Coulomb (SR) calculated by manual and Gromacs are inconsistent

GROMACS version:gmx2020.3
GROMACS modification: Yes

In order to figure out the calculation of Coulomb in gromacs, I manually calculated the Coulomb-14 and Coulomb (SR) values of an ethanol molecule and compared them with the calculation results of gromacs.
In gromacs calculations, the box size is about 4nm. coulombtype = cut-off, rcoulomb = 2. The manual calculation method of Coulomb-14 is f=0.5, if i, j is 1,4. The calculation method of Coulomb (SR) is f=0, if i, j is 1,2 or 1,3; f= 0.5, if i, j is 1,4; otherwise f=1.

The manual calculation result of Coulomb-14 is consistent with gromacs, but the calculated value of Coulomb (SR) is inconsistent.

Gromacs calculation
Coulomb-14=-30.969759,Coulomb (SR)= 36.379906.

Manual calculation
Coulomb-14=-30.973276,Coulomb (SR)= 5.3962189

May I ask what causes this problem?

solved. Coulomb (SR) does not include Coulomb-14, so the energies of i, j=1, 4 should be removed when calculating by hand.

In the “GROMACS calculation” Coulomb (SR) does not include Coulomb (1-4), while it does in your manual calculation.
Best regards