To the best of my knowlegde, you can not set the density matrix constraint from the scripter, since it is an new experimental feature, but I think it works really great!
So you have edit it manually in the script - either by using the script editor or an external editor.
In principal there is no limit on how high you can set voltage, however it depends on your machine, as
the voltage goes up, the electric field becomes stronger, and the SCF might become unstable if the the electric field becomes to strong, however this can be compensated by increasing the length of the system in the Z-axises. Therefore the highest voltage you can apply, becomes an issue of the computer size rather than the algorithm ( higher voltages will also increase time of the contour integral, but it scales perfect with the number of CPU).
The largeste voltage I have ever succeded to converge, is a 20 V bias calculation across an MTJ junction, but the system was quite long along the z-axis, and it required some work in order to determine the sufficient size in order to make the system work.