My circuit is configured so that I can use a parallel port to directly access any 16-bit address that the microprocessor can address. 1 control pin enables the high-address latch (A8 to A15) and 1 control pin enables the low-address latch that is shared with the microcontroller (A0 to A7). ALE is overridden, and the processor is stuck in reset (intentionally) when programming takes place.

1 control pin determines whether programming takes place, or not. I call it "master reset". If this pin is not active, then no programming takes place, and the parallel port is effectively "disconnected".

Although I have no status ports in use, the way I verify that everything works is that I send the code in programming mode, and then watch the CPU go to work.

What I noticed is that in order to successfully program a byte into the eeprom, I had to write a byte to it (bit set low) for 100 or so microseconds, and then I set the bit high, but it must remain high for at least 1/2 a second.

Once again, timing can control alot.

my design does not involve tying PSEN and RD together with an AND gate, because I think the system won't work due to excessive track lengths.

I also use single-sided boards, and the more parts I add, the longer the wire lengths (most likely).