The high penetration of electric vehicles (EVs) charging stations (CSs), together with the progressive availability of distributed energy resources, increases the risk of grid overloading and power-quality degradation. Real-time control has been advocated by the recent literature as an alternative to costly grid reinforcement. Performance of most of the existing control methods is assessed via simulations, and real-field validation is rarely performed. We present the experimental validation of a recently proposed real-time control method for EVs CS. This method works at sub-second scale and allows the CS to adapt to the rapidly changing state of the grid caused by highly volatile energy resources, such as photo-voltaic (PV) plants. The method fairly allocates the available power, proportional to the EVs needs, while minimizing the EVs battery wear due to frequent charging power variations. It tracks an aggregated power-setpoint that comes from the main grid controller by solving a mixed-integer optimization problem. Our main goal is to show that the method works in the field, i.e., it can control the charge of commercial EVs that are connected to a real grid through a CS. The field validation has two challenges. The first one is to study the real-time capabilities of the method and by analysing how fast it computes the control power-setpoints. The second one refers to the handling of the non-ideal response of EVs to the control power-setpoints due to implementation and reaction delays and inaccuracies. The experimental results demonstrate the performance of the method and show that it can be deployed in the field.