3D LES - Error: Divergence detected in AMG solver
 

I am working on a 3D LES and after almost 30 iterations I constantly have the continuity residual blowing up.
The message I get is: "Error: Divergence detected in AMG solver".

I was initially using a second order implicit scheme and relaxation factors of 10^(-6).
I tried switching to first order implicit and relaxation factors of 10^(-4), but still the same problem occurs.
Any ideas?

I should also mention that my mesh is structured and that I am using periodic boundaries. By reading other threads I understand that this might also be a source of the problem.