A Flapping Toy
Kirk T. McDonald
Joseph Henry Laboratories, Princeton University, Princeton, NJ 08544
(November 30, 2000)
1 Problem
Deduce the frequency of small oscillations of the flapping toy shown in the figure below,
supposing the central mass m moves only vertically, and the motion of the others masses is
only in the x-y plane.
The flapping toy can be approximated as a mass m connected by two massless rods of
length 2c to two masses m/2. The centers of the two rods are suspended from a horizontal
plane by massless strings of length b, with distance 2a between the upper points of suspension.
Lengths a and c similar, and both are smaller than b
2 Solution
This problem was suggested by Frank Moon.
A toy such as that modeled here can be purchased from several vendors [1, 2].
The static equilibrium configuration of the toy is such that its center of mass is such
that the center of mass is as low as possible. Regarding half of the central mass m as being
attached to the left rod, and the other half to the right arm, we see that the center of mass
of each rod system is at the point of attachment to the string.
If c ≤ a then there is only a single equilibrium state, with angles α and β, shown in
the figure above, given by sin α0 = (a − c)/b ≤ 0 and β 0 = 0. However, in this case angle
α is always less than or equal to α0 and does not oscillate about the equilibrium value; an
analysis of small oscillations should be performed in terms of angle β.
If c > a the strings are vertical
√ at equilibrium. There are now two equilibrium states,
with α0 = 0 and sin β 0 = ± c − a2 /c. The frequency of small oscillation about either
2
equilibrium state will be the same, but the amplitude of the oscillation of the central mass
1
√
must by less than c2 − a2 to be called “small”; otherwise the motion of the central mass
will extend beyond both of its equilibrium points.
This system has only one degree of freedom if the motion is restricted to the x-y plane
and the central mass moves vertically. In this case, the left and right rods make the same
angle β to the horizontal, and the two strings make the same (small) angle α to the vertical.
The system is equivalent to two identical systems consisting of a string, a rod, and two
masses of m/2 at the ends of the rod, in which one mass is constrained to have x = 0.
To analyze the motion, it suffices to consider only one of the two subsystems, say the one
at x ≥ 0. The center of mass of the subsystem is at the point of attachment of the string to
the rod. Thus, the potential energy of the subsystem is,
mgb 2
V = mgycm = −mgb cos α ≈ −mgb + α . (1)
2
This form suggests that it will be simpler to analyze the system in terms of coordinate α
rather than β. However, as previously noted, small oscillations about equilibrium in angle
α occur only when c > a.
2.1 Analysis for c > a Using Angle α
In this case the potential (1) describes oscillations about α = 0 with effective spring constant,
keff = mgb (c > a). (2)
To relate α and β we use the constraint that the central mass is at x = 0,
0 = a + b sin α − c cos β, (3)
a + b sin α
cos β = , (4)
c
√
c2 − (a + b sin α)2 c2 − a 2
sin β = 1 − cos β =
2 ≈ = sin β 0, (5)
c c
noting that b sin α c for very small α. To relate β̇ to α̇ we return to eq. (4),1
b cos α
− β̇ sin β = α̇, (6)
c
b cos α bα̇
β̇ = − α̇ ≈ − √ . (7)
c2 − (a + b sin α)2 c2 − a 2
The kinetic energy of the center of mass of the subsystem (which moves in a circle of
radius b) is,
mb2 2
Tcm = α̇ . (8)
2
1
Comparing eqs. (3) and (7) we see that β̇/α̇ → −∞ when c ≤ a and β ≈ β 0 = 0. This confirms that
we cannot use coordinate α to analyze the motion for c ≤ a.
2
The kinetic energy of the rotational motion relative to the center of mass of the subsystem
is,
2
I β̇ mc2 2
Trel = = β̇ . (9)
2 2
The total kinetic energy is, using eq. (7),
mb2 2c2 − a2 2
T = Tcm + Trel ≈ α̇ . (10)
2 c2 − a 2
The effective mass of the subsystem in this case is,
2c2 − a2
meff = mb2 (c > a). (11)
c2 − a 2
The frequency of small oscillation is, recalling eq. (2),
keff g(c2 − a2 )
ω= = (c > a), (12)
meff b(2c2 − a2 )
which is small compared the frequencies g/b or g/c of simple pendula of the same
characteristic lengths as components of the flapping toy.
For analyses of small oscillations when c ≤ a we must use angle β rather than α as the
coordinate. Following these analyses, we return to analyze the case c > a using angle β.
2.2 Analysis for c < a Using Angle β
To eliminate coordinate α in favor of β, we write eq. (3) as,
c cos β − a
sin α = , (13)
b
b 2 − (c cos β − a)2 b2 − (a − c)2 c(a − c)
cos α = 1 − sin2 α = ≈ − β 2 . (14)
b b 2b b2 − (a − c)2
Then, the potential (1) can also be approximated as,
c(a − c)
V = −mgb cos α ≈ −mg b2 − (a − c)2 + mg β2. (15)
2 b − (a − c)
2 2
For c < a the potential is springlike with effective spring constant,
c(a − c)
keff = mg (c < a). (16)
b2 − (a − c)2
To express the kinetic energy in terms of β̇ rather than α̇ we rewrite eq. (6) as,
cβ̇ sin β cβ̇ sin β 0
α̇ = − ≈ = 0, (17)
b cos α b
3
where the approximation holds for small oscillations about the equilibrium β 0 = 0 when
c ≤ a. That is, only the rotational kinetic energy is significant when using coordinate β and
c ≤ a,
mc2 2
T ≈ Trel = β̇ , (18)
2
and the effective mass for coordinate β is,
meff = mc2 (c < a). (19)
The frequency of small oscillation is, recalling eq. (16),
keff g(a − c)
ω= = (c < a), (20)
meff bc b2 − (a − c)2
Thus, the frequency of small oscillations goes to zero as c approaches a either from above
or from below.
2.3 Analysis for c = a Using Angle β
If lengths a and c and equal, then from eq. (14),
c2 β 4 c2 β 4
cos α ≈ 1 − 2 ≈ 1 − 2 (c = a), (21)
4b 8b
and the total (conserved) energy in terms of coordinate β is,
2
mc2 β̇ mgc2β 4 mgc2β 4max
E ≈ Trel + V = + = (c = a), (22)
2 8b 8b
where β max is the maximum angle during an oscillation. Hence,
dβ 1 g
β̇ = = β 4max − β 4 (c = a). (23)
dt 2 b
During one quarter of an oscillation β varies between 0 and β max, so the period τ is given
by,
βmax
b 2 8 b 1 dz 1 2b 2
τ = 4 4 dβ = √ = Γ (1/4)
0 g β max − β 4 β max g 0 1 − z4 β max πg
10.48 b
= (c = a), (24)
β max g
using z = β/β max and Gradshteyn and Ryzhik 3.166.16 [3].
4
2.4 Analysis for c > a Using Angle β
When analyzing the case that c > a using coordinate β the approximation in eq. (14) is not
appropriate (since the resulting potential (15) has no minimum). Instead, we make a Taylor
expansion of cos α as a function of β about its equilibrium value, given by cos β 0 = a/c,
b2 − (c cos β − a)2
cos α = , cos α(β 0 ) = 1, (25)
b
d cos α c sin β(c cos β − a) d cos α(β 0 )
= , = 0, (26)
dβ b b2 − (c cos β − a)2 dβ
d2 cos α c cos β(c cos β − a) − c2 sin2 β c2 sin2 β(c cos β − a)2
= − ,
dβ 2 b b2 − (c cos β − a)2 b[b2 − (c cos β − a)2]3/2
d2 cos α(β 0 ) c2 sin2 β 0 c2 − a 2
= − = − , (27)
dβ 2 b2 b2
such that,
c2 − a 2
cos α ≈ 1 − (β − β 0 )2, (28)
2b2
and the potential can be approximated as,
c2 − a 2
V = −mgb cos α ≈ −mgb + mg (β − β 0 )2 (c > a). (29)
2b
The effective spring constant for this analysis is,
c2 − a 2
keff = mg . (30)
b
When considering the kinetic energy in terms of coordinate β for c > a we cannot neglect
the contribution from the center of mass motion (as was possible for c < a). Recalling
eq. (refe8a,
cβ̇ sin β cβ̇ sin β 0
α̇ = − ≈ , (31)
b cos α b
so the kinetic energy can be approximated as,
mb2 2 mc2 2 m(2c2 − a2 ) 2
T = α̇ + (1 + sin2 β 0)β̇ = β̇ . (32)
2 2 2
The effective mass for this analysis is,
meff = m(2c2 − a2 ), (33)
and the frequency of small oscillations is,
keff g(c2 − a2 )
ω= = (c > a), (34)
meff b(2c2 − a2 )
as found previously in sec. 2.1.
5
References
[1] [Link]
[2] Flying Toucan, [Link]
[3] I.S. Gradshteyn and I.M. Ryzhik, Table of Integrals, Series, and Products, 5th ed. (Aca-
demic Press, 1994), [Link]