Keep the Beam delay fixed as 0.5, and beam width will always
be 90% of RF pulse - Beam Delay.
This way we always have valid beam pulses and the beam end
never falls out at the same moment than RF end