Then again, if that is all true, a diode on the exciter wire such that power could go only to the regulator and not the other way should solve the problem. So maybe its an open diode in the alternator?