It also depends on how many miners there are and how much of the overall hashrate is comprised of high power ASICs.  If only 10% of the total hashrate (currently 15TH I think) comes from these ASICs, the differential in difficulty will not be a direct correlation to their individual hashing power, but as a percentage increase of total power on the network.  1.5TH would be about 500 A5 Dashmasters, if the batch is that size, so that's 10% of the network.  Add another 750gh from an approximately comparable size Antminer D3 batch size since it's 1/2 as powerful.  So that's 2.25TH of the total new network hash power total of  17.25TH on the network after the first batches of these new models goes online (unless the batches are MUCH larger, which seems unlikely).  So what we're really looking at difficulty wise is a 14% increase in network hashrate give or take a couple percent, so I would expect a differential increase comparable to the increase in total net hash, rather than one directly correlated in a 1:1 ratio based on how much more powerful these particular devices are.

There are also other factors to consider, such as miners leaving the network because the difficulty goes up too high for their profit margins to hold (vis a vis Ethereum), and that will reduce the overall network capacity, slowing the rate of difficulty increase for a while at least.