Giulio,
I have not investigated your model thoroughly, but I see a few possible causes. In the first model, "prova" there does not seem to be any forces or prescribed displacements. As far as I can see, you have put a constraint on the surface that is also a contact surface. This is not something I would have tried. In addition, there is a large initial gap in your contact. This can be problematic in a contact analysis, especially if you, as in your case, do not include large deformations.
I'm not sure how the structure is loaded/constrained in reality, so it is difficult for me to anticipate where the capsule will contact the "PM1A30019" component, and how those contact forces are transmitted to constraints.
Suggestions:
-model parts/assembly so that there is minimal initial gap in contacts.
-make sure locations for constraints and contacts are well defined, and don't mix them, i.e. put contacts and constraints on the same surface.
-Include large deformations in your analysis (necessary in statically indeterminate models with contacts) - but this does not allow weighted links. You will have to model your link differently.
-if possible, use symmetry, so as to eliminate one possible direction of motion. Provided you do not want to analyze "out of plane" motion of the capsule, then there is a symmetry plane at the centre of the PM1A30019-part.