Convergence problems using fmincon with bounds

I'm using fmincon with linear and nonlinear equality and unequality constraints. The solution of the problem is bound to certain intervals which I implemented using the linear unequality constraints. I thought it would be better to use lower and upper bounds instead, but when I substitute the linear constraints representing the bounds by lower and upper bounds, fmincon requires much more iterations to solve the problem.

Why is the convergence so much slower when using bounds instead of linear constraints? Is there maybe an option which could help to speed up the convergence time?

Thank you!

Edit:

All values need to be positive. I just realized that using the lower bounds to constrain x to positive values increases the convergence speed dramatically. However, usuage of other bounds will slower the convergence again.