This happens because in the wideband SINR calculation for the 3D channel model case, the antenna gain is not included in the pathloss. This causes the negative values. In the 3D model case the antenna gain is included in the channel matrix, so the wideband SINR is not a relevant parameter to this model.
Thanks for you reply, i tried to input very high tx power, however, SINR never exceed zero !, all UEs SINR now near zero accumulated in vertical dots, with throughput values widely different, which make no sense.
As I have mentioned, the wideband SINR for the 3D channel model case is not a relevant parameter so you should ignore it.
The reason why you get negative SINR values is because you get strong interference aspecially from two neighbouring sectors. As the pathloss does not include the antenna gain, each interferer radiates as it would have an omni-directional antenna.