Virtual acoustic environments are becoming a widespread tool in hearing research. Free-field reproduction allows the design of more realistic experiments, where participants can turn their head and move around without the need for individualized HRTFs or head tracking. A limitation of commonly used sound field synthesis techniques, in particular Ambisonics, is their decreasing accuracy with increasing distance from the center of the loudspeaker array. While participants can move around in the reproduced sound field, it is not possible to control the sound field they receive at off-center positions to the degree required for hearing research. In this work, we consider a two-dimensional 36-loudspeaker array, where the loudspeakers are typically equalized in the center of the array. We investigate the possibility of moving the Ambisonics sweet-spot by equalizing the loudspeakers at different points inside the loudspeaker array. Simulations show that at least 50% of the sweet-spot size is preserved when moving the equalization point. An adaptation of the relative loudspeaker positions for the Ambisonics decoding reduces the parallax shift observed in Ambisonics systems.