[en] The inherent numerical properties of large thermal-hydraulic system codes may be usually verified by means of simpler codes, dealing with selected cases of restricted practical application. The authors provided an example of such methodology in their previous work, in relation with the RELAP5 series of codes. The problem considered was the single-phase, natural circulation flow in a simple loop, under unstable flow conditions. Neutral stability curves were derived for the system. The effect of the number of nodes in the expected behavior of the system was consequently analyzed. The ad hoc codes previously developed have been further refined, to analyze de problem under two approaches: a.-a nodal one, based on a finite-difference approximation and, b.-a modal one, based in a modal decomposition of the governing equations in Fourier series. Theoretical values for the steady state parameters are obtained for both approximations. The nodal code was used as the standard approximation and different schemes of different order have been used. Then, the effect of the number of nodes in the damping of the system was quantitatively determined. The modal code, with the number of modes ranging from 30 to 100, was used as an approximation free of numerical diffusion. In this case, the energy equation was solved considering also a constant diffusion term, allowing a partial simulation of the numerical diffusion of the nodal approximation. In this way the analysis allowed the modal analysis to be performed including an average value of the diffusion arising from the up-winding in the nodal solution. Results show that the inclusion of this diffusion reasonably accounts for the damping of the solution, allowing a qualitative recovering of the nodal behavior. System non-linearity naturally precludes the exact coincidence of the results obtained. A similar analysis may be used to assess the effect of the number of nodes of a given discretization on the dynamics of more complex thermal-hydraulic systems. (author). 6 refs., 5 figs