Peter connected his display as if it was memory and used the A8 signal for the c/d on the display and the rd and wr signals from the cpu connect to the lcd. Thats why he could just read or write to a memory address - the cpu itself generates the correct signal sequence. Since you did not do it this way, your code needs to generate the memory cycle sequence. Thats why there's a few more lines in the code I posted. Look at the datasheet for the memory cycle timing - that should explain all. No magic involved.

The output of the bintohex program is what I would expect - it is called Intel Hex. Maybe you need a different bintohex program!