[Coco] CC-Five (was Re: Pseudo CoCo4???) (LONG)

[jdaggett at gate.net]

My thoughts on an off the shelf development boar is just that, a develeopment 
board. Most if not all have some limitations. Either insufficient ram, low color 
resolution on the VGA port, and then add cost. Most companies that put out 
development boards are rather low volume production and are rather expensive. 
Costs are more dependent on the FPGA used. Some FPGAs go as high a 
$2000+ in  small quantities. 

Probably the most complete development board so far that I like is the Xess XST-
3S1000. The following are some of its features:

Features:

    * XC3S1000 FPGA
    * XC9572XL CPLD
    * 32 MByte SDRAM
    * 2 MByte Flash
    * 100 MHz oscillator
    * Parallel port
    * Keyboard/mouse PS/2 port
    * 512-color VGA port
    * 10/100 Ethernet MAC+PHY
    * 20-bit stereo codec
    * stereo in/out ports
    * 8-bit video decoder
    * dual RCA video input ports
    * USB 1.1 peripheral-mode interface
    * RS-232 serial port
    * IDE hard disk interface
    * pushbuttons (6)
    * DIP switches (2)
    * LED digits (3)
    * LED bargraph
    * dual daughterboard connectors
    * prototyping area
    * DC and ATX power connectors
    * 5V / 3.3V / 2.5V / 1.2V regulators

Cost is $398. There is a huge library of GPL HDL code that can be used and 
adapted to any need that we would need. The only limitations that I see on t his 
board is that it uses 9 bit color for the VGA. 16 bit would be better but is far better 
than the 6 bits that is current to the Coco3. The Spartan 3-1000 is a one million 
gate equivalent FPGA and is by far plenty for what we need. For judging how 
much room there is extra, I have currently  a DS2FT board that has a Spartan 2E 
300K equivalent gate FPGA and all of John Kent's System09 fits in it with room to 
spare. 

The CPU(MC6809) portion of System09 only takes up 59% of the slices of a 
Spartan 3-250. That is one fourth the size of the FPGA on the XESS board.  An 
enhanced GIME chip would take up not much more than about 10%  of the 
resources of the Spartan 2-250 chip. There may still be some optimization of 
John Kent's code toimprove real estate and/or increase speed performance. 
Currently I believe that the CPU portion could run upwards to 16MHz. Maybe 
some optimization may get it to the 25 to 35 MHz. range. 

just some thoughts. 

james



On 22 Jan 2007 at 19:54, Joel Ewy wrote:

> jdaggett at gate.net wrote:
> > In doing that  you may limit the hardware or cause an increase in
> > cost for hardware to be backward compatible. 
> >
> > Just about anything now to go into a FPGA will require extra
> > hardware to interface 3 volt logic to 5 volt logic.
> >
> > james
> >
> >   
> Hmm.  How difficult a chore would that be?  What if you didn't bother
> with designing the board that has the FPGA on it at all and simply
> planned to use off-the-shelf FPGA prototyping or evaluation boards.
> Design a CoCo personality board that would have the bus and peripheral
> connectors on it, and the DRAM sockets, and the A/D converters, and
> all the stuff that wouldn't be found on the FPGA boards.  Include the
> logic level conversion on that board and try to design for the future,
> so that upgrading would mean replacing only the FPGA board.  Is the
> logic level conversion intrinsically difficult/expensive, or is it
> just another design consideration?
> 
> JCE