Diode matrices - next we'll be wire-wrapping it.

Mike, unfortunately you have some ingrained misconceptions. We've all tried to tell you this and offered compelling evidence. Please read the following carefully - don't try to argue the case as you did with my lcd code - it works as I have implemented it also it is not a new idea - I'm not the first person to do it nor the last.

Please consider.....

You have an eeprom, by just adding some simple logic you can divide the eeprom address space into two (equal or unequal) spaces. One has the bootloader code that stays there, the other has your application code that you can reprogram at will. One port pin can swap between these two spaces. On reset the port pin is high, so that selects the bootloader space. The bootloader can use various means to detect whether you want to load new code or simply execute the application code by pulling the port pin low and jumping to location 0(this needs to be done by code in ram). You could use a switch on a port pin for this. This is how I do it on my boards. Thus you only need to program the eeprom out of the board once to install the bootloader. Unless the board gets hit by lightning or the power supply fries everything, the bootloader will most likely remain. Some eeproms have lock mechanisms that make losing the data even less likely and you could install a switch on the eeprom write signal if you're really paranoid.

The actual method you use for the code loading is up to you. For me I do it via the serial port using the XMODEM protocol and I use Hyperterminal on my PC to do the file transfer. A 16K file takes about a minute to load at 19200baud.

No magic invovled, just basic logic and some software. I can even post the bootloader if you're interested. Its small and in assembler.

I used a PLD for the logic, as nowadays having a handful of random logic is just a pain when doing a pcb - you've been getting some of this experience of late.