Self-interaction in spin-orbit coupling? In spin orbit coupling, in an atom motion of electrons about nucleus generates magnetic field and we consider this field to interact with magnetic moment of electron. It sound strange as in electrostatics a field is generate by a charge particle but this field does not interact with its origin charge. Doesn't this contradict ?
 A: A good picture is to jump on board the electron and ask yourself what is going on in the reference frame in which it is momentarily at rest. In that frame there is a magnetic field whose source is the nucleus of the atom. It can be calculated conveniently by a Lorentz transform from the rest frame of the nucleus. Anyway, there it is. The magnetic dipole of the electron then interacts with this magnetic field. The only difficulty with this approach is that you have to then allow for the Thomas precession, which is a precession of the axes of the sequence of inertial frames in which the electron is momentarily at rest.
The other perspective is to stay in the rest frame of the nucleus, where there is just the electric field of the nucleus. Then the moving magnetic dipole of the electron presents an electric dipole moment. But I think this second way is not so physically intuitive as the first.
In either case, you are correct that no (physically small) charge or dipole interacts with the field that it itself generates. Rather, a property of the electron interacts with a field produced by the nucleus.
A: An electron has a magnetic moment. When the electron moves it is perceived as having an electric dipole moment as well. This interacts with the nuclear electrostatic field.
A: It's in the name $\textbf{spin-orbit}$ coupling. The source of the magnetic field is the orbital angular momentum, whose origin was discussed by Andrew's answer. And the dipole that is interacting with it is due to the spin. So the field does not interact with the origin of the dipole as the source of the two are different. 
