Hard stop modeling - two Integrator blocks vs. Integrator Second-Order block

Hi,

I am attempting to understand what stands behind Integrator Second Order block, which makes it possible to model second-order differential equations with limits accurately (the description of the problem is here: https://blogs.mathworks.com/simulink/2014/01/22/how-to-model-a-hard-stop-in-simulink/). Are there numerical issues which cause that two first-order Integrator blocks give uncorrect results (nonzero 1st derivative values when going into saturation)?