Experimental errors in neutron residual strain imaging are often underestimated, mis-understood or simply ignored. The choice of beam defining optics can play a large role in the magnitude of these errors. In this paper we show that the use of a traditional slit beam defining system with large specimens can lead to several sources of errors. Large specimens usually require necessarily large distances between the slits and the instrument reference point (gauge volume). At large distances, the slit system can lead to significant underestimation of the residual strain magnitude as well as poor definition of the gauge volume. We show that for large specimens, the use of radial focussing collimators reduces these effects, leading to significant improvement in measurement reliability. Specifically, we show application of radial collimators at the SALSA instrument of the Institut Laue Langevin neutron source in Grenoble, France.