Fine tune Q_new calculation, improve initial condition of q_lpd_old_exc via prev_Q_new.