Bil Herd wrote:
> What kind of app is this targeting?  Cartridge, C64, or damm near anything?
> Is it a matter of getting hold of the R/W line and the host can participate
> in the reprogramming? 
>   
My primary goal is a way to create KERNAL replacements for the various 
CBM machines.  2364's are nowhere to be found, the Motorola equivalent 
is almost extinct, so an 2764 to 2364 adapter seemed a good first 
attempt.  But, as I looked at EPROM prices, I found they are getting 
expensive as well, and it's tougher to mass produce them (I have to 
either ship preprogrammed EPROMs to China for assembly, or solder them 
up here after programming them. 

So, for the DIY crowd, it seems worthwhile to offer a 2364 to JEDEC-std 
adapter board with a header and a socket, at a nominal price.  But, for 
the rest who want a finished solution, I decided to go this route.  This 
can be mass produced economically, can be easily programmed after 
assembly, and has about the same profile height as a plastic DIP ROM.
> Can the host be put into tristate reliably and something grab the buses, for
> example a 24 bit shift register driven by 2/3 wire? (Let the programmer do
> all of the magic byte stuff, the adapter just s/p translates).  Twice the
> pins if you have to mux.
>   
My initial idea on that route is SoftROM:  
http://www.jbrain.com/vicug/gallery/softrom/SoftROM_PCB  A 64kB SRAM 
with tristate buffers around the perimeter and a uC to drive.  The uC 
will hold the machine in reset while dloading the code into the SRAM, 
and then release the RESET and enable the SRAM lines.
> I assume that anything that adds price at all is out of the range, you'd be
> looking a pin count for the address and data lines at a minimum price,
> whether done with TTLish shift registers or PAL/PGA/uProc.
>   
SoftROM can load from SPI-based FLASH or SD card, though a ZigBee or 
other idea is easy to implement over a RS232 or I2C setup (or ZigBee)
> Too bad there isn't an AVR type with 512k flash. you think your talking to a
> mem, your really talking to a proc.
>   
I think the XMEGAs will go there, but it's tough to do all the work in 
250nS (4 instructions @ 16MHz), and the XMEGA RAM is not readable by an 
external source.  I could do dual port, but it's not cheap.

Jim

       Message was sent through the cbm-hackers mailing list