FMOLS and DOLS panel estimation in eviews8

[Imene]

Dear eviews users and moderator, I have estimated an FMOLS an DOLS equation (Pooled and Grouped) on cointegrated series, my coefficients are efficients but my errors seem correlated according to the CORRELOGRAM (Qstatistic and squared residuals) and the Durbin watson is far from 2.
So how can I check robustness in co-integrating regressions estimated by FMOLS, DOLS on Eviews 8? Can I accept my models?