Thank you, Kevin. IRIS isn't the best to estimate magnitudes, but it works fine in this task. And it is free.
The quality of the photo, and some procedures that I didn't, like dark, flat and bias subtractions are needed to get good precision.
It works better if you use to register the variation of magnitude of a star, were the true value isn't important, since you want the amount of the variation. And this is just that I need, by now.