Problem in converging scf when spin direction of all atoms fixed

[megha]

Hi,
I have a system whose unit cell has 28 atoms ( 4 magnetic atoms and rest non-magnetic). I fixed spin directions (theta and phi) for all the 28 atoms. My system never reaches to convergence. (Some parameters I put are: initial scaled spins as 0.2 for non-magnetic atoms and 1 for magnetic atoms, fixed_spin_energy=5*eV) Please suggest possible solutions.
Thanks

[Anders Blom]

In general noncollinear calculations can be hard to converge, it may be better to first do a collinear calculation to get a better starting guess for the populations and spins, then run the noncollinear model based on that.