Dave that's a very good point. There is no harm in latching the data, at worse it would have no effect, and possibly be very benificial.
I have come up with the idea of using two 74LS374's instead of the 74LS244 for the onboard bidirectional data bus to the 8255. The LS374's latch their inputs on clock low to high so we can use the RD* and WR* signals comming from the GAL. Fortunately the 8255 also uses WR* and RD* (lows).
The SELECT_BOARD* will now go low for EITHER a port read or write. The S1oo bus sOUT & sINP status lines will keep the U108 & U110 gates open as long as possible and the RD* and WR* will latch the data. Probably not necessary to have data FROM the 8255 latched but may as well do both.
Could you take a glance at the circuit include here.
On Tuesday, February 10, 2015 at 9:23:42 PM UTC-8, yoda wrote:Hi John
My comment is I am not sure that the 8255 is not the root of the problems at higher frequencies. Having to have the data valid for a min of 30 nsec after pWR* is de-asserted is going to be a problem. The GAL will help some but I think you would be served better by latching the data on the board. This is the only board so far that I have not been able to get to work with the 68k board. I will certainly take a couple of boards and see if it works better but if you are going to do a tweak on the board at least latch the data on write to 8255 if you insist on keeping it 8255 based.
IDE Wr Data latch 2.jpg
Description: JPEG image