[Coco] The Tri-Annual CoCo 4 Thread

[Mark McDougall]

On 14/02/2014 8:08 AM, Steve wrote:

> Nick, I looked up the part on line there is only 119,808 bits of ram on the
> chip. You will need more that 1,000,00 bits for the 128K of ram on the CoCo
> 3. Chips with that size of ram run $250 and up.

You wouldn't use the FPGA's on-chip RAM for the Coco memory; you'll end up 
with a very expensive FPGA that is 80% empty! Besides, on-chip memory is 
used for other purposes as well, such as FIFO's, line buffers, ROM's (eg. 
character set) and in some cases logic for components that you have no 
control over. Much better to use external SRAM that is cheap and expandable 
into the MB!

> What if design a board that holds the FPGA, RAM and the other interfaces
> (ports) to plug in CoCo equipment and have it fit a CoCo case.  The you
> would have a CoCo replacement.

I've actually discussed this idea with my business partner over the years. 
The idea was to use an FPGA module, much like the Altium Nanoboard design 
(of which we have a few) that could plug into custom base-boards for 
different purposes. eg. Coco, TRS-80 Model I, Apple II, C64 etc.

It's actually no different in concept to the Intel-based CPU modules that 
you can buy, with on-board RAM and the bridge chipsets. I've worked on 2 
designs with these modules; the base-boards were simply I/O connectors and 
custom peripherals such as SATA, UART, video, battery-backed SRAM, magnetic 
card readers, etc etc.

Returning to the concept applied to the Coco and other retro devices; the 
conclusion we eventually came to was that it would be cheaper to produce, 
and easier to design, if we simply produced a single board for each variant, 
and then lifted the relevant FPGA portions from the schematic each time. 
Whilst it may appear somewhat 'future-proof' to have a plug-in FPGA module, 
you're still constrained by the up-front design of the I/O broken out from 
the module. And chances are, if you are adding major functionality that 
requires a larger FPGA, you also need more I/O.

The approach that I, personally, would take, would be to design-in an FPGA 
that is more than adequate for current wish-lists, at the right price-point 
of course, and simply choose a system and develop the board from there. Next 
step is to produce off-shoots for other systems.

As I've said previously, my current (non-Coco) project is a rather complex 
FPGA board that has a tonne of 5V I/O on connectors... perfect for 
prototyping designs such as above. My intention is to some day get around to 
doing exactly that, and the 1st systems I'll be targeting are the Coco and 
TRS-80. But don't hold your breath, the initial project is very complex (I'm 
hoping it will have some commercial value), I'm doing it in my spare time 
(though I'm working to change that!), and could still easily be another 2 
years in the making.

Regards,

-- 
|              Mark McDougall                | "Electrical Engineers do it
|  <http://members.iinet.net.au/~msmcdoug>   |   with less resistance!"