The well-known magic numbers of stable nuclei can vanish and new ones appear in the neutron-rich nuclei far from stability. The GXPF1 effective interaction predicts the appearance of an N = 32 subshell gap and an N = 34 shell closure in Ti and Ca. The investigation of these shell gaps developed into a question of the appropriate effective charge in the pf shell. The trend of the measured transition matrix elements in the neutron-rich Ti isotopes is reproduced with effective charges derived from an analysis of isospin analogue states in ⁵¹Fe and ⁵¹Mn, e_p ≈ 1.15 and e_n ≈ 0.8, that differ from the standard effective charges, e_p = 1.5 and e_n = 0.5. The proton and neutron effective charges are mixed in the case of Ti, and to separate the two components the neutron effective charge is determined via the intermediate-energy Coulomb excitation of ⁵⁰Ca. To reduce the uncertainty in the measurement a model is developed to simulate in-beam response functions for a position sensitive NaI(Tl) γ-ray detector. The effective charge is found to be e_n = 0.77(13), indicating that the enhanced neutron effective charge derived in the upper pf shell is applicable in the lower pf shell.