Ian, I'm afraid my NatSemi data book only goes up to MM5290 (16Kx1 Dynamic)
but the '-3' variant is 200nS as is the MM5298-3 and it lists the '-4'
variant as 300nS. Unless I had no option I wouldn't use the slower chips
whereas the 150nS would do fine although without any real feel for the
tolerances of the chips, board design wait circuitry (if exists) or the
refresh controllers it is not really possible to say with any degree of
certainty.