Thanks a lot Niko, I was quite surprised to see these graphs here. They didn't gather much attention at my forum
The actual samples are taken every round hour - 12:00, 13:00, etc. So the graph slopes aren't quite accurate - especially around the lamp on/off time. I'm currently developing another firmware revision with settable log periods for accurate data acquisition - you will be able to collect data every 3 minutes for example.
The graphs itself actually don't show something extraordinary. It's just interesting to see how the theory works 'live'. I haven't seen such graphs on the net yet so I decided it will be useful to publish the data.