You are alone with this problem. Finding the right size for a PR can be really hard. Too big, and no one really wants to review it because it is too time consuming. Too short, and it fails to show the bigger picture. We struggle with exactly the same thing, with a bias towards too large PRs.

I am not even sure I can always agree with myself here. I tend to favor small PRs myself, but I also really dislike PRs that are only part of the solution. PRs should not only be about critiquing the actual code lines, it should also evaluate and discuss the overall architecture, security aspects, performance, readability, tests and probably other things as well. This is difficult to achieve with small PRs.

The solution is never to "just fix your process", in my opinion. The correct process depends on the people involved, the culture in the team and organization, the importance of the service in question, among other things. Finding the correct process is a never ending task :)