This study investigates the sound mechanism of unexpected tonal sound emitted from a rotor-only axial fan ingesting air from a seemingly free atmosphere in a large plenum. Even if a plenum fully satisfies the current standards with respect to the unblocked space upstream of the fan, blade-passing frequency (BPF) tones with a considerable variation of amplitudes can be observed. The tonal sound from an isolated axial fan rotor and its underlying mechanism has hitherto not been studied. Besides the standard aerodynamic and aeroacoustic performance tests, sophisticated local flow field measurements are conducted employing hot-wire anemometry and flush-mounted blade pressure transducers. Navier-Stokes based numerical methods with subsequent acoustic models and the Lattice-Boltzmann method are applied to gain further insight into the details of the flow and acoustic field. It is shown that slow but large-scale room airflow is induced in the inlet plenum which eventually produces a spatially and temporally inhomogeneous inflow ingested by the fan. This distorted inflow interacts with the rotating fan blades, causes fluctuating blade forces and eventually tonal sound. A practical conclusion is that the state-of-the art measurement procedure of acoustic fan performance can be erroneous. Accurate, reproducible and test rig independent acoustic performance data require a careful management of the inflow to the fan. Typically, inlet plenums required for standardized fan noise measurements are anechoic chambers. Large-scale room airflows were observed in a variety of those chambers, which substantially differed in shape and size. None of the chambers provided an inflow quality as obtained by utilizing a hemispherical flow conditioner in the immediate inlet of the fan.