I have generated a phase space just below the linac head in PRIMO (PENELOPE variant, https://www.primoproject.net/primo/main). I then took this as the source in a simulation with simple geometry to get a percent depth dose. In Geant4/Gate and in penEasy the PDD produced is incorrect but when used as the source in PRIMO I get a depth dose that agrees with water tank measurements. The position of dmax is the same for all curves. The geometry is simply 1 m of air and 40 cm of water to represent standard water tank measurements. Attached are the depth doses, does anyone have any suggestions?