Rotational equations of motion for an applied force $\vec{F}$ located at $\vec{r}$ relative to the center of mass _C are:
$$ \vec{r} \times \vec{F} = {\rm I}_C \dot{\vec{\omega}} + \vec{\omega} \times {\rm I}_C \vec{\omega} $$
where ${\rm I}_C$ is the 3×3 rotational inertia matrix. So lets look at the frisbee example:
- $\vec{\omega} = (0,\Omega,0)$ rotation about the y-axis
- $\vec{r} = (0,y,0)$ along the axis of rotation
- $\vec{F} = (F,0,0)$ perpendicular to the axis of rotation
- ${\rm I}_C = \begin{bmatrix}
\frac{m}{4} R^2 & 0 & 0 \\
0 & \frac{m}{2} R^2 & 0 \\
0 & 0 & \frac{m}{4} R^2
\end{bmatrix}$
- $\begin{pmatrix} 0 \\ 0 \\ -F y \end{pmatrix} =\begin{bmatrix}
\frac{m}{4} R^2 & 0 & 0 \\
0 & \frac{m}{2} R^2 & 0 \\
0 & 0 & \frac{m}{4} R^2
\end{bmatrix} \begin{pmatrix} \dot{\Omega}_x \\ \dot{\Omega}_y \\ \dot{\Omega}_z \end{pmatrix} +\begin{pmatrix} 0 \\ 0 \\ 0 \end{pmatrix} $ equations of motion
- $\dot{\Omega}_z = -\frac{4 F y}{m R^2} $ and all other angular accelerations are zero
So the components of rotation are going to change over time, starting from the z-axis.
Note that any perturbation is not necessarily stable. See this post: https://physics.stackexchange.com/a/17507/392
Also if the force is through the rotation axis ($y=0$) then there is no rotation. $\dot{\Omega}_z =0$