rdrast has said the same as one of my previous posts, I agree just make each bit in the N register 0-7 equal the relevant output i.e. 8-15 (assuming that the ML has the ability to use discrete bits from an integer file I have not installed my licences so can't check at the moment).