[Coco] CoCo 3 68000/RAM board project

bkheath at gmail.com bkheath at gmail.com
Sat May 9 01:17:35 EDT 2009



On Fri, 8 May 2009, Andrew wrote:

>>  Message: 1
>>  Date: Thu, 07 May 2009 10:39:47 -0400
>>  From: jdaggett at gate.net
>>  Subject: Re: [Coco] CoCo 3 68000/RAM board project
>>  To: CoCoList for Color Computer Enthusiasts <coco at maltedmedia.com>
>>  Message-ID: <4A02BA73.21591.786DE5 at jdaggett.gate.net>
>>  Content-Type: text/plain; charset=US-ASCII
>>
>>  All
>>
>>  The CPU module containing the FPGA for the 6809 is laid out and needs
>>  verification. The memory module is also done. It will be a 1Megabyte by
>>  16 10 nS SRAM. Another board could be done using SDRAM. Now the Main
>>  board and peripheral boards are left to design and lay out.
>>  What I am doing is shrinking the size of the main Coco board down to
>>  100mm by 80mm size. Taking a fes of Sock's ideas also into consideration.
>>  The CPU  FPGA will be capable of one 6809 plus room left over for a math
>>  coprosser or a second 6809.  The GIME FPGA is also large enough to handle
>>  VGA desktop of 800x600 pixels at 8 bit color. That is able with the
>>  2Mbytes of ram. 1Mbyte for video desktop and at least one 640x480
>>  hardware window. Lower resolutions can have more windows.
>>  Coding status for the GIME chip is all the registers are done. Need to do
>>  the video section with windowing, and the linear address generater for
>>  the video ram access. Also the multipexor between the CPU and video
>>  addresses.
>>  As for the math coprosser, I have a cordic processor started. It
>>  eventually will handle all division, multiplication and several trig
>>  functions. I have yet to dicide wthether it will be just integer or
>>  handle floating point yet. That will depend on how much room is left
>>  after the core of the cordic is left.
>>  james
>
> James,
>
> Personally, I would want to see the second 6809, 2 meg of ram, but with a 
> limit of 320x400 with 256 colors (Mode X, more or less), and hardware 
> sprites. If not a second 6809, then a music/sound chipset or something, 
> which the CoCo has sorely lacked forever, barring the SSC and/or Orc-90 
> (although I hesitate to call the Orc-90 a music/sound chipset - most of the 
> functionality was in the software, the hardware being only a very nice set 
> of DACs for stereo sound).

I like the idea of a second 6809. Since you've already got an
MMU and fast ram why not do that Motorola dual proc design with
the two E clocks 180 out of phase. Set aside a register set in
the MMU for "shared" blocks (which could also be none). Then
software could configure it with one for I/O one for processing,
or one for block devices one for user interface, lot's of
possibilities, even independent EDB and OS9 with the shared
memory emulating a fast comm link.

'Course I wanted to build a set-up like this back when my
hardware bench was available, still do if the chance comes.

There should be room, the CPU-CPU memory arbitration logic was
pretty simple, though it does involve buffers to isolate another
set of address and data lines to keep it from interfering with
the interleaved display memory.

Just some thoughts.


Brett K. Heath



More information about the Coco mailing list