You are on page 1of 29

NORMAL MODES WAVE MOTION AND THE WAVE EQUATION Hilary Term 2007 Dr P G J Irwin NORMAL MODES Overview

w 1. What is a normal mode? 2. Simple Example: 2 identical coupled pendula. 3. General solution as a superposition of the normal modes. 4. Alternative methods of solution: Normal Modes and Matrix formulation 5. Energy of Motion 6. Extension to non-identical pendula 7. Diagrammatic Representation 8. Coupled Pendula damping 9. Forced oscillations 10. More than 2 components 11. N very large

NM 1

What is a Normal Mode? For a system of N coupled 1-D oscillators there exist N normal modes in which all oscillators move with the same frequency and thus have fixed amplitude ratios. N.B. Normal mode is for whole system. Even though uncoupled angular frequencies need not necessarily be the same, for N.M. the effect of coupling is that all bodies move with the same frequency. If initial displacements a) correspond to N.M. then oscillations continue in normal mode. b) dont correspond to N.M. then get non-harmonic motion of each component. In 1-D, system with N components gives rise to N normal modes. If each oscillator is allowed to move in -D, then N normal modes exist. Start with simple system of two coupled pendula - will discuss more interesting cases later. Notational definitions Uncoupled angular frequencies (of jth oscillator): j Coupled angular frequencies (of ith normal mode): i Displacement (of jth oscillator): x j = ( x0 ) j cos ( i t + i )

where

( x0 ) j

are the relative amplitudes are the normal mode frequencies (same for all j). extra phase delays (for 2-component system 1 = 2)

i i

Usual to fold the phase delay into the relative amplitudes (i is often 0 or p), i.e. x j = ( x0 ) j cos i t where ( x0 ) j can then be negative.

NM 2

Coupled Pendula Consider two massless rods of length l, which have bobs of mass m attached to the end, which are themselves connected by a spring.

x l F x

y l F

Assumptions: 1) Assume that spring obeys Hookes law and thus that the restoring force varies linearly with extension, i.e. F = k ( y x ) 2) Assume displacements from equilibrium positions are small such that the restoring force due to gravity for each pendulum is x mg sin = mg l and acts along line of masses. N.B. here 1 = 2 = g / l The equations of motion are then: mx = mgx / l + k ( y x ) and my = mgy / l k ( y x ) Check: Add equations

(1) (2)

mg ( x + y) l (x + y)/2 is the centre of mass of the two bobs and thus this equation must not involve k (equal and opposite forces). This is indeed the case. m( x + y) =

NM 3

Standard method for 2-component normal mode systems

Look for solutions where both masses move with the same angular frequency :
x = x0 cos (t + ) y = y0 cos (t + )

eit } ) ( = Re { x0

eit } ) ( = Re { y0

then

x = 2 x0 cos t = 2 x y = 2 y0 cos t = 2 y
and substitute into Eqs. 1,2:
mg + k x0 + ky0 = m 2 x0 l mg and kx0 + k y0 = m 2 y0 l

Putting A = g / l + k / m and B = k / m , these equations may be rewritten

( A ) x + By and Bx + ( A ) y
2 0
2 0

0
0

=0 =0

(3) (4)

So we have two simultaneous equations for x0, y0 with constants equal zero. either x0 = y0 = 0 (not very interesting) or we need to choose such that the two equations are consistent (i.e. are the same). In such circumstances all we can determine is the ratio of x0/y0: x0 B A 2 = = y0 A 2 B which means that So

( A )
2

2 2

= B2

and thus and so

( A ) = B
12 = A + B
= g /l = 2 or or or
2 2 = A B

= g / l + 2k / m = 2 + 2k / m

NM 4

These s are the normal frequencies which depend on the spring constant k as:

2 2

22 12
k

So much for the normal frequencies, how do we find the amplitudes x0 and y0 and thus describe the motion of the normal mode? For each value of 2 substitute back into either Eq. 3 or Eq.4.: Substituting for 12 = A + B into Eq.3 gives: Bx0 + By0 = 0

x0 = y0 x0 =1 y0 and thus the solution for the positions x and y are: x = C cos (1t + ) = D cos 1t + E sin 1t
y = C cos (1t + ) = D cos 1t + E sin 1t

where C and 1(note the phase difference is the same for both masses) or equivalently D and E are arbitrary constants determined by the boundary conditions. In this mode, both pendula move in phase and thus the length of the spring remains constant. Hence the oscillation frequency is simply the single pendulum frequency.

NM 5

2 = A B into Eq.3 gives: Substituting for 2 Bx0 + By0 = 0

x0 = y0 x0 = 1 y0 and thus the solution for the positions x and y are: x = F cos ( 2t + 2 ) = G cos 2t + H sin 2t
y = F cos ( 2t + 2 ) = G cos 2t H sin 2t

where F and 2 (or equivalently G and H) are again arbitrary constants. In this mode, both pendula move in antiphase and thus the length of the spring is constantly varying. Hence the oscillation frequency includes the spring constant k.

NM 6

General solution.

Because the equations of motion (Eqs 1,2) are linear (under the approximations we have assumed), the general solution of the displacements x and y are a linear combination of the normal mode solutions, i.e.:

x = D cos 1t + E sin 1t + G cos 2t + H sin 2t y = D cos 1t + E sin 1t G cos 2t H sin 2t and differentiating x = 1 D sin 1t + 1 E cos 1t 2G sin 2t + 2 H cos 2t y = 1 D sin 1t + 1 E cos 1t + 2G sin 2t 2 H cos 2t Examples.
Example 1 Normal Mode Excitation Suppose that at t = 0, x = a, y = a and the masses are initially at rest. Equating the initial positions to Eqs 5 and 6: x ( 0) = D + G = a
y (0) = D G = a

(5) (6) (7) (8)

which clearly gives D = a, G = 0. Equating the initial velocities to Eqs. 7 and 8: x ( 0 ) = 1 E + 2 H = 0


y ( 0 ) = 1 E 2 H = 0

giving E = H = 0.

x = a cos 1t y = a cos 1t and thus we have excited the first normal mode only which is to be expected given the initial displacements. In addition, once in this normal mode, the system will remain in it indefinitely.
Example 2 Normal Mode Excitation Suppose that at t = 0, x = y = 0, and the masses are given initial velocities x = v , y = v . Equating the initial positions to Eqs 5 and 6: x (0) = D + G = 0
y (0) = D G = 0

Hence the solution for t > 0 is

which clearly gives D = 0, G = 0. Equating the initial velocities to Eqs. 7 and 8: x ( 0 ) = 1 E + 2 H = v


y ( 0 ) = 1 E 2 H = v

giving E = 0, and H = v / 2 and thus the subsequent motion is: v x = sin 2t

2
v

y=

sin 2t

NM 7

And thus we have excited the second normal mode only, in which the system will remain. The motion in these two normal modes may also be summarised by the following figure:

where here the coupling is such that the frequency of the 2nd mode is higher than that of the first.

NM 8

Example 3 Non-Normal Behaviour - Beats Suppose that at t = 0, x = a, y = 0 and the masses are initially at rest. Equating the initial positions to Eqs 5 and 6: x ( 0) = D + G = a
y (0) = D G = 0

which clearly gives D = a/2, G = a/2. Equating the initial velocities to Eqs 7 and 8: x ( 0 ) = 1 E + 2 H = 0
y ( 0 ) = 1 E 2 H = 0

giving E = H = 0. Hence the solution for t > 0 is

a a cos 1t + cos 2t 2 2 a a y = cos 1t cos 2t 2 2 and thus both normal modes are excited. The solution for both x and y is then determined by the beating of the two normal frequencies 1 and 2. x=
Eqs 9 and 10 can be re-written from standard trigonometrical identities as:

(9) (10)

+ 2 2 x = a cos 1 t cos 1 t = a cos ( t ) cos ( t / 2 ) 2 2 + 2 1 2 y = a sin 1 t sin t = a sin ( t ) sin ( t / 2 ) 2 2 + 2 and = 1 2 which is a standard beat between two waves. where = 1 2

NM 9

acost/2

asint/2

2/

4/

N.B. Transfer of energy between two pendula. Note also that one period of the envelope equals two beats.

NM 10

Alternative methods of solution: Normal coordinates or Decoupling

The equations of motion for the coupled pendula are again: mx = mgx / l + k ( y x )
my = mgy / l k ( y x )

(1) (2)

For many coupled oscillator systems it is possible to make linear combinations of the equations of motion which may independently solved for a linear combination of the position variables, in this case x and y. If we add Eqs 1 and 2 we find:

m( x + y) =
or

mg ( x + y) l

g q1 (11) l where q1 is a normal coordinate here equal to q1 = ( x + y ) / 2 . Eq. 11 describes simple harmonic motion which may be trivially solved to give: q1 = D cos 1t + E sin 1t where 12 = g / l is the first normal frequency found earlier. q1 =
Similarly, if we subtract Eqs 1 and 2 we find: mg m( x y) = ( x y ) 2k ( x y ) l or g 2k q2 = + q2 l m where q2 is another normal coordinate, equal to q1 = ( x y ) / 2 . Eq. 12 again describes simple harmonic motion for q2 and thus q2 = G cos 2t + H sin 2t
2 where 2 = g / l + 2k / m is the second normal frequency found earlier.

(12)

To extract our original position variables x and y we note that x = ( q1 + q2 ) / 2 and

y = ( q1 q2 ) / 2 and hence
1 ( D cos 1t + E sin 1t + G cos 2t + H sin 2t ) 2 1 y= ( D cos 1t + E sin 1t G cos 2t H sin 2t ) 2 which are identical to the general solutions (Eqs 5 and 6) derived earlier since the constants are arbitrary. x=
This method can lead to quick solutions for the normal frequencies if the suitable linear combination of parameters can be spotted. For simple cases like this it is easy but not for more complicated systems This technique is also known as decoupling. NM 11

Energy of Motion

Decoupling is also instructive when the energy of the system is considered. Consider first the potential energy of the coupled oscillators. Consider the forces acting on particle 1. mx = mgx / l + k ( y x ) This force may be related to the partial derivative w.r.t. x of a potential energy function V: V mx = mgx / l + k ( y x ) = x integrating we find: mg 2 1 2 V= x + kx kxy + f ( y ) 2l 2 where f is an unknown function of y. The force acting on particle 2 may be similarly integrated w.r.t. y to give a second expression for the potential energy. my = mgy / l k ( y x ) hence mg 2 1 2 V= y + ky kxy + g ( x ) 2l 2 where g is an unknown function of x. Combining these two expressions, the total potential energy of the system is given by: 1 g k V = m + ( x 2 + y 2 ) kxy . 2 l m Consider now the kinetic energy K. This is given simply by: 1 K = m ( x2 + y 2 ) 2 While the expression for K is straightforward, that for V is rather more complex. However if we substitute for the normal coordinates: x = ( q1 + q2 ) / 2 and

y = ( q1 q2 ) / 2 , then the V may be re-expressed as:


V=
1 g 2 1 g 2k 2 m q1 + m + q2 2 l 2 l m 1 1 2 2 q2 = m12 q12 + m 2 2 2

where 1 and 2 are the normal mode angular frequencies. Similarly K may be rewritten as:

NM 12

1 2 K = m ( q12 + q2 ) 2
and the total energy is then:

E=

1 1 1 1 2 2 2 m12 q12 + m 2 q2 + mq12 + mq2 2 2 2 2 1 1 2 1 1 2 2 q2 + mq2 = m12 q12 + mq12 + m 2 2 2 2 2 = E1 + E2 =

N 1

En

HENCE THE TOTAL ENERGY OF THE SYSTEM IS THE SUM OF THE ENERGIES OF THE NORMAL MODES. (Example of Parsevals Theorem)

NM 13

Alternative methods of solution: Matrix method

The standard and normal coordinate approach are both fine for systems of two coupled oscillators. However, what if the number of oscillators is greater than two? The equations of motion for the coupled pendula may be rewritten as k g k x + y = x m l m k g k x + y = y m l m These equations may be written in terms of a matrix and column vectors as k g k l + m x m x (13) = k g k y y + m l m or x = x x where x is the column vector and A the square symmetric matrix in Eq.13. y We again look for normal mode solutions where all elements oscillate with the same frequency, i.e.
x = x0 cos (t + ) y = y0 cos (t + )

or where

x = x 0 cos (t + )
x x x = and x 0 = 0 y y0

Differentiating we find that

x = 2 x 0 cos (t + ) = 2 x .

Substituting into Eq. 13 we find:


x = 2 x

or or

( + I) x = 0
2

( + I) x
2

=0

(14)

where I is the unit matrix. The solutions of Eq. 14 are identical to those described earlier in section on the standard method of solution, i.e. either x0 = 0, which is not very interesting, or the equations must be consistent which implies that the determinant of the matrix ( + 2 I ) must be equal to zero. Hence

NM 14

k 2 g k l + m m =0 k g k 2 + m l m

leading to:

2
and hence

k g k + = m l m as before.

12 = g / l

or

2 2 = g / l + 2k / m

The normal mode amplitudes x0 are again found by substitution for the normal mode frequencies into Eq. 14 as before and hence for the two respective normal modes: 1 1 x 0 = C or x0 = F 1 1 or 1 x = ( D cos 1t + E sin 1t ) 1 or 1 x = ( G cos 2t + H sin 2t ) 1 and hence in general x 1 1 x = = ( D cos 1t + E sin 1t ) + ( G cos 2t + H sin 2t ) y 1 1 The advantage of the matrix method is its general applicability, and the ease with which it may be applied to systems with more than two normal modes.

NM 15

Non-Identical Pendula

Consider the non-identical coupled oscillator system below

y x l1 F x The equations of motion are: mx = mgx / l1 + k ( y x ) and my = mgy / l2 k ( y x )


x = x0 cos (t + ) y = y0 cos (t + )

l2 F

(1) (2)

Again look for solutions where both masses move with the same angular frequency :

then

x = 2 x0 cos (t + ) = 2 x y = 2 y0 cos (t + ) = 2 y

and substitute into Eqs. 1,2:


mg + k x0 + ky0 = m 2 x0 l1 mg and kx0 + k y0 = m 2 y0 l2

Putting: A = g / l1 + k / m = 12 + k / m B = k / m C = g / l2 + k / m = 22 + k / m these equations may be rewritten ( A 2 ) x0 + By0 = 0 and Bx0 + ( C 2 ) y0 = 0 These equations can be solved in the same way as before to give the result:

NM 16

2 2 22 ) + ( 2k / m ) 2 x0 m 2 = ( 12 22 ) ( 12 22 ) + ( 2k / m ) 2k y0 N.B. ( x0 / y0 )1 for mode 1 and ( x0 / y0 )2 for mode 2 are related by:

2 = ( 12 + 22 ) + 2k / m 2

2 1

x0 x0 = 1/ y0 1 y0 2 N.B. This is not always true! It just happens to be so in this case since the pendulum bobs have the same mass.

Simple checks that these formula are correct. Adding or subtracting we get the two modes: Add: A 1) if 12 = 22 then 2 = 2 + 2k / m , Subtract: B

or

2 = 2

2 2 x0 ( 1 2 ) m = k y0 -

2) as k 0 then 2 = 12

or or or

2 = 22
x0 k = 2 2 y0 m ( 1 2 ) 0

3) k very large then

2 = 2k / m
x0 1 y0

or or

2 =

1 2 1 + 22 ) ( 2

x0 1 y0

Spring very strong forget gravity

Spring unstretched

NM 17

Nature of Normal Mode as k/m changes

Variation of frequency and normal mode amplitudes for coupled non-identical pendula where natural frequencies are 1 = 1.0 and 2 = 0.3 respectively.

Curves A correspond to taking positive choice for the sign in the equations, and Curves B correspond to the negative choice. For Curves A: Small k/m Large k/m For Curves B: Small k/m Large k/m i.e. SIMPLEST TYPE of MOTION OF SYSTEM DEPENDS ON VALUE OF COUPLING Shorter pendulum ~ at rest, longer pendulum oscillates. Pendula oscillate in phase with similar amplitudes Longer pendulum ~ at rest, smaller pendulum oscillates. Pendula oscillate in antiphase with similar amplitudes

NM 18

Diagrammatic Representation of Normal Modes

Normal mode motion specified by x0/y0. Can draw as a unit-length vector on Cartesian axes. Two Normal Modes Two vectors. Consider case of non-identical pendula, For k/m 0 For k/m Intermediate k/m
x0/y0. x0/y0.

A(+) 1

or or
2 k / m

B() 0 1

y0 = tan = x0 ( 12 22 )

2 1

22 ) + ( 2k / m )
2

NM 19

Non-N.M. Behavour Non Identical Pendula

For non-identical pendula OF EQUAL MASS we found earlier that


x0 x0 m 2 2 = 1/ = ( 1 2 ) + 2k y0 1 y0 2

2 1

2 2 22 ) + ( 2k / m ) = r

Hence the general solution is:


x 1 r x = = ( D cos 1t + E sin 1t ) + ( G cos 2t + H sin 2t ) y r 1 x a Suppose at t = 0, x = = and x = 0 . Having zero initial velocities means that y 0 the sine terms disappear. Hence a 1 r x = = D +G 0 r 1 which may be solved to give
2 2 x (t ) = a cos 1t + r cos 2t / (1 + r )

y ( t ) = ar [ cos 1t cos 2t ] / (1 + r 2 )

This is a little more difficult to simplify but it can be shown that


1 r2 x ( t ) = a cos ( t ) cos ( t / 2 ) a sin ( t ) sin ( t / 2 ) 2 1+ r y ( t ) = 2ar sin ( t ) sin ( t / 2 ) / (1 + r 2 )

It can be seen that


x varies between

and

1 r2 a 2 1+ r

2r a 2 1+ r Hence get incomplete transfer of energy


y varies between

and

NM 20

Figure showing beats of non-identical pendula. Note the incomplete energy transfer.

NM 21

Coupled pendula damping.

Consider our coupled pendulum system again. Suppose both bobs are also subject to a friction force equal to times their velocity. Then the equations of motion must be modified to: mx = mgx / l + k ( y x ) x (15) and my = mgy / l k ( y x ) y (16) These equations may be solved by any of the methods previously. Using the method of normal coordinates or decoupling, we can add or subtract these equations to give decoupled equations in terms of the normal coordinates. Adding Eqs. 15 and 16 gives g ( x + y) = ( x + y) ( x + y) l m or

m which may be solved to give q1 = ( D cos 1t + E sin 1t ) exp ( t )


where

q1 +

q1 +

g q1 = 0 l

g 2 2 2 = = = 12 2 and 1 2 2 l 4m 4m 2m where 1 is the frequency of the corresponding normal mode calculated without friction.
1 =
Subtracting the two equations we find g ( x y ) = ( x y ) 2k ( x y ) ( x y ) l m or g 2k q2 + q2 + + = 0 m l m which may be solved to give t + H sin 2 t ) exp ( t ) q2 = ( G cos 2 where
g 2k 2 2 2 2 = + = 2 = 2 2 2 2 2 4m l m 4m Hence expanding q1 and q2 in terms of x and y:

and is as before.

t + H sin 2 t ) exp ( t ) x = ( D cos 1t + E sin 1t + G cos 2 t H sin 2 t ) exp ( t ) y = ( D cos 1t + E sin 1t G cos 2

NM 22

Forced oscillations

Find response of system [= two identical coupled pendula] to external force on one or more elements. Flashback Corresponding problem for single oscillator d 2x dx m 2 + + kx = F cos t dt dt solution consists of complementary function and particular integral. C.F. = P.I. = transient. Dies away as e 2 m because of resistive term. Determined by initial conditions. steady state response to external force. resonant response at
k 2 for small . m 4m

For zero damping ( = 0) transient does not die away with time. k , grows with t. Resonant response at = m Back to coupled pendula With force F cos t acting on particle 1, the equations of motion become g mx + x + m x + k ( x y ) = F cos t l g my + y + m y + k ( y x ) = 0 and l

(17) (18)

We have found the transient response already so all we need do now is find the P.I.. Adding the equations g F ( x + y ) + ( x + y ) + ( x + y ) = cos t m l m or
q1 +

q1 +

where q1 = x + y . To find the P.I. put q1 = Re A exp ( j t ) then:

g F F q1 = cos t = Re exp ( j t ) l m m

g 2 + j + A = F / m m l
and thus
A=

( F / m ) exp ( j1 )
2 g 2 + l m 2 1/ 2

( F / m ) exp ( j1 )
2 2 2 2 ) + ( 1 m 1/ 2

= B1 exp ( j1 )

NM 23

where tan 1 =

( / m )

2 1

2 )

and 1 is the corresponding undamped normal mode

frequency. Hence the normal coordinate q1 is given by: q1 = B1 cos ( t + 1 ) Subtracting the equations of motion (Eqs 17, 18) gives:

( x y) + ( x y) + ( x y) +
m
or

g l

2k F ( x y ) = cos t m m

F g 2k F q2 + + q2 = cos t = Re exp ( j t ) m m l m m where q2 = x y .To find the P.I. put q2 = Re A exp ( j t ) then: q2 +

g 2k 2 + j + + A = F / m m l m
and thus
A= = B2 exp ( j2 ) 2 2 1/ 2 g 2k 2 2 + 2 + 2 ) + 2 ( m m l m ( / m ) where tan 2 = 2 and 2 is the corresponding undamped normal mode ( 2 2 )
2 1/ 2

( F / m ) exp ( j2 )

( F / m ) exp ( j2 )

frequency. Hence the normal coordinate q2 is given by: q2 = B2 cos ( t + 2 ) Reconstructing x and y from q1 and q2:
x= q1 + q2 = B1 cos ( t + 1 ) + B2 cos ( t + 2 ) 2 = ( B1 cos 1 + B2 cos 2 ) cos t ( B1 sin 1 + B2 sin 2 ) sin t = C cos ( t + ) y= q1 q2 = B1 cos ( t + 1 ) B2 cos ( t + 2 ) 2 = ( B1 cos 1 B2 cos 2 ) cos t ( B1 sin 1 B2 sin 2 ) sin t = D cos ( t + )

where
C 2 = ( B1 cos 1 + B2 cos 2 ) + ( B1 sin 1 + B2 sin 2 )
2 2

D 2 = ( B1 cos 1 B2 cos 2 ) + ( B1 sin 1 B2 sin 2 )


2

tan =

( B1 cos 1 + B2 cos 2 ) ( B1 sin 1 + B2 sin 2 )

NM 24

tan =

( B1 cos 1 B2 cos 2 ) ( B1 sin 1 B2 sin 2 )


THIS IS VERY COMPLICATED!

Fortunately you will probably never meet forced coupled oscillators WITH damping. If the damping is set to zero then the equations are greatly simplified. The phase angles 1 and 2 are both zero and thus: F /m F /m q1 = 2 cos t and q2 = 2 cos t 2 2 2 1 Hence combining to get x and y: 2 1 F 12 + 2 2 2 x = ( q1 + q2 ) = cos t = A ( ) cos t 2 2 2m (12 2 )( 2 2 )
2 1 F 2 12 y = ( q1 q2 ) = cos t = B ( ) cos t 2 2 2m (12 2 )( 2 2 )

NM 25

N-coupled oscillators

Extend discussion to N coupled oscillators. Consider the transverse oscillations of N particles of mass m spaced equally along a flexible, elastic, massless string, which is under tension T.

(reproduced from French, 1971). Assume the particles are displaced by small distances yi. If the displacements are small then the length of the string between the particles is increased to l = l / cos l (1 + 2 / 2 ) . Hence if the displacements are small l l and the tension in the string remains constant. Consider the pth particle above. The force acting in the y-direction is:

F = T sin p 1 + T sin p
which may be approximated to:

T T y p y p 1 ) + ( y p +1 y p ) ( l l

Hence the equation of motion of the pth particle is: 2 2 y p + 2 0 y p 0 ( y p+1 y p1 ) = 0


2 = T / ml . We can write a similar equation for each of the N particles and where 0 thus we have N coupled differential equations and thus N normal modes.

Substituting a normal mode solution y p = Ap cos ( t + ) we find:

2 + 2 0 ) Ap 02 ( Ap+1 + Ap1 ) = 0

(19)

This is very hard to solve. Instead we shall assume a solution of the form Ap = C sin p
Ap must be zero for p = 0 and p = N + 1 since the ends are fixed which implies: ( N + 1) = n

NM 26

where n is any integer, then Ap +1 + Ap 1 = C sin ( p + 1) + sin ( p 1)


= 2C sin p cos

But from Eq.19

Ap +1 + Ap 1
and hence solving for :

( =

2 + 2 0 ) Ap 2 0

n = 2 0 sin

n 2 ( N + 1)

Hence the motion of the particles in the nth normal mode (where n varies from 1 to N) may be written as: y p = Ap cos ( nt + n )

pn = Cn sin cos ( nt + n ) N +1 pn = sin ( Dn cos nt + En sin nt ) N +1 and the general solution is again a linear combination of the normal modes:
yp =

N n =1

pn sin ( Dn cos n t + En sin nt ) N +1

What do these modes look like? Here is an example for four coupled oscillators.

(reproduced from French, 1971) Clearly there 4 normal modes in all. Note that n = 6, 7, 8, 9 repeat patterns of n = 4, 3, 2,1 with opposite sign.

NM 27

N very large

As N becomes very large, the chain of coupled oscillators becomes more and more like a continuous string.
1/ 2 n n T n = 2 0 sin = 2 sin ml 2 ( N + 1) 2 ( N + 1)

for finite n, as N becomes large:

T n 2 ml

1/ 2

n T = 2 ( N + 1) m / l

1/ 2

n ( N + 1) l

substituting the linear mass density = m/l and the total length of the string L = (N+1)l we find

n n = L
For particle displacements

1/ 2

pn y p = sin ( Dn cos n t + En sin n t ) N +1

but

pn pln n x = = N + 1 ( N + 1) l L where x = pl is the distance along the string. Hence the general solution is:
yp =

n =1

n x sin ( Dn cos n t + En sin n t ) L

In this case there are an infinite number of normal modes, the coefficients of which may be found by Fourier Analysis (2nd Year Syllabus).

NM 28

NOTES

NM4. eit Why is x0 cos (t + ) = Re x0 ? = a + ib . is itself complex, i.e. x0 The point that is easy to overlook here is that x0 Hence: eit = ( a + ib )( cos t + i sin t ) x0 = a cos t b sin t + i ( b cos t + ia sin t ) = x0 cos (t + ) + ix cos (t + ) NM 27 We have:

Ap = C sin p
Ap +1 + Ap 1 = C sin ( p + 1) + sin ( p 1) = 2C sin p cos

Ap +1 + Ap 1
Substituting for As we find:

( =

2 + 2 0 ) Ap 2 0

( 2C sin p cos =
or

2 + 20 )

02

C sin p

2 202 cos = 2 + 20

and

2 = 202 (1 cos )
2 = 40 sin 2 2

Hence

= 20 sin 2

NM 29

You might also like