Oops, I do too have an upper bound. So I claim 0 < C1 < f(N) < C2 < 90 degrees for all large-enough N, and would be interested in whatever the best C1, C2 you can cook up, would be, and also in specific values of f(N).