Effect of tight convergence on band gap

[Kalpna Jain]

Dear sir,
I make my incar file for geometry optimisation of a one dimensional polymer like this:
SYSTEM=POLYMER
PREC=Normal
IBRION=2
EDIFF=1.0e-06
EDIFFG=1.0e-05
INIWAV=1
ISTART=0
ICHARG=2
After that I submit the job for band structure by using the CHGCAR CONTCAR file and perform a self-consistent run using 10 KPOINTS in line mode in direction of polymer extension.

I get the almost correct value of band gap. but the dE value in OSZICAR file was gone only upto 0.392E-02 whereas I had given the value of EDIFFG 1E-05.

To correct this problem, I use the incar file with NSW=200 & ISMEAR=0 AND NSW=200 & ISMEAR=1 respectively in two different jobs to see the results.
the value of dE goes to E-05. But the band gap is incorrect in both cases.
I am finding difficulty to understand the reason for this.

Is tight convergence bother the band gap or there is any other mistake.
Please guide me.

[graeme]

Every calculation should be converged with respect to the input parameters.

[Kalpna Jain]

dear sir,
please explain this in detail.
why tight convergence is giing wrong band gap .