“??”, you’re right. at the moment, I was confused with the calculation of expectation of two indep, r.v.s. So the correct one is (assuming x & y are stat-indep.):
fz(z) = conv( afx(z) , bfy(z) ) = int{-inf,inf}fy(t)fx(t-y)dt.
Pr(z<t) = int(-inf, t)fz(z)dz.