Divergence of nonself-consistent HSE06 calculations

[zzhlax]

Dear vasp users,
Recently, I used HSE06 hybrid functional to calculate the difference charge density of hematite (the difference between self-consistent and nonself-consistent charge densities). For the self-consistent calculation, it is ok. But the nonself-consistent electronic calculation can't be converged at all whenever the parameter TIME changed (from 0.4 to 0.01). My INCAR and POSCAR files are given below. Any advice will be highly appreciated.
zzhlax

INCAR:
SYSTEM = Fe2O3:AF
PREC = Accurate
ENCUT = 400
IBRION = -1
NSW = 0
EDIFFG = -0.01
NELM = 60
NELMIN = 2
EDIFF = 1.0e-05
VOSKOWN = 1
NBLOCK = 1
ISPIN = 2
MAGMOM = 6*0 -5 5 5 -5
INIWAV = 1
ISTART = 0
ICHARG = 12
LWAVE = .FALSE.
LCHARG = .TRUE.
ADDGRID = .FALSE.
LVTOT = .FALSE.
ISMEAR = 0
SIGMA = 0.1
LREAL = .FALSE.
NPAR = 72 ! data distribution, compute speed, usually set as number of nodes.
ISYM = 0
# Hybrid functional
LHFCALC = .TRUE.
ALGO = Damped
TIME = 0.2
HFSCREEN = 0.2
# ENCUTFOCK = 0
NKRED = 2

POSCAR:
Fe2O3:AF
1.0
5.4265999794 0.0000000000 0.0000000000
3.0907860352 4.4603844028 0.0000000000
3.0907860352 1.6185827201 4.1563468333
O Fe
6 4
Direct
0.444400012 0.750000000 0.055600047 !O
0.944399953 0.555599988 0.250000000
0.750000000 0.055600047 0.444400012
0.250000000 0.944399953 0.555599988
0.055600047 0.444400012 0.750000000
0.555599988 0.250000000 0.944399953 !O
0.144659996 0.144659996 0.144659996 !Fe
0.355340004 0.355340004 0.355340004
0.644659996 0.644659996 0.644659996
0.855340004 0.855340004 0.855340004 !Fe

[admin]

1. In the DFT calculation remove LWAVE=.FALSE.
2. Start the non-scf HSE calculation from WAVECAR (ISTART=1).

[zzhlax]

1. In the DFT calculation remove LWAVE=.FALSE.
2. Start the non-scf HSE calculation from WAVECAR (ISTART=1).

Thanks very much. Should the DFT calculation also be nonself-consistent?

[admin]

First step is a standard self-consistent DFT calculation producing scf wavefunction.

[zzhlax]

First step is a standard self-consistent DFT calculation producing scf wavefunction.

OK, thanks.