Got your point here. But the behavior was extremely undeterministic. I guess it was some oscillating phenomenon. Sometimes it was working and sometimes not and if you touched the pins with a finger you could somehow switch it off an on, really strange. I connected RFM69HW modules with cables and without changing the sketch everything is running perfect now. So I guess with those modules there must be something wrong. But I do not have an oscilloscope to analyze dynamic things properly nor I want to search for problems which obviously disappear with other modules.