[Coco] CoCo3FPGA, FPGA Development, and GIME replacement

[John Kent]

Hi Joel,

On 6/07/2011 1:41 AM, Joel Ewy wrote:
> Gary has put an integrated 'hardware' Drivewire client into 
> Coco3FPGA.  This uses an unmodified DECB ROM image.  The CoCo doesn't 
> know it's not talking to a FDC.  And CoCo3FPGA can talk Drivewire at 
> something like 430K.  At the current state of development there are no 
> mass storage options other than Drivewire for CoCo3FPGA.  I think it 
> would be nice to see some kind of flash memory device so the thing can 
> stand on its own.  But if/when Gary puts the source back out there, 
> somebody will likely add that if he doesn't himself.
>>>> Would I have to run the development system on windows?  If yes, no
>>>> deal.
>>> The software runs under Linux.  It's a 970M download.
>> Great, not a problem as I have 4 terrabytes of drives here.  URL?
>>
>
> This is the only thing I have bookmarked:  
> http://www.xilinx.com/support/download/index.htm
>
> Now, I only downloaded the 900M one instead of the 4G one where they 
> have high resolution JPEGs and a Bio of each logic gate.  When I 
> originally set up my CoCo3FPGA system I tried an earlier version of 
> Webpack on Win2K.  I finally figured out how to get that to work just 
> to upload the bit file.  That machine has suffered a hardware failure, 
> and I'm just not using MS-Windows enough these days (other than fixing 
> other people's malware-infested boxes) to bother reconstituting it.  
> Gary's got a new version with pretty 4096-color graphics, so I 
> downloaded the Linux version (I'm running Ubuntu 10.04 on this 
> machine) for the upgrade.  Unfortunately, the Digilent board requires 
> hardware modification for the new colors, and that simply hasn't 
> gotten done yet.
>
> Like John K. says, you've got to register with them and do a deal with 
> Rumplestiltskin for your firstborn child, but no money changes hands, 
> and they haven't come to collect Henry yet.
>
> JCE
>

One thing worth explaining is that so far Gary has only made the binary 
or bit file of CoCo3FPGA available, so all you are doing with the 
development software is using the programmer section to download the 
binary to the FPGA or configuration Flash memory. I do have an old copy 
of Gary's verilog code that he gave me ages ago but things have changed 
a fair bit since then. I think he does plan to release the source code 
when he gets it to a point where he is ready to distribute it.

My system09 project is a system on a chip that is similar to the old 
South West Technical Product Computer 6809. It does have a color display 
although it does not have the same resolution graphics as on theCoCo.  
Most of the software for the SWTPc was text based. It has PS/2 keyboard, 
RS232 port and extended memory. System09 does however fit in a 200Kgate 
FPGA boards, so it's much smaller than Gary's CoCo3FPGA.

I've ported system09 to a variety of FPGA boards, some of which have a 
16 bit IDE port which I have connected up to a Compact Flash card 
running in True IDE mode. I'm running the old Flex 9 software and I've 
written IDE drivers for it that allow me to interface an IDE hard disk 
or CF card.

It should be possible to add an IDE port on the end connector of the 
Digilent Spartan 3 starter board. Alex Freed cut and jumpered the tracks 
of a Digilent test point header board to rewire the Spartan 3 starter 
board to a 40 pin IDE socket for his Apple II implementation. It's then 
possible to use a commercial IDE to CF card adapter to interface to a CF 
card. Digilent however no longer sell the test point header board, so 
you'd need some other way to transpose the pins. I haven't bothered to 
do it with my system09 because I can't find an elegant way to do it.

The CF card operates at 3.3V. If you hooked up a Hard Disk drive running 
at 5v you'd need series 100ohm protection resistors in the signal lines.

The point is that it wouldn't be hard to implement an IDE interface on 
the Spartan 3 starter board and that would give you stand alone 
operation. There are 32 uncommitted pins on the end connector of the 
Spartan 3 starter board which are more than sufficient for a 16 bit data 
bus, 3 address lines, 2 chip selects and 2 read/write signals.

The DE1 board has an SD card socket on it however implementing an SD 
card interface from the little information I have seen is much harder 
than writing and IDE driver. The IDE driver simply has to set up the 
sector cylinder and head registers issue a read or write command to the 
command register and poll a couple of status bits to see when data is 
ready on the data bus. The issue then is how to interface it to disk BASIC.

Anyway I am rabbiting on ... 3.00am here in Melbourne Australia. Time 
for bed.

John.

-- 
http://www.johnkent.com.au
http://members.optusnet.com.au/jekent