Check this thread for the details on how the solver treats multiple steps: Understanding multistep analyses
Wasn’t the BC that you deactivated in the second step also constraining the model ? Even if you release just one DOF and added force there, load control may cause convergence issues.
What you did is similar to the common approach of establishing contact with displacement control in the first step and then switching to force control in the second one. This involves some special considerations mentioned in the last point of the thread linked above as well as here: Hertz contact with force control - CalculiX (official versions are on www.calculix.de, the official GitHub repository is at https://github.com/Dhondtguido/CalculiX).
CalculiX has the *RESTART feature for that, but it’s not supported in PrePoMax and has to be used by submitting the model outside of it with some input deck modifications: How to use the *Restart command in my case? - CalculiX (official versions are on www.calculix.de, the official GitHub repository is at https://github.com/Dhondtguido/CalculiX).