Regarding coulumbic cutoff in pair_style lj/charmm/coul/long

Hi
I am trying to run a simulation of a negatively charged molecule with water and sodium ions.
The system is periodic in the x and y direction and I use reflecting walls to restrict the system in the z direction.
The simulation boxsize is [xlo,xhi,ylo,yhi,zlo,zhi]=[-0.1,15,-0.1,15,-0.1,70]

I use pair_style lj/charmm/coul/long and 2D ewald summation for the long range electrostatics with the following parameters (in real units ) :

pair_style lj/charmm/coul/long 10.0 11.0 15.0
kspace_style ewald 1.0e-4
kspace_modify slab 3.0

According to the lammps documentation : lj/charmm/coul/long args = inner outer (cutoff)
The third parameter is the Coulombic cutoff.

What one would expect is on changing the Coulombic cutoff, the individual values of ecoul (pairwise electrostatic energy) and elong (the long range contribution to the electrostatic energy) would change but ecoul+elong should be a constant.

But when I run the simulation with different values of the cutoff I get different values of the electrostatic energy (ecoul+elong) for the same system!
Increasing the Coulombic cutoff, gives extremely high ecoul values.
I am not sure why that happens, what am I missing here?

Also is there any restriction on the Coulumbic cutoff and is it related to the simulation box size in anyway ?

Thanks
Sweta

Paul can probably look at this. Your description
is correct - the total Coulombic energy should
be insensitive to the choice of Coulomb cutoff. Maybe
there is something different about the 2d slab case,
but I don't think so. Can you try the same thing
with PPPM to see if the behavior changes?

Steve