I removed these by snipping the pins, removing the body, then desoldering the pins one at a time. I clear the holes with a solder sucker, solder wick and a pin drill. I fitted a pair of sockets and 74ACT244 chips, hoping the better speed might help.

Alas it did not, and putting 74LS244 chips back in there did not restore functionality. Reset just produces the tiled pattern of the uninitialised DRAM.

There are various possible explanations. I may have damaged the board or left some solder swarf that is shorting a pin. I had a good look and only found one piece of loose swarf.

At this point I am getting a bit fed up with the amount of time this is soaking up. More time than I can spare, or it is really worth. I have reverse engineered the PAL chips which was my original goal. The new PCB is not a 'faithful copy' of the original, the new board has been autorouted with thinner wires and smaller via holes. Perhaps making it more fragile.

I decided to do a quick connectivity check that all the data pins connect to the CPU. All good except for the D0 line. I heated U30 pin 18 hoping that the solder would wick to the other side of the board and its tracks. This restored previous functionality. 

Swapping the LS for ACT did not cure the fault, in fact it made things worse. So I returned to using LS chips.

The only difference between these chip families is speed, which suggests timing is the issue.