Measurement of very low frequency sources is rather difficult to perform because very few testing rooms remain anechoic in the 10-100 Hz frequency range. To overcome this problem, the field separation method has been proposed a few years ago. This technique consists in measuring two acoustic quantities on a closed surface surrounding the tested source. To take advantages of spherical harmonic functions, the measurement surface should be spherical. The acoustic quantities can either be acoustic pressure measured on two concentric (half-) spheres or pressure and velocity measured on a (half-) sphere. Then, by using spherical harmonic expansions, contribution from the subwoofer is separated from reflections on the walls of the testing rooms to recover half or free space conditions. This paper focuses on the choice of the measured data set. Simulations are performed using p-p and p-v data approaches. The effect of microphone spacing for the double pressure layer formulation will be investigated. Finally, measurement results obtained on different subwoofers, i.e., frequency response and directivity, will be shown to highlight simulation results.