16-741 Mechanics of Manipulation Project

A Semiquantitative Analysis of Tippe Top Inversion

Kevin Yoon
April 25, 2006

Introduction

The Tippe Top is a type of spinning top (Fig. 1) that exhibits the strange
behavior of inverting itself such that it spins on its stem after being spun on
its spherical base. The mass of the top is distribued symmetrically about
the stem axis and the center of mass, C, is located on the stem axis while
not coinciding with the geometric center of the sphere. (See Fig. 2.)

Figure 1: A typical tippe top.

The goal of this paper is to provide a brief semiquantitative analysis of the
inversion phenomenon based largely on Cohens work ([1]) which is widely
regarded as one of the first rigorous mathematical treatments of the subject.
Cohens model, derived from Newtonian mechanics, has in recent years been
improved upon by investigators (some of whom approached the problem
from a spectral anaylsis perspective as in [5]). A few assumptions are made
in this paper as they were made in [1] and previous works. Namely, the
axis of angular momentum, l, is assumed to be vertical when in reality this
situation is highly unlikely. It is more probable that l will precess about the
1

Figure 2: Tippe top. (Taken from [1])

vertical at some angle. Also, the horizontal velocity of C, vh , is neglected
here and assumed to be zero. Anyone who has ever spun a top knows this
is also rarely the case. Cohen does address the effect of vh on the equations
of motion derived in [1] but does not go into as much detail as to derive
a formula for vh . These assumptions, however, are not detrimental to the
validity of the explanation found in Cohens model which still illustrates a
very complete picture of the forces and torques at work.

An explanation using energy

The cause of the tippe tops seemingly paradoxical behavior (i.e. transitioning from a gravitationally favorable state to an unfavorable one) was
the subject of much debate before Del Campo [2] argued that frictional
forces played a major role in tippe top inversion. Figure 3 shows that the
z-component of the angular momentum remains dominant both before and
after inversion, meaning that the direction of rotation with respect to the
fixed coordinates of the tops body is reversed. Also, the elevation of the
center of mass means that the kinetic energy of the top decreases in order
to balance the increase in potential energy. Thus, both the total angular
velocity and the total angular momentum decrease during inversion. Given
that a torque is required to reduce angular momentum, it follows that fric2

tional forces must be responsible, since the only other forces acting on the
top are gravity and the normal force exerted by the table, both of which are
parallel to the z-axis and therefore incapable of producing torques along the
z-axis.

Figure 3: Tippe top inversion. (Taken from [1])

Phase I: Inversion on sphere

Figure 4: Coordinate frames. (Taken from [1])

We first define three sets of coordinate frames (Fig. 4a) as defined in
The ex , ey , ez axes are fixed in the laboratory frame with ez pointing
normal to the table surface. The e1 , e2 , e3 axes are fixed with respect to
tops body with e3 pointing along the stems axis. The en , en0 , e3 axes
defined as

[1].
up,
the
are

en = ez e3 /|
ez e3 |
en0 = e3 en

(1)

The unit vectors en , en0 are fixed in the plane in which e1 , e2 rotate (Fig.
4b). It can be seen that the vector en is also fixed in the plane spanned
by ex , ey remaining horizontal at all times. Rotation about the stem axis is
denoted by ; nutation, or the inclination from the vertical, is denoted by
; and precession, or rotation about the vertical, is denoted by . From Fig.
5, we can also obtain the relationship
ez = sin
en0 + cos
e3

(2)

Figure 5: ez in terms of e3 and en0

We can then express the angular velocity of the en , en0 , e3 frame as measured from the laboratory frame:
en +
ez
=
en + sin

=
en0 + cos
e3

(3)

( n en + n0 en0 + 3 e3 )
If we momentarily neglect the translational motion of the top, it is clear
(given an initial counter-clockwise applied rotation when looking down on
the top) that the sliding frictional force, Ff , at the point of contact, T ,
opposes the motion of T in the table surface plane along the
en axis. The
resulting torque, Nf , about the center of mass is

Nf = r Ff
= (a
e3 R
ez ) (|Ff |
en )
= (a
e3 R(sin
en0 + cos
e3 )) (|Ff |
en )

[using Eq. 2]

= |Ff |[a(
e3
en ) + R(sin(
en0 en ) + cos(
e3 en ))]
= |Ff |[a
en0 + R(sin(
e3 ) + cos
en0 )]
= |Ff |[(Rcos a)
en0 Rsin
e3 ]

(4)

This frictional torque Nf is the only torque with components about the
e3 and en0 axes since normal forces can only affect the torque about en .
Equation 4 also reveals that without the frictional force at T , there are no
torques that can cause inversion further verifying the key role of Ff . We
can see that for < cos1 (a/R), the torque about en0 is positive and the
torque about e3 is negative. It follows then that n0 would increase and
3 would decrease over time. Thus, the ratio 3 /n0 = cot (from Eq.
3) decreases over time, meaning that increases, or the top tips. When
cos1 (a/R) /2, the torques about both e3 and en0 are negative.
However, the magnitude of the torque about e3 is larger so provided that the
inertial moments I3 and In0 about e3 and en0 respectively are approximately
the same (or I3 < In0 ), 3 decreases more rapidly than n0 , so will continue
to decrease. Similarly, if > /2, the torques are still negative so 3 and n0
must continue to decrease with 3 becoming more negative since it passes
through zero at = /2.

(a) Rotation axis does not coincide

with center of sphere. (Taken from
[7].)

(b) Point of contact slides around rotation axis providing the torque necessary for inversion. (Taken from [7].)

Figure 6: Motion of point of contact

When = /2, the magnitude of the translational velocity of T is at a
maximum since the radius of the circular path that the contact point T
5

traverses around the rotation axis going through the center of mass (Fig.
6) is at a maximum. This translational sliding is also a source of friction.
In fact, at = /2 it is the only source of friction because 3 = 0 so there
is no friction due to spinning. Translational velocity was ignored when
considering < /2 because the friction at T due to the spin about the
stem axis (spinning friction) was in roughly the same direction as, and for close to zero - relatively insignificant compared to, the friction due to
translation (sliding friction). When > /2, however, the spinning friction
acts in the opposite direction of the sliding friction, but since the sliding
friction is larger in magnitude (at least initially when n0 >> 3 ) the total
friction still points in the
en0 direction. The resulting Nf continues to
exert negative torques about e3 and en0 meaning that 3 and n0 continue

to decrease. Since 3 = cos

and n0 = sin,
must continue to increase.

Phase II: Rising on the stem

Once increases to the point that the stem touches the ground (Fig. 7),
there is a transfer of weight to the stem and the behavior becomes identical to
what happens when an ordinary top is spun. The radius R0 of the stem that
contacts the table is much smaller than that of the sphere of the tippe top,
hence it has a smaller moment arm, R0 sin, about which frictional forces
can retard the rotation about the stem axis allowing the top to continue
spinning.
It can be seen that when the stem touches the table, the velocity of the
contact point T 0 slides across the table surface at a greater velocity than
T since it traverses a circular path with a larger radius. Furthermore, due
to the smaller stem radius R0 as compared to R, the spinning velocity at
T 0 due to rotation about e3 is smaller than it is at point T . For some
arbitrary normal force, this translates into a greater frictional force in the

en0 direction at T 0 than at T . Because of this higher potential frictional

force, if is sufficiently greater than zero at the time the stem touches the
table, there is a transfer of weight from the sphere to the stem while the
torques Nf and N0 f about C (as shown in Fig. 7) decrease both n0 and
3 . If the stem has a spherically rounded tip of radius R0 , then the same
argument used in Section 3 (where a now represents the distance from C to
the spherical center of the stems tip) can be used to explain the continued
increase in , or the rising of the top on its stem. Eventually, n0 and
3 decrease enough such that sliding friction dominates over the spinning

Figure 7: The tippe top with both the sphere and stem in contact with the
table
friction at T 0 . The result is an abrupt change in the direction of the dominant
frictional force at T 0 and a torque N0 f that points in the opposite direction
as the vector depicted in Fig. 7 which stabilizes the top.

Conclusion

Tippe top inversion is a surprisingly complex phenomenon which has been

analyzed in greater detail in [4] and [5] among others. Fortunately, it is possible to explain (at not too high a level) in terms of Newtonian mechanics.
This paper outlines the nature of the forces involved in tippe top inversion,
albeit based on several assumptions. It reinforces the key role of friction
over other parameters such as inertial moments, although clearly these parameters and others (including initial angular velocity, initial , and length
of stem) also influence inversion to some degree.

References
[1] Cohen, R.J., The Tippe Top Revisited, American Journal of Physics, vol.
45, p. 12, January 1977.
[2] Del Campo, A.R., Tippe Top (Topsy-Turnee Top) Continued, American
Journal of Physics, vol. 23, p. 544, 1955.
[3] Or, A.C. The dynamics of a tippe top, SIAM Journal of Applied Mathematics, vol. 54, p. 597-609, 1994.
[4] Soodak, H., A geometric theory of rapidly spinning tops, tippe tops, and
footballs, American Journal of Physics vol. 70, p. 815, August 2002.
[5] Bou-Rabee N.M., J.E. Marsden, and L.A. Romero, Tipee Top Inversion as
a Dissipation-Induced Instability, SIAM J. Applied Dynamical Systems, vol.
3, no. 3, p. 352-377, 2004.
[6] Hugenholtz, N.M., On tops rising by friction, Physica, vol. 18, p. 515-527,
1952.
[7] http://www.fysikbasen.dk/TippetopENGLISH.php