dipole fdtd

I'm not sure how to describe oscillating electric point charges in FDTD scheme of the time domain Maxwell equations, but I would like to model the near-fields produced by an oscillating electric dipole and electric quadrupole. Then I will compare these solutions to analytical ones. I think a dipole could be just represented as a point current and then in one FDTD grid set a hard source as

Ex(i,j,k) = dt*sin(omega*t)/eps0

I'm not sure if this is correct. How about a quadrupole then?

That is more or less the standard method to implement dipoles.

For a quadrupole I would use four cells E(i+0/1,j+0/1,k) and insert
identical sine waves with opposite signs at two corners. I never tried it
but don't see any reason why this should not work.

Four cells? Wouldn't that mean four separate current densities? Or do you mean two cells with opposite signs? That would make more sense if a dipole is described with only one cell.