 Dl :=-3

• c) Explain why the two values of B2 :=.-8 C2 :=1 D2 :=5

\1 and x3 have opposite signs.
(O-t -lot \ -y-t
i n
Bl • e 12-e " •e

COt -COt
n n
x2 :=lCl-e +C2-e -e

cot n -co-t \ -ft

T^-^ n ' n
e -|-D2-e 1-e

Overdamped Oscillator

^iax(xl) =6.382 max(x3)=2

-nin(xl) =0.219 min(x3) =-1.683

n
Time
78 Harmonic Oscillators Chap. 3

y Example 3.1

Consider a damped harmonic oscillator and graph its motion using the following data and the
initial conditions, where m = mass, k = spring constant, and b = damping constani

m:=0.05 k: = 5.0 xO:=.Ol v0:=0 N : = 20 t:=0..N

(a) bi:=0.1 (b)b2:=l (c) b3:=5

Solution
Approach using Eq. (3.63):

(a) Underdamped oscillator with b = 0.1

bl :=0.1 7I :=- b — -yl = 1 coo := * co0 = 10

2m

col :=/\|(n0 - 7l Xll :=-71+A/712-COO2 X12 :=-7l - A / 7 1 2 - toO2

tol =9.95 M l = - 1 +9.95i A.12 = - 1 -9.951

A and (|) are determined

3
from initial conditions. See A:=0.1 <t>:=-0.1 xll t :=Ae -cos(ool---t-(|)
\ 3
the alternative approach
below.

(b) Critically damped oscillator with b = 1.0

b2 := 1 72 : = - b - 72 = 10 coO := — co0 = 10
2-m

2 2
co2 :=^/coO - 7 2 X21 :=- ~ 2 -coO 2 >.22:=-72-^72 2 -coO 2

co2=0 =-10 X22=-10

A:=0.1
3
x22,:=A-e -cos co2— +• 1
(c) Overdamped oscillator with b = 5.0

b3:=5 73:=-b- 73=50 coO := - co0 = 10

2-m ^m
Sec. 3.4 Damped Harmonic Oscillator 79

X32:=-y3-Jy32- coO2

X32 =-98.99
C03 =48.99i

x33t := i t
A:=0.1 -cos co3—

o.i#

xl 1

x33,

-0.1
5 10 15 20
t

Under, critically, over damped

col =9.95
co2=0 co3=48.99i
max(xll) =0.1 max(x22) =0.1 max(x33) =0.1 +0.004i

-5
min(xll) =-0.071 min(x22) =0 rain(x33) =5.915-10

Alternate approach
Using Eqs. (3.60), (3.61), and (3.62):
i a) Underdamped oscillator with y = 1.0

Guess A'-.l-m §'= • l r a d

Given .l=A-cos(())) 0=-A-cos((()) - 10-A-sin(<|>) ())<lrad

Evaluate the constants
?v using the initial /ol
Find A
conditions and Eq. (3.61). ( '*> = L A A:=.l A=0.1 ' =-0.1

3
xl t :=A-e -cos col —
80 Harmonic Oscillators Chap. 3

(b) Critically damped oscillator with 7 = 10

Guess Al :=.l A2:=.l

Given Al + A2= -Al-99-A2=0

/ 0.101 \
Find(Al,A2) =
\-0.001j Al :=0.1( A2:=l
\ Y2
3
-- -e

(c) Overdamped Oscillator with 7 = 50

Guess Bl :=. B2:=.l

Given Bl=0.10 B2 - 10Bl='0

'0.1
Find(Bl,B2) = ( ) B,:=», B2 : = -0.001 X.31-1
\ / x3t :=B1•e

0.1 ¥•

xll

-0.05 -0.05

It is clear from these two graphs that both approaches give the same results. The third
treatment mentioned in Section 3.4 also will yield the same results.
EXERCISE 3.1 Discuss the motion of a harmonic oscillator with the same initial condition
as in the example, except that (a) the mass is changed to 1.0 kg while keeping the k the same
and (b) the k is changed to 10 N/m while keeping the mass the same.
Sec. 3.4 Damped Harmonic Oscillator 81

Energy Considerations

The total energy E(t) of a damped harmonic system at any time t is given by
£(0 = £(0) + Wf (3.71)
where £(0) is the total energy at time t = 0 and Wf is the work done by friction in the time in-
terval 0 to t. Assuming the dissipative frictional force/ = - bx = - bv, we can calculate Wf
as follows:

(3.72)

Thus the rate of energy loss by friction may be written as

dW,
t\ =- -bv1 (3.72)
dt \ dt
which is negative and represents the rate at which energy is being dissipated into heat. Since Wf
< 0, E, continuously decreases with time and may be calculated in the following manner:
£(?) = K(t) + U(t) = \mx2 + \kx2 (3.73)
From Eq. (3.62),

y
x(t) = - sin(<u,?

Let us assume that the system is lightly damped so that ylo)x < 1, and neglecting the second
rerm on the right in the preceding expression for x, we can substitute for x and x in Eq. (3.73):
£(0 = \Ale-1
Since we assumed light damping, we may write <Wj — a>l = k/m; hence this equation takes the form
] 2
E(t) = 2kA e~2yt (3.74)
while the initial energy of the system is obtained by substituting / = 0 in Eq. (3.74); that is,
F = -kA2 n 75*1
Thus
£(0 = Eoe~2yt (3.76)
lyt
That is, the energy decreases (or decays) exponentially at a much faster rate (e ~ ) than the rate
at which the amplitude decreases or decays (e ~ yt).
The time T in which £ decreases to lie ( = 0.368) of its initial value is called the char-
acteristic time or decay constant and may be evaluated by substituting E(t) = E<Je and t = rin
Eq. (3.76):
 82 Harmonic Oscillators Chap. 3

or 2yr =

That is,
1 2m m (3.77)
7
~ 2y
2 ~ 2b b
If y is very small, T —> °°, and if y is very large, T —»0.
Also, using Eq. (3.76), we may write the logarithmic derivative of £ as

d idE
(3.78)

(l/E)(dE/dt) represents the fractional rate of decrease in energy. Since the rate of energy loss is
proportional to the square of the velocity [Eq. (3.72b)], the loss in energy is not uniform. dEldt
will be maximum when x is maximum (near the equilibrium), and it will drop to zero when x
is zero near maximum amplitude. The plots of E and dEldt are shown in Fig. 3.9.

3.5 QUALITY FACTOR

The quality factor Q, or simply Q value, is a frequently used term in mechanical oscillatory sys-
tems, as well as electrical oscillatory systems. Q is a dimensionless quantity and represents the
degree of damping of an oscillator. The quality factor is defined as 2ir times the ratio of the en-
ergy stored to the average energy loss per period. Thus

energy stored in the oscillator

= 2TT (3.79)
average energy dissipated in one time period
If P is defined as the power loss or the rate at which the energy is dissipated, and the time pe-
riod of oscillation Tx = ITTIM^ we can write the denominator as PT{ = P27r/a>1, and hence
Eq. (3.79) as

(3.80)
P2TT/CO, P/(o

But is the time of motion for 1 radian. Thus

energy stored in the oscillator

Q = (3.81)
average energy dissipated per radian
As should be clear, for the lightly damped oscillator, Q will be very large, while for a heavily
damped oscillator, it will be very small. We can calculate the Q value of the lightly damped os-
cillator as follows.
The energy of the oscillator and the rate at which it loses energy are given by [Eqs. (3.76)
and (3.78)]

E(t) = (3.76)
Sec. 3.5 Quality Factor 83

Figure 3.9
—
Below are the graphs of energy N:=ioo n:=0..N t
n
E and the rate of energy loss
A:=l k:=20 b:=.5
ET = dE/dt versus time t for a
damped oscillator ^Q
b
For clarity, instead of y: = col := O - y
2-M
graphing ET, we have graphed
ET-5
y - 0.25 coO = 4.472 col = 4.465

(a) Why is E positive and l

A 2 "2"Tt.
ET negative?
*
E_ :=—-A e "['M-col -(sin^col-t + <t>)) t- k-cosfcol\ + ty
z
n

(b) Is the rate of loss of ET : = -b- col •A2)-e n

J- sin(a>l-tn + (()) -t-—• cos(ool-t
energy ET in phase with [ ' col
or out of phase with
energy E? What is the
significance of this?

\ ,

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

t_

Energy and rate of energy loss

 Harmonic Oscillators Chap. 3
84

(3.78)
=
and Yt - ^
Thus the energy dissipated in time Af will be

\dE (3.82)
= 2yE Ar
dt
If Af is the time for 1 radian of oscillation, Af = \lo)x; hence

O = — = E
= ^L (3.83)
A£ 2y£/ft)! 2y
For light damping, cox — co0; hence

(3.84)

If y is small, Q will be large, and vice versa. Ordinary mechanical systems, such as loudspeak-
ers and rubber bands, are heavily damped and may have Q values from 5 to 100. On the other
hand, systems such as tuning forks and violin strings may have a Q value as high as 1000. A
typical microwave cavity resonator has a Q value of about 104. Systems with extremely light
damping are excited atoms (Q = 107), excited nuclei (Q — 1012), and gas lasers (Q =* 1014).

y Example 3.2
Consider a critically damped oscillator of mass m, damping coefficient b, and initial
displacement A. Calculate the rate of energy dissipation and the total energy dissipated
during the time interval t = 0 and t = m/b.

Solution
According to Eq. (3.68), the solutions of the x=(Bl-t-B2-t)-e"Tt
critically damped oscillator for x and v are
v=(B2- (Bl + B2-t)-Y)-exp(-yt)
(where 7 = b/2m)
t=0 x=A v=0
Using the initial conditions at t = 0, x = A,
and v = 0, the values of the two constants B1=A B2=yA
B1 and B2 are as shown

The resulting equations for x and v are x=A 1 + t -e

2-m
(using Y = b/2m)

_dx_ b 2 -A jt-e "\2-m)' t

dt 4-m 2
 Sec. 3.6 Forced Harmonic Oscillator (Driven Oscillator) 85

Substituting the values of v above

into Eq. (3.72), the time rate of change
2 b5
of energy dissipation through damping is ——s v=
16-m4

2.2
4-7 -t -exp(-2-yt)-h4-yt-exp(-2-yt) + 2
Integration with respect to t Wi=—•
gives the energy dissipated in time t
t :=i
Using the values given in
Example 3.1,we calculate the 2
-ta -exp(-2-Yt) + 4- y t-exp(-2-yt) + 2-exp(-2-yt)J
energy dissipated between Wi :=—-
8
t = 0 and t = m/b = l/2y.
Wi
= -0.115
Wi =-1.839 16

, newtonsec __ b
= 1•sec
b :=.l m:=.05-kg y: = 2-m
—

A:=.01-m A=5-104-kg t :=— t=0.5-sec

2-Y

5
P = rate of energy dissipation
> v -£r 6 9
-kg 3 'sec 3
16-m4/ • P = 2.299* 10

WI : = - 8
.025 WI =9.197-10 -kg3-sec 3

EXERCISE 3.2 Make plots of W and P versus t for the situation given in Example 3.2.

15 FORCED HARMONIC OSCILLATOR (DRIVEN OSCILLATOR)

-. free oscillator will oscillate forever. But, in reality, every system has some damping present
i-.e energy is dissipated, say in the form of heat) and the system will eventually stop oscillat-
- z. To maintain the oscillations, energy from an external source must be supplied at a rate equal
:he energy dissipated by the oscillator in the damping medium. Such motion in which energy
>upplied externally is called forced oscillations or driven oscillations, while the system is
. ^ied a. forced oscillator or a driven oscillator. If the system is acted on by a driving force Fd
-.-•in the net force, F net , acting on the system is given by

(3.85)
86 Harmonic Oscillators Chap. 3

where F, = -kx, Ff= -bx

and from Newton's second law F net = nix. Equation (3.85) cannot be solved unless we know
the form of the applied force Fd. Since we have been limiting our discussion to linear oscilla-
tors, it is easier if we assume that the driving force has a sinusoidal form given by

Fd = Fo 0) (3.86)

We have good reasons to assume this form for the driving force. First, many actual situations
involve just such a force, as, for example, the response of a bound electron when electromag-
netic waves are incident on it, that is, in the scattering of light from bound electrons. Second,
any periodic function of time can be represented as a sum of several harmonics (or sinusoidal)
terms. Using the techniques of Fourier series, one can solve for the motion of the system under
any periodic driving force (as discussed in Chapter 4.).
We may combine these equations and write the following equation that describes the mo-
tion of a driven harmonic oscillator:

mx + bx + kx = Fo cos(a>f + 80) (3.87)

This is an inhomogeneous, second-order, linear differential equation. The solution of Eq. (3.87)
is given by the sum of two parts according to the following theorem:

Ifx,{t) is a particular solution of an inhomogeneous differential equation and the comple-

mentary function xh(t) is the solution of the corresponding homogeneous equation [that is,
Eq. (3.87) with the right side equal to zero], then x(t) = xt(t) + xh(t) is also a solution of |
the inhomogeneous differential equation.

Thus the general solution of Eq. (3.87) is

x(t) = xft) + xh(t) (3.88)

where xh is the solution of the homogeneous equation

mx + bx + kx = 0
From Section 3.4, the general solution of this homogeneous equation is given by any one of the j
three forms, Eq. (3.60), (3.61), or (3.62); hence

xh(t) = (3.89a)

xh(t) = sin Ccos a>xi) (3.89b)

xh(t) = (3.89c)|

Since the oscillations of a damped oscillator eventually decay to zero, the xh part of the solu-l
tion is called the transient term. After a certain time, the xh part of the solution is of no conse-l
quence; hence, for a steady-state solution we must concentrate on finding the particular solutionl
 Sec. 3.6 Forced Harmonic Oscillator (Driven Oscillator) 87

According to Eq. (3.87), the applied force varies sinusoidally, so we expect the resulting
steady-state solution x,{t) to vary sinusoidally. A solution of the form x = A cos a>t would have
^een perfectly acceptable if the left side of the equation did not have an x term. To take care of
JUS situation, we must have a solution of the form

x = A cos(cot± 4>) (3.90)

let us assume a solution of the form

xt = A cos(cot - 4>) (3.91)

To calculate A and <f>, we substitute for xt in Eq. (3.87), and after setting 0O = 0, we get

-ma>2A cos(cot - 4>) - bcoA sin((ot - cp) + kA cos(a>t - <f>) = Fo cos cot

Rearranging,

(kA cos (/> — mco2A cos <f> + bcoA sin <f>) cos cot

- (kA sin cf> - ma>2Asm cj> - bwA cos <f>) sin cot = Fo cos cot

- :r this to hold for all values of t, the coefficients of the cos cot and sin cot terms on each side
-list be separately equal. That is,

(k — raw2) cos </> + bco sin <p = —- (3.92)

A
(k — mco2) sin <f> — bco cos </> = 0 (3.93)
z
rom Eq. (3.93), we obtain an expression for the phase angle to be

bco bcolm
tan tp = 2 (3.94)
k — mco klm — co2
_ >ing the usual notation klm = col a n d 7 — b/2m, we get

2 Tco
t a n cf> = (3.95)
con — co
Tom which we obtain

2yco
sin tf) = (3.96)
2
V(co - co2)2

cos cp = (3.97)

s we substitute these in Eq. (3.92), we get

FJm
A = (3.98)
- co2)2 + Ay2, 2co
,.2
 Harmonic Oscillators Chap. 3
88

Thus a particular solution of the inhomogeneous equation is

FJm (3.99)
xt(t) = -j===^===== cos(wr -
V{oi - OJ2)2 + 4 y 2 « 2
(3.100)
where
Q

[A slightly different procedure for obtaining the general solution is more convenient when
the driving force Fd given by Eq. (3.86) is written in exponential form as
Fd = Foe'(w'+('») (3.101)

and we can obtain the same results.]

Using the solution given by Eq. (3.99) together with the homogeneous solution xh, given
by Eq. (3.89c), we get the general solution

X = Xh + X,

(3.102)
<f>h)
As required, this solution contains two arbitrary constants (of integration) Ah and 4>h> while <j> i
not a constant and is given by Eq. (3.100). The first part of the solution oscillates with a natural
frequency cox. Because of the damping, the oscillations die out for large values of time, that is,
for t > 1/y. The homogeneous solution xh is called the transient solution, while the particular
solution Xi is the steady-state solution. The general solution x will be independent of the influ-
ence of the initial conditions except in the beginning when the transient term is still contribut-
ing. Figure 3.10 illustrates this for two special cases: (a) for u> < co{, that is, the driving fre-
quency is less than the natural frequency; (b) for u> > o>,, that is, the driving frequency is greater
than the natural frequency. For both cases, the plots of the homogeneous solution xh versus t as
well as plots of the particular solution x, versus t are shown. The resultant of these two, that is,
the plots of xn = xh + xt versus t are also shown. As is clear from these plots, the transient solu-
tion xh is effective only in the beginning and decays to zero as time passes, while the steady-
state solution remains constant with time. Thus the transient solution effects the general solu-
tion only in the beginning. Furthermore, if w < a;., the transient term xh causes distortion of the
resulting sinusoidal waveform as shown in Fig. 3.10(a). On the other hand, if co > o>,, the tran-
sient term xh, instead of causing distortion, has the effect of modulating the oscillations due to
the force function as shown in Fig. 3.10(b). Of course, in both cases, after the transient term has
died out, the oscillations are governed by the force function. In addition to the relative values of
a; and (ou initial conditions will also affect the detailed motion, but only in the beginning. It is
important to note that the transient terms play an important role in electrical circuits. In design-
ing such circuits, it is necessary to avoid peak voltages and currents when initially the circuits
are closed.
Since for t > 1/y, x — JC,-, we shall concentrate on the discussion of the steady-state solu-
tion, that is, the particular solution xt given by Eqs. (3.99) and (3.100). This solution is inde-
pendent of the initial conditions. !
Sec. 3.6 Forced Harmonic Oscillator (Driven Oscillator) 89

Figure 3.10

Below are two special cases of the

influence of the transient term on the N:=30 n : = 0..N t :=-
steady-state solution.
A:=l y:=l <t>:=0 :=3
(a) When the driving frequency co
(=10) is greater than the M w:=10 F0 :=50
natural frequency col (2.828), this
leads to distortion, as illustrated. 2yo)
: = atan
coO - co 0)1 := 0)0 - y

)>i =-0.216 col =2.828

FO
n M
xh : = A-e -cos(col-t •cosfco-tn-)-(()ij

2 2 2 2l
coO - co j -h 47-co J

Distortion

Timet

• v*
90 Harmonic Oscillators Chap. 3

(b) When the driving frequency co (=3) N:=30 n :=0..N t := —

is less than natural frequency n
10
col (9.95), this leads to modulation. A:=l y:=l
coO:=lO
as illustrated

M:=l co:=3 F0:=50

l
2
2yco 2 2
:=atan col ::==(U
! <c o O - y
ooO - co

col =9.95
|>i = 0.066
FO
-yt M
xh n :=A-e "-cos(col-t n -|-0 xi :=-
2 2\z 2 2
X ;=xh Xi
coO - co j -i- 4-y -co
n n+ n
Modulation

xh
\

n
0

-1
V
0 0.5 1 1.5 2.5
t
Time t

If the driving frequency is equal to the natural frequency, what effect it will have on the
amplitude?

Example 3.3
I
Consider a damped oscillator, for which y = <U(/4, acted on by a driving force F = Fo cos cot. Find the
general solution x(t).
Solution
The second-order differential equation describing the driven oscillator is

mx + bx + kx = FQ COS cot
 Sec. 3.6 Forced Harmonic Oscillator (Driven Oscillator) 91

or x + 2 y i + cohc = — cos a>? (i)

where y = wg/4 and

to2 = (to2 - y 2 ) m = 0.97 to0 (ii)

The transient (or the homogeneous) solution

xh(t) = e^yl{Al cos to,? + A2 sin to,?) (Mi)

lakes the form

xh(t) = e~^'\Ax cos 0.97? + A2 sin 0.97?) (iv)

Let us assume the particular solution for the applied force F = FQ cos cot to be

xft) = B{ cos &>? + B2 sin &>/ (v)

x;(?) = — ooB{ sin <y? + coB2 cos to? (vi)
JT,(?) = — co By cos cot — co2B2 sin w? (vii)
Substituting these three equations in Eq. (i) and rearranging gives

p
( - w 2 B ! + 2ywB 2 + a cos <ot + ( - w 2 5 2 - 2ya)B1 + w^B2) sin <a = — cos (viii)

Equating the coefficients gives

p
((o0 — (o )By + 2ycoB2 — — (ix)
m
-2ywBj + (a>l — co2)B2 = 0

Solving these equations for B, and B 2 in terms of Fo gives

B, = — • (xi)
- w 2 ) 2 + 4y2io2]

m[(a>l — &i2)2 + 4y 2 w 2 ]
(xii)

Hence the general solution is given by

x(t) = xh{t) + x,(?)

, cos 0.97? + A2 sin 0.97?) + 5 , cos w? + B2 sin to? (xiii)

where fi, and B2 are given by Eqs. (xi) and (xii), while At and A2 are to be evaluated using the initial con-
ditions in Eq. (xiii) by the usual procedure (see Example 3.1).

EXERCISE 3.3 Complete the example for the driving force F = Fo sin cot with initial conditions t = 0,
.r(0) = 0, and i(0) = 0, that is, calculated, and A2. Also graph xh, xt, and xn = xh + x-t.
 92 Harmonic Oscillators Chap. 3

3.7 AMPLITUDE RESONANCE

The amplitude A and the phase angle <f> of steady-state motion according to Eqs. (3.98) and
(3.100) are

FJm
A = (3.103)
- colY + 4y2,lto.2
2yco
<f) = t a n (3.104)
O>Q — CO2

For a fixed value of a»0, the variations in A and cp with the driving frequency co for different val-
ues of y are shown in Fig. 3.11. As illustrated, the behavior of these quantities strongly depends
on the ratio co/to0.
As stated earlier, cp represents the phase difference between the driving force F and the re-
sulting motion x\ that is, it represents a delay between the action and the response. As shown in
Fig. 3.11, this phase lag, which is </> = 0 when co = 0, increases to <f> = TT/2 for co = a>0 and
reaches <$> = tr as co —> °° ; that is, at very high frequencies the oscillations of the system are
180° out of phase with the driving force. It is interesting to note that as y —> 0 the phase change
occurs more and more rapidly, and in the extreme case when y = 0, the phase changes suddenly
from 0 to 7T at o> = co0.
From Fig. 3.11, it is clear that, depending on the values of y, there is a certain driving fre-
quency at which the amplitude A has a maximum value. The frequency at which the amplitude
is maximum is called the amplitude resonance frequency cor. This frequency cor may be calcu-
lated from Eq. (3.103) by setting

= 0 (3.105)
dco
Upon solving the resulting equation, we get

co = cor = (co20 - 2y2)m (3.106)

which states that as the damping coefficient decreases the resonance frequency increases, and
in the limit as y —> 0, cor —> coQ, decreases the natural frequency of a free oscillator. In the case
of extremely small damping, we can express the right side of Eq. (3.106) by using the binomial
theorem; that is,

cor = to0
2 co2

or (3.107)
to0

Equations (3.106) and (107) for a driven oscillator may be compared with the case of a damped
oscillator discussed previously; that is,

co, y22\) 1/2

Sec. 3.7 Amplitude Resonance 93

) Figure 3.11
N:=40 n : = 0..N m:=0..4
ia) The graph below shows
Amplitude resonance in the variation of
Y •=
amplitude A versus frequency ratio co/coO coO : = 2 co :=—+• .01 m
n
10
for different values of y. 0 0
1 ~2

F0: = 6 M := 5 2 1
FO 3 1
M 4 T

Amplitude Resonance

0 0.333 0.667 1 1.333 1.667

0)0

(b) The graph below shows the

N:=39 n:=l..N i:=41..8O m:=1..3
phase angle variation
in the phase angle § versus frequency
n m
ratio oVcoO. r\ A i
coO :=4 co_ : = 10
— co. : = 10
—

7I :=1 72 :=2 73 :=3

94 Harmonic Oscillators Chap. 3

2 - 7 -CO •y -co.
'm i
: = atan : = atan -h 71
coO - (co coO - (co.

Phase Angle

Oi.3

0.5 I
What is the effect of using
<0n C0 n (0n CO; CO; CO;
values of y greater than 1 in (a)
0)0' raO' coO' 0)0' coO' coO
and less than 1 in (b). Frequency ratio

and, for small y,

O), =- CO 0 (3.108)

while for a free oscillator a>l = klm. Thus w, lies to the right of wr, while co0 is still farther away
from <on as shown in Fig. 3.12 where A2lA\ = 1 at co = cor.
Thus the maximum amplitude A = Ao that occurs at co = (or may be obtained from
Eqs. (3.103) and (3.106) to be

A= (3.109)
2yVa>l-y2
In the case of small damping, we assume that y —> 0; hence

Ao - - - - (3.11U)
2myu>0 bco0
It is clear that if b is small or y —> 0, the amplitude Ao becomes very large. For undamped sys-
tems, b = 0 and hence Ao = °° ; but there are hardly any systems that are undamped.
Sec. 3.7 Amplitude Resonance 95

Figure 3.12

The graph below shows the relative positions of the resonance frequency cor, the natural
frequency col, and the natural free frequency coO. Q(co) are the positions of the resonance
amplitude for frequencies cor, coO, and col. The graphs are the ratio of the square of the
amplitudes versus con - cor.
n
N:=50 n:=0..N co :=— F0:= 2 coO:=7 y:=l M:=l
n
5
FO
M
A :=-
:=coO- 1-2- c o l l = coO -
coO 2 , \2f 2 , \2
coO - c o -h 4 - y - c o
coll =6.928
jrl =6.856
FO
2
col :=coO- M
2coO
coO

coO - (co) J -h 4-y -(co)

Q(co) :=
_x =6.857 col =6.929
FO
FO i:=0..2
\0 ; = - 2-M-Y-COQ
AO =0.143
2-M-ycoO
x. : =
X.
> i are different values of Q for different i

frequencies. x[ and yj are plotted against cor Q(cor)

6.857 1.021
\j - cor. Note the small differences in the col Q(col)
6.929 1.015
positions of the resonance frequencies. coO Q(coO)
7 1

Positions of Resonance Frequencies

\
0.5

-a
-2 -1

O)n-0)r,x.-(Or

How do you explain the gradual change in Q for different values of co?
 96 Harmonic Oscillators Chap. 3

The fact that the amplitude of the oscillations of the system is very large at the resonance
frequency has both desirable and undesirable effects. In the case of electrical circuits that are
used in tuning radios and in certain types of musical instruments such as organ pipes, it is de-
sirable to have a large response for a small driving force. On the other hand, it is very undesir-
able to have a very large amplitude in mechanical systems, such as in the springs of an auto-
mobile or in the spring mounting of an electric motor. There the aim is to minimize the
transmission of vibrations.

3.8 ENERGY RESONANCE

In most practical situations involving oscillating systems in nature, the quantity observed ex-
perimentally is energy and not amplitude. Also, the total energy of an oscillating system is pro-
portional to the square of the amplitude near resonance; hence we should look for the variation
of A2 versus co. Once again we assume that o>0 is fixed. For steady-state motion, the amplitude
A is constant, and we may write

x = A cos(wt + </>) (3.111a)

and v = x = — coA sin(wf + (f>) (3.111b)

which gives us

K(t) = \mv2 = -2 sin2(wf (3.112)

U(t) = \kx2 = \kA2 cos2(cot (3.113)

E{t) = K(t) + U(t)

= \A2[m(x)2 sin2(wf + < / (3.114)

Let us now calculate time averages of K(t), U(t), and E(t) in the case when A changes with &>,
and remembering that {cos2(oot + 4>)) — (sin2(a>£ + </>)) = \ for an average over one period.
Substituting for A from Eq. (3.103) into Eq. (3.112), we get

2
FlJm2 (3.115)
K(t) = 2mco
o20 - co2)2 + 4 y V
That is,

(a"
(3.116)
N 2 2 2
"' 4 m (OJ 0 - co ) + 4 y V

The value of OJ for which (K) is maximum is obtained by setting

d{K)
= 0 (3.117)
du>
 Sec. 3.8 Energy Resonance 97

which gives

OJ, = &>n (3.118)

That is, the kinetic energy resonance occurs at the natural free frequency co0.
Since U(t) = \kx2, the potential energy resonance must occur at the same position as the
amplitude resonance; that is, the potential energy resonance frequency cou from Eq. (3.106) is
,,2\ 1/2
(3.119)
I: should not be alarming to find that the kinetic and potential energies resonate at different fre-
quencies. This is because we are dealing with a damped oscillator, which is a nonconservative
>> stem. The energy is continuously being drained by the damping medium.
The average total energy (E) from Eq. (3.114) is

= \A2mco2{sm2{cot {kA2(cos2(tot

= \A2(mco2 + k) = \mA\co1 + (3.120)

Substituting for A from Eq. (3.103), we obtain

1 Fl O)2 +
(3.121)
4 m (co2 - col)2 +
For very weak damping, y < co0, and we may write

co22 _i_ /2 „_

- co0)

Vsing this in Eq. (3.121) and noting that £ is a function of <u, we replace E(t) by E(a>) and
simplify

1
(3.122)
m (co — io0) + y
Rewrite this as

1
(3.123)

:- here the function L(co) contains all the necessary frequency dependence of (E(co)). A plot of
function L(co) is called a resonance curve or Lorentzian. Figure 3.13 shows several such plots
: ?r different values of y. Note that for large y the function is effectively zero except near the res-
: nance frequency co0.
The maximum height of the resonance curve occurs at co0 and is equal to 1/y2. This value
-ill fall to one-half its maximum value when [fromEq. (3.122) or (3.123)]

(co - co0)2 = y2 (3.124)

98 Harmonic Oscillators Chap. 3

Figure 3.13
Below is the graph of the resonance curve n : = 0..200 co_ :=n coO :=50
(or Lorentzian), that is, the graph of L(co)
versus co for different values of 7.
73 :=3
(a) What causes the change in the values
ofL? LI :=-

(b) What determines the width of the

energy resonance curve? L2 :=-
-coO
(c) What does the following analysis
indicate about the location of the
L3 :=-
resonance frequency on the ©-axis? (an - coo')

max(Ll) = L1
50
=

LI
n
0.75
\ I
1\
L2n 0.5
n ,
max(L3) =0.111 L3 50 =0.111
I L3
n
L1 45 = 0.038 Ll 5 5 =0.038

— 0.25
L245 = 0.034 L255 = 0.034
40 45 50 55 60
n

L3 45 = 0.029 L3 55 = 0.029 Frequency

or co = ±y (3.125)

This equation states that the resonance curve drops to half its maximum value at w+ = io0 + y
on the higher frequency side of to0 and at w_ = w0 - y on the lower frequency side of co0. Thus
the full width of the curve at half-maximum, called the resonance width, is given by to+ + co^ =
2y. We call Aw the resonance width; hence

Aw = 2y (3.126)
 Sec. 3.9 Rate of Energy Dissipation 99

As y decreases, the width A« of the curve also decreases, which means that the resonance curve
becomes higher and narrower. This implies that the range of frequencies to which a system will
respond becomes narrower; that is, an oscillating system becomes increasingly selective in
frequency as is obvious in Fig. 3.13.
The frequency-selective property of an oscillating system is characterized by the quality
factor Q. In Section 3.5, we defined the quality factor as the ratio of the total energy stored in
the oscillator to the energy dissipated per radian of oscillation. For a lightly damped oscillator,
we showed [Eq. (3.84)] that

(3.84)
2y
When such an oscillator is driven, we get a resonance width Aa> = 2y, and we may write

ton resonance frequency

(3.127)
Aw frequency width of resonance curve
We can prove that Q = wjly by directly applying the energy definition of Q to a driven oscillator.
It becomes clear that systems with a high value of Q have a very narrow resonance width
and hence are very highly selective to a frequency response when external driving forces are ap-
plied. If the resonance curve is sharp, a system will respond only when the driving frequency is
equal to the resonance frequency. As mentioned previously in Section 3.5, Q may vary from 5
to 100 for mechanical systems, to — 108 for atomic systems, and to — 1014 in the case of gas
lasers. Such facts have been utilized for defining and making time standards using atomic
clocks.

3 9. RATE OF ENERGY DISSIPATION

Finally, let us calculate the rate at which energy is being dissipated, which should be equal to
:he rate at which work is being done. Starting with the general equation for a forced oscillator,
with 0O = 0,

mx + bx + kx = Fo cos a>t (3.87)

and multiplying both sides by x,

m'xx + bx2 + kxx = FQX COS a>t

We may write this as

a I mx
II + + bx2 = (Fo cos wt)x (3.128a)

(K + U) + bx2 = (Fo cos ot)x (3.128b)

dt
100 Harmonic Oscillators Chap. 3

That is,

Time rate of \ / Rate at which Rate at which energy

=
change of A' energy is being is being supplied by
andC/ I dissipated the driving force

By substituting for x and x from any particular solution [Eq. (3.90)] we can prove that the av-
erage power (P) at which the driving force does work is

dW\
(P) = {^-J = (xF0 cos cot)

or (P) = \XQFQ sin <f> (3.129)

where xm is the maximum velocity, which occurs at 10 = coQ. Furthermore, we can show that
(after transients have died out)

(P) = (3.130)

as it should be.
Let us consider the right side of Eq. (3.128b), which, after substituting for x, may be writ-
ten as

x0 Fo cos cot = — FocoA cos cot sin(cot — <f>)

= F0coA(cos2 cot sin <f> — cos cot sin cot cos (3.131)

The first term on the right is positive, which means that the driving force is sypplying energy, j
while the second term is negative, which means that the driving mechanism is receiving energy, I
that is, the driving force is alternately supplying and absorbing energy. This may appear strange,
but that is what actually happens. But it may be pointed out that the average value of cos2 cot is
^, while that of (sin cot cos cot) is zero. Hence, on the whole the driving agent supplies more en-
ergy than it absorbs.

PROBLEMS
3.1. Prove the following results for a simple harmonic oscillator; that is, prove Eqs. (3.49) and (3.50): j

N 'space 6^*^ ' * 'space 3

<£>space = Wspace + Wspace = (E\ime

3.2. A liquid in a U tube is in equilibrium. When the liquid is slightly displaced, it executes simple har- |
monic motion. Calculate the frequency of such oscillations.
Problems 101

Equilibrium
position

w = A(2z)pg

Figure P3.2
3.3. A small mass m is in a box of mass M that is tied to a vertical spring of stiffness constant k. When
displaced from its equilibrium position y0 to yi and let go, it executes simple harmonic motion. Cal-
culate the reaction between m and M as a function of time. Does the mass m always stay in con-
tact with the box? If not, what determines that it will not be in contact with Ml Calculate the value
of h as measured from the equilibrium position for which the contact is lost.
3.4. A wooden block of cross-sectional area A and mass density p when floating in water displaces a
volume V. If a person of mass m jumps on this block, calculate the period of small oscillations.
3.5. One end of a spring of spring constant k is fixed while the other end is pulled horizontally with a
force F for a time t0, starting from its equilibrium position x0. Show that

F( Tk x [k
x = x0 + — cos-W —(t- f0) - cos-W — t
k \ Vm vm
3.6. A block of mass M is tied to a horizontal spring of stiffness k. A small mass m is placed on M. The
coefficient of friction between M and m is (i. For what value of the amplitude A will the mass m
slip away from Ml Clearly describe the conditions that lead to this.
3.7. A mass of 0.2 kg is attached to a spring having a spring constant 12 N/cm. The mass is displaced
6 cm and then released. Calculate v0, To, vmm, and total energy. Graph these as a function of time.
3.8. In Problem 3.7, if the mass is released with a velocity of 5 cm/s from its displaced position, cal-
culate v0, To, vmax, A, the maximum potential energy, the maximum kinetic energy, and the total
energy.
3.9. Suppose the motion in Problem 3.7 is taking place in a viscous medium. In 16 seconds the am-
plitude reduces to half of its initial value. Calculate (a) the damping constant, (b) the period Tx
and frequency v0, and (c) the decrement of motion. What are the significant differences in the two
situations?
3.10. A pendulum with a time period T spends time Af between x and x + Ax. Calculate the value of At/T
and make a plot of AtlT versus x for different values of amplitude A. Calculate the area under each
curve. What is the significance of this result? Note:

A?
T
3.11. Derive Eq. (3.62) directly from Eq. (3.60).
102 Harmonic Oscillators Chap. 3

3.12. Explain why the constants in Eq. (3.60) are complex, while B and C in Eq. (3.61) and A and <fr in
Eq. (3.62) are real?
3.13. An oscillator when undamped has a time period To, while its time period when damped is T. Sup-
pose after n oscillations the amplitude of the damped oscillator drops to 1/e of its original value.
Show that

1/2 1
1
4TT n

and
8TTV

3.14. Consider a harmonic oscillator of mass m under a restoring force — kx and damping force —bx. It
is displaced a distance +A and then released with zero velocity. Find the equation representing the
underdamped, critically damped, and overdamped motions.
3.15. Discuss Problem 3.14 if mass m is released with velocity v0 from the displaced position.
3.16. In the case of a damped harmonic oscillator, find the position of x(t) when its values are maximum
and minimum.
3.17. Discuss the motion of a damped oscillator subject to a constant force Fo.
3.18. In the case of a critically damped oscillator for which the velocity at the equilibrium position is v0
and that is subject to a force Fo cos u>t, calculate the value of x(t) and make a graph for some mean-
ingful values of the constants.
3.19. In the case of a critically damped oscillator for which the velocity at the equilibrium position is v0
and that is subject to a force Fo sin u>t, calculate the value of x(t) and make a graph for some mean-
ingful values of the constants.
3.20. Consider a damped oscillator for which y = OIQ/4, having a velocity D0 at x0 at t = 0. After a time
t = 2TT/O)0, a force F = Fo cos u>t is applied. Calculate x(t) and make a graph for some meaningful
values of the constants.
3.21. Consider a forced oscillator for which x = 0 at t = 0 and F = Fo cos(wf + 0). Find the particular
solution and the general solution and make a graph for some meaningful values of the constants.
3.22. Consider a damped oscillator for which y = OIQ/4. If the driving force is given by

F = A, cos oit + A3 cos 3a>t

calculate the value of x(f).
3.23. One end of a massless spring of natural length l0 is attached to a fixed point, while the other end is
attached to a mass m. The whole system is being immersed in a viscous medium that exerts a force
on m proportional to the velocity of m ( = kv). The force constant of the spring is of the form k2/4m.
Initially, the spring is stretched to a length x0 and released. Find the position of m as a function of
time; evaluate all the arbitrary constants in terms of the initial conditions. Calculate the rate at
which the total energy is changing at time t — rnnlk, and graph this as a function of time.
3.24. A mass of 0.1 kg is attached to a spring of stiffness constant 20 N/cm. The mass is displaced 10 cm
and released in a viscous medium. In 12 seconds the amplitude reduces to Me of its initial value.
Calculate (a) the damping constant, (b) w, (c) the decrement of motion, (d) the characteristic time,
and (e) x(t), and graph it.
3.25. In the case of the forced oscillator considered in the text, using the solution x = A cos(a>t + ff),
solve for xs.
Suggestions for Further Reading 103

3.26. Prove Eq. (3.124); that is, the maximum height of the resonance curve falls to one-half its value
when y2 = (co — a>0)2.
3.27. Show that the maximum height of the resonance curve is lly2.
3.28. Prove Eq. (3.129); that is,

(P) = \xmFQ sin 4>

3.29. In Eq. (3.129), substitute the value of sin <fi and state the final result in terms of &>, « 0 , and y.
3.30. Derive Eq. (3.130).
3.31. Consider a damped oscillator with m = 0.2 kg, k = 100 N/m, and b = 5 N-s/m. The oscillator is
driven by a force F = (1.6 N) cos 20?.
(a) If the displacement is given by x = A cos(wf — 8), what are the values of A and 5?
(b) Calculate the energy dissipated in one cycle.
(c) What is the average power input?
(d) Graph the energy and energy dissipated versus time.

SUGGESTIONS FOR FURTHER READING

ARTHUR, W., andFENSTER, S. K., Mechanics, Chapter 6. New York: Holt, Rinehart and Winston, Inc., 1969.
BARGER, V., and OLSSON, M., Classical Mechanics, Chapter 1. New York: McGraw-Hill Book Co., 1973.
BECKER, R. A., Introduction to Theoretical Mechanics, Chapter 7. New York: McGraw-Hill Book Co., 1954.
DAVIS, A. DOUGLAS, Classical Mechanics, Chapter 3. New York: Academic Press, Inc., 1986.
FOWLES, G. R., Analytical Mechanics, Chapter 3. New York: Holt, Rinehart and Winston, Inc., 1962.
FRENCH, A. P., Vibrations and Waves, Chapters 3 and 4. New York: W. W. Norton and Co., Inc., 1971.
HAUSER, W., Introduction to the Principles of Mechanics, Chapter 4. Reading, Mass.: Addison-Wesley
Publishing Co., 1965.
HOCHSTADT, H., Differential Equations—A Modern Approach. New York: Holt, Rinehart and Winston,
Inc., 1964.
KJTTEL, C , KNIGHT, W D., and RUDERMAN, M. A., Mechanics, Berkeley Physics Course, Volume 1, Chap-
ter 7. New York: McGraw-Hill Book Co., 1965.
KLEPPNER, D., and KOLENKOW, R. J., An Introduction to Mechanics, Chapter 10. New York: McGraw-Hill
Book Co., 1973.
MARION, J. B., Classical Dynamics, 2nd. ed., Chapters 3 and 4. New York: Academic Press, Inc., 1970.
MURPHY, G. M., Ordinary Differential Equations and Their Solutions. New York: Van Nostrand Rein-
hold, 1960.
ROSSBERG, K., Analytical Mechanics, Chapter 10. New York: John Wiley & Sons, Inc., 1983.
SLATER, J. C , Mechanics, Chapter 2. New York: McGraw-Hill Book Co., 1947.
STEPHENSON, R. J., Mechanics and Properties of Matter, Chapter 5. New York: John Wiley & Sons, Inc.,
1962.
S ™ O N , K. R., Mechanics, 3rd. ed., Chapter 2. Reading, Mass.: Addison-Wesley Publishing Co., 1971.