This paper presents a new version of the fast decoupled loadflow in that a more broad range of power systems can be solved. The key lies in the different way in which the resistances are ignored and in a different iteration scheme. In the standard algorithm the resistances are ignored while building the B-prime loadflow matrix; this paper shows it is preferable that the resistances are ignored in the B double-prime matrix instead of the B-prime matrix. For normal test systems there is hardly any difference in the number of iterations, however, the new algorithm iterates faster if one or more problematic R/X ratios are present. An iteration scheme with strict successive P and Q iterations prevents cycling convergence behaviour which can be found in some low voltage systems. In this paper the advantages of the new version (to be indicated with the BX scheme) are demonstrated with runs on IEEE test systems with both uniformly and nonuniformly scaled resistances or reactances. R-scaling up to 3 is always possible, sometimes values up to 5 can be used. X-scaling of at least 0.1 is possible without losing convergence and with iteration counts which are significantly lower than with the standard scheme.