The article describes a control architecture for solar furnaces where active cooling is employed to improve the tracking of a time-varying temperature reference. This capability is important during the decreasing phase of the temperature reference where heat loss must be increased. The results of two different control methodologies, exact linearization and model predictive control with integral action, are shown with active cooling that is done in coordination with the command of the shutter which adjusts the solar incident power. The controller parameters are computed from the temperature dynamics which is identified off-line from collected process data. This approach is used to avoid online adaptation mechanisms of the controller parameters, that may cause stability problems during the controller startup, and may melt the testing material sample. The novelty of the present work is to present a control architecture that coordinates the operations of the shutter together with the application of active cooling. This methodology improves temperature reference tracking and increases the usability and the operation of solar furnaces.