Wave
equation with no damping
The wave equation expressed
in polar coordinates with no damping is given by
![]()
For a drum, boundary conditions are:
, ![]()
, ![]()
, ![]()
To prevent a discontinuity, we assume that:
and
![]()
Since the displacement of the drumhead must be finite
at all points, we further assume that:
, ![]()
The analytic solution is:

This is quite complicated. In
order to better visualize this result, we isolate a single term,
![]()
where
is the nth zero of the mth order Bessel
function of the first kind.
The function S(r)
divides the drum into sections, and F(t) gives the time dependent
oscillation of each section. For example, v22(r,θ,t)
gives:

The signs in each section
correspond to the relative sign of v22(r,θ,t) in
that section. The lines are stationary with time.
Below are some animations of
different modes.
λ02
λ
21
λ 22
This behavior can be seen experimentally. The boundaries between sections are stationary, so particles placed on the drumhead will collect these boundaries as the drumhead vibrates. To illustrate this, the drumhead was lightly seasoned with pepper, and tapped first in the center and then along the edge.

It is important to notice
that these are only individual modes, the motion of a real drumhead will be a
linear combination of all modes. Below is an animation of four modes
superimposed.

The presence a particular
mode is explained by orthogonal eigenfunctions. The initial conditions can be
expressed as an orthogonal sum of all possible vibration modes
(eigenfunctions). The more similar an initial condition to a particular mode,
the more prominent that mode will be.
Striking the drum in the
center will only give rise to λ0n modes, since there is
no θ dependence. Striking the drum off center deforms the drumhead
in such a way that λmn modes are excited.
If the drum is struck very
near the edge, there is more energy in high order modes then if the drum is
struck near the center. This explains why a drum sounds higher pitched if
struck near the edge.
Wave
equation with damping
The wave equation with linear
damping can be written

where k is constant.
For simplicity, the initial conditions are assumed independent of theta, and
the equation becomes

The boundary conditions are
the same as above. For the initial conditions, it will be assumed that the drum
stick contacts the drumhead, the drumhead depresses, and then springs upward,
and the stick leaves the head. The instant the stick leaves the head, t
equals zero.

Mathematically, initial
conditions are
Using the separation v(r,
t) = f(r)T(t),
we obtain

where λ2
is constant. This results in two related ODEs.
![]()
![]()
The first ODE is Bessel’s
equation, with a solution of
![]()
Applying boundary conditions
gives
![]()
, αn
is the nth zero of the Bessel function, and a is the radius of the drum
(15 cm).
The time portion of the
solution is:
![]()
Notice that damping not only
causes the solution to decay with time, it also affects frequency content. The
fundamental frequency and envelope function can be used to find the values of k
and c.
The final solution is
,
,
. (AA10)
Where
and
.
It is interesting to note
that an is found using the orthogonality of the Bessel
function, which is analogous to finding a Fourier series.
The frequency spectrum is
. The first few frequencies are listed below.
|
n |
fn (Hz) |
|
1 |
220 |
|
2 |
507 |
|
3 |
795 |
|
4 |
1083 |
Below is a plot of the
Fourier spectra for the measured and calculated data.

There various possible causes
for the presence of extra peaks in the measured Fourier spectrum. Even if the
drum is stuck in the center, modes other than λ0n could
be present because of non-uniform tension in the drumhead. However, the
additional measured frequencies do not correspond to any eigenvalues. A likely
cause is non-uniform tension on the drum head, leading to different regions
vibrating at slightly different rates.
The sound generated by the
analytic solution can be played.
Vibration
of a Guitar String
The vibration of a guitar
string is solved by the 1-dimensional wave equation.
![]()
As a is defined as the
length of the guitar string (in meters), the boundary conditions will be
defined as:
![]()
The boundary conditions are
both equal to zero because the string is fixed at both ends.
Separation of variables gives
a solution of

where a = .8 m. The
value the constant c can be found by noticing that the fundamental
frequency is
. The experimental data gives f1
= 73 Hz, resulting in c = 116.8 . The string will be plucked
a distance b from the end. The initial conditions are therefore
![]()
For simplicity, we will let d0
= 1, since this is only scales the magnitude and has no effect on frequency
content. With a = .8 m and b = .15, the constants An
and Bn are:
![]()


·
The dampening of an
oscillating system not only causes decay with time, the frequency content is
also affected.
·
Guitar strings have
eigenfunctions consisting of sines and cosines.
·
Drums have
eigenfunctions consisting of sines and cosines multiplied with Bessel functions
of the first kind.
·
The eigenvalues (and
therefore frequencies) of a guitar string are all integer multiples of each
other, which gives the guitar string a musical quality.
·
The eigenvalues of a
drum are not integer multiples of each other, so a drum is not a harmonic
instrument. In order for a drum to have a definite pitch (such as a timpani),
one vibrational mode must largely dominate the rest.
·
For any linear
system, the system response will only contain eigenfunctions that are present
in the initial condition or forcing function.