I have a repository that handles an entity with a lot of information that is filled using different tables, but the same connection.
To encapsulate things better, I have created "subrepositories" that fill parts of the entity.
Is it against the pattern or perfectly right?
To answer your question, I think it makes sense to make a clear distinction between the data mapper that maps tables to objects and the repository:
A repository
mediates between the domain and data mapping layers
Whereas a data mapper
moves data between objects and a database
So if I understand your question correctly, you are actually talking about the data mappers.
Yes, creating data mappers that can be composed into larger mappers is a good approach. It promotes separation of concerns and makes the individual mappers simple.
Repositories on the other hand should be used per aggregate. Create one repository for each aggregate in your domain. This also means that the repository only works with whole aggregates, and not with entities. To achieve its goal, the repository can use the (composed) data mappers. This is how the two concepts relate.