I finally managed to solve the mystery and guess what, I was right that the master was changing its Modbus adress (wasnt hard to find). I fact I told myself that if the two inverters keep the same adress on the same bus there ould be a conflict and I would have no response.. So I tested address 5 and it worked. Why 5 ? well already why do they put a fixed adress 4 in thier thing.
In short for those who like me would be faced with this problem you have the solution.
And I would to thank you Coulomb for your explaination it makes things more clear.