The Deep Underground Neutrino Experiment (DUNE) is a long-baseline neutrino-oscillation experiment aiming at measuring CP-violating phase and neutrino mass ordering. It uses the Liquid Argon Time Projection Chamber (LArTPC) technology with four giant modules of 17-kt each. The recently proposed vertical drift (VD) concept has been selected for the second module. In the VD concept, a suspended cathode separates the module into two equal volumes. The electrons drift towards the anodes located at the top and bottom of the TPC in a 500 V/cm electric field. The anodes are made of two double-sided perforated PCBs etched with strips at different orientations. Those strips will record the passage of the electron clouds via induced and collected signals. The anodes of the top volume will be equipped with accessible electronics, while the anodes at the bottom of the TPC will have embedded electronics immersed in LAr. In order to test and optimize the VD concept, cosmic data were taken using different demonstrators of increasing size hosted at the CERN Neutrino Platform. This poster will focus on the data acquired over the last months with a 3×3.4 m2 anode plane over a drift distance of 25 cm mounted with the two electronics. The performance of the system was analysed using LARDON, a fast python-based software written for 3D-track reconstruction in LAr. In particular, results of the noise characterisation and calibration of the system will be discussed. The signal-over-noise ratio measured with muons will be presented. Finally, first results about the induction-signal studies, from simulations and data, will be showed.