This paper investigates the modelling accuracy of small-scale solar energy systems, in particular the effect of changing the time resolution. Accurate modelling of solar energy systems is important in designing cost-effective solutions and in research into micro-grid functionality. Currently, most modelling uses a resolution of one hour when simulating the energy in small scale renewable energy systems. Within each hour, these models take an average of the load demand and solar generation values. An hourly time resolution is ignorant of the large variations in a daily load profile that result from high power appliances and also misses short-term variations in solar generation. Few researchers using hourly modelling consider the implications in terms of the simulation accuracy. In this paper, 3 load profiles are modelled with grid connected solar systems, one with and one without a battery, totalling 6 models. Each model is simulated for 4 seasons of the year, modelled at hourly, 5 min and 1 min time resolution. The results show that the hourly model was a poor predictor of battery behaviour, over-estimating the battery state of charge (SoC) by up to 10%. It is also shown that, for systems without battery, the quantity of energy exported and self-consumed changed by an average of 5% and 8% respectively. The study concludes that increasing time resolution from hourly to 5-min resolution in energy models would be advantageous, increasing their accuracy in terms of battery behaviour and predicted self-consumption of solar energy. In particular, modelling at 5-min instead of hourly resolution shows the full benefits of installing battery storage with solar systems.