I have tried different timings, but non works very well. Too small integrating makes the reading(s) unstable. I'm aware of the saturation problem, so I don't think that's why my readings are wrong.