At present, the BeiDou system (BDS) enables the practical application of triple-frequency observable in the Asia-Pacific region, of many possible benefits from the additional signal; this study focuses on exploiting the contribution of zero difference (ZD) ambiguity resolution (AR) to the precise point positioning (PPP). A general modeling strategy for multi-frequency PPP AR is presented, in which, the least squares ambiguity decorrelation adjustment (LAMBDA) method is employed in ambiguity fixing based on the full variance-covariance ambiguity matrix generated from the raw data processing model. Because of the reliable fixing of BDS L1 ambiguity faces more difficulty, the LAMBDA method with partial ambiguity fixing is proposed to enable the independent and instantaneous resolution of extra wide-lane (EWL) and wide-lane (WL). This mechanism of sequential ambiguity fixing is demonstrated for resolving ZD satellite phase bias and performing triplefrequency PPP AR with two reference station networks with a typical baseline of up to 400 and 800 km, respectively. Tests show that about 90 % of the EWL and WL phase bias of BDS has a consistency of better than 0.1 cycle, and this value decreases to <80 % for L1 phase bias for Experiment I, while all the solutions of Experiment II have a similar RMS of about 0.12 cycles. In addition, the repeatability of the daily mean phase bias agree to 0.093 cycles and 0.095 cycles for EWL and WL on average, which is much smaller than 0.20 cycles of L1. To assess the improvement of fixed PPP brought by applying the third frequency signal as well as the above phase bias, various ambiguity fixing strategy are considered in the numerical demonstration. It is shown that the impact of the additional signal is almost negligible when only float solution involved. It is also shown that by fixing EWL and WL together, as opposed to the single ambiguity fixing, will leads to an improvement in PPP accuracy by about 20.6 % on average. Attributed to the efficient resolution of EWL + WL within about 2 min in Experiment I, the 0.5 m level positioning can be achieved in 10 min for both horizontal and vertical, compared to 50 min for horizontal and 30 min for vertical by the NONE/EWL/WL fixed solution. While, for Experiment II, the improvement in the convergence can only be seen for the horizontal as the TTFF takes about 40 min for EWL and WL to be resolved.