Recent progress in radiometry allows the development of diffuse fraction models based on data stored at high-frequency sampling. At the present only a handful of such models exist in the literature. In this paper, we present a new and simple model for estimating diffuse fraction developed on 1-minute resolution data. The new model (PB) is extensively validated on data from the Baseline Surface Radiation Network, for both diffuse fraction (k(d)) and direct normal irradiance (G(n)). The results show that PB performs well compared to other similar models, despite its structural simplicity. k(d) is estimated with the best accuracy in locations with temperate climate (TM), the average (median) nRMSE of the tested TM stations being 18.38% (17.79%). The G(n) estimates are found to be most accurate at arid locations (AR), the average (median) nRMSE over all tested AR stations being 17.29% (16.96%). We find that the performance of the models varies strongly as a function of the climate zone or when we switch between estimating diffuse and direct normal irradiance components, possibly changing the accuracy-based rankings among models, a fact which should be taken into consideration by users when selecting appropriate models.