I was playing with modifying the supported 3COM NICs to support larger than 128kB chips.
It's been a success so far, and seems to be functioning well with 256kB chips. I haven't tried with 4Mbit/512kB yet, and I don't remember the last time I've seen an 8Mbit [DIP] one. I need to add 28pin DIP and 32pin PLCC sockets yet - that should cover the whole 28F/29F family.
It will need some work for the 512kB chips to work. On <512kB chips, pin 1 is either Vpp or /RST, so it has to be high. On the 512k chips, it's A18, so it has to change. For now, I have a pullup on it, and a switch to the bankswitching latch(A18). So when the switch is open, there is no support for 512k chips, but smaller should be fine... I should be able to take care of this in software though. I guess just have it turn the line on at init, and drop it if the probe determines a 512k+ chip.
There's only a few bits of the latch used, so the others can be used for general purpose outputs. Right now I have two of the bits tied to LEDs, and the code set to turn on one or the other, depending whether it is in a write or read routine. With some hardware work, off the other outputs - I should be able to switch higher voltages to Vpp for the 27xxxx EPROM, I hope, but that's for another day.
Here's a photo of the current card. Not very pretty, but once I get the kinks out, I should be able to get some (much smaller) add-on PCBs made, that will just plug into the stock plug on the NIC.
http://electrontube.org/3com_mod/etherlink_mod.jpg
Schematic, as is: http://electrontube.org/3com_mod/current_sm.png
Schematic (minimalistic version): http://electrontube.org/3com_mod/minimal_sm.png
Code changes: http://electrontube.org/3com_mod/3comhack.diff
I think for the demo board, Maybe I'll add another decoder and an 8255.. then I'll have 3 bytes of GPIO - for whatever.
Schematics, pictures, etc, related to this are CC-SA if you want them for anything. Anyways, any ideas, comments, criticism, etc, would be welcome :-)
-Mark
Oh yeah - one thing I've noticed, after read, write, or erase, the shutdown routine is called, but it *isn't* called after a probe. (on 3com NICs at least). Is that correct behaviour..? I noticed this, because the latch leaves LEDs on after a probe, but not after R/E/W (the shutdown function clears the bankswitching latch).