The neutron production in ASDEX Upgrade (AUG) neutral beam injection (NBI) heated discharges is dominated by beam-target fusion reactions. Hence, the neutron rate (NR) and energy distributions are footprints of the fast ion distribution. This motivates to establish a reliable neutron rate calibration. Comparisons at AUG between the experimental NR and the one predicted by the TRANSP code show systematic variations from campaign to campaign. Potential reason for this is the delicate absolute calibration of the neutron detectors. Therefore, a different calibration technique was performed, enabling longer acquisition time, uniform geometry, better statistics and thus less uncertainty. A toy train carrying a radioactive source (238Pu/B) over two radial positions on the equatorial plane shows a periodical NR on the epithermal 3He neutron detector. The calibration results are compared to a neutron transport simulation using the Monte Carlo (MC) code Serpent. Preliminary comparisons for one source position on the outer railway track show a discrepancy factor of about 130 in the position of least material inside the simulation, in the direct line of sight to the detector. For a better understanding of these results, two additional measurements were performed. The results were again compared to a detailed Serpent simulation. This paper describes the calibration set-up for the neutron measurements in AUG, provides a brief simulation background on reaction rate estimations and a survey on the comparison between the measured and calculated neutron rates.