Optimal PV based EV charging is performed either by predicting or by measuring PV output. However, due to the uncertainties associated with the variables involved, both the charging methods are not only disposed to significant voltage limit violations but also may cause PV energy harvest reduction. Hence, a methodology is proposed in this paper, which utilizes the measured PV output of a given sample and the supplied historical ramp to predict the PV output of the immediate next sample using a non-iterative method. The charging rates of the EV population are, subsequently, adjusted in the interval between the successive samples based on the predicted PV output with the help of a proposed SOC based charging strategy. The proposed methodology has been tested at the University of Queensland (UQ)'s solar and parking facilities coupled with its electric grid. The results show that aside from reducing the probability of voltage limit violations (PW), the proposed methodology can increase the PV energy harvest. Moreover, it is cost-effective as compared to the conventional method such as onsite battery energy storage deployment. (C) 2017 Elsevier Ltd. All rights reserved.