Dear @phialberta,
The nbdbuf is useful if you have bands than the occupied ones, but if you set nband to the number of occupied ones, then you should not use nbdbuf as it will disregard the last occupied bands then, which you want to take into account in the residual of the SCF.
Regarding the negative d, couldn’t it be negative, are the elastic constants negative too?
You have to check if the relaxation is well done with the pressure and check if your calculation is converged, see this post:
Best wishes,
Eric