Hi Larry,
thanks again for pointing me in the right direction, it was in fact a mistake in my setup, with replacing the AQ I made some rewiring and the two wires of the front motors which were previously crossed (A was to M10 and B to M3) are not crossed anymore, but I used the same mixing table again, thus A and B were swapped which explains, the behaviour a small change on roll, caused it to try to react and instead increasing the angle on roll.
I have now swapped the ports for those two motors and now this trend to flip on roll disappeared.
The maiden flight can come, most probably next tuesday, as I need some free room for it and my graden is rather small.
Next step will be to make the connector for the current sensor.
Confidence in the system and my setup is back.