[Coco] COCO DTV
msmcdoug at iinet.net.au
Thu Feb 15 20:45:35 EST 2007
Here's my $0.02 worth...
The CDTV was a spin-off from Jeri Ellsworth's original project to
implement a C64 in an FPGA. The CDTV was, as James pointed out, a
commercial development that involved taking the FPGA design and putting
that into an ASIC - a process that can only be undertaken by a
commercial enterprise with a lot of capital and that is expecting to
sell hundreds of thousands of units. I bought mine for (IIRC) AUD$30.
What also came out of Jeri's work was the C-one, which looks like a
conventional PC motherboard but instead of an x86 CPU, it has a couple
of FPGAs - the idea being that these can be configured on-the-fly to
replicate 'retro' computers such as C64, VIC20, Amstrad etc. The C-one,
although technically more complex and the CDTV, is actually much cheaper
to produce, and is perhaps commercially viable if you can sell a few
hundred of them. I think they cost around AUD$700 each.
OTOH you have a plethora of FPGA 'evaluation' or 'development' boards
which aren't purpose-built, but rather attempt to provide an FPGA on a
board with an array of generic peripherals and I/O for people to 'play'
with. These boards can cost as little as USD$99 up to several thousand,
depending on the capabilities. These are what most of us are using to
develop our FPGA designs atm.
The abundance of FPGA development boards and the sheer volume of silicon
being shipped by vendors these days has brought the FPGA into the realm
of affordability for the average hobbyist in the last few years. Bigger
and cheaper FPGAs means more powerful options in the hands of tinkerers
and hackers. Modern FPGA dev boards are more than capable of Coco 1/2/3
So where does that leave the Coco DTV?
An ASIC implementation is obviously out of the question.
An FPGA implementation is certainly within the realms of possibility.
Certainly there are people even on this list that would be capable,
given the time, to produce a Coco 1/2/3 in an FPGA. So it becomes then
an issue more of economics rather than feasibility.
If you want to design a custom board, then that costs $$$. If you want
to stick an FPGA on it, in small volumes, that costs $$$. If you think
you can sell a couple of hundred, you might break even. But they will be
expensive - they certainly won't cost AUD$30. Something between that and
the cost of a C-one which I believe is actually being sold at a loss.
And you likely won't get a 'joystick' form-factor either. Without an
ASIC and with an FPGA, power conditioning, I/O for PS/2, CF/MMC/SD etc
you'd be looking at a "little box" rather than something that fits in
Of course all of this is completely relevant to the discussion of the
Coco 4/5 on this group that started a few weeks ago.
My prediction: not a lot will happen until we have fully-functional Coco
1/2/3 designs running on FPGA development boards. Only then will we
start to enhance the designs for modern-day peripheral connectivity and
features of the Coco itself. Then it will be a matter of either
endorsing a specific FPGA dev board that is particularly suitable for
our needs, or perhaps looking at development of a purpose-built board or
daughter/adapter boards if there is enough interest.
I haven't even touched on the option of doing the Coco DTV thing as a
software emulation in, say, an ARM. That's a whole new kettle of fish.
And as an aside: Jeri has become the 'Woz' of the C64 and FPGA 'retro'
community. Sure she was streets ahead of the competition and will always
be revered for that reason (and being a decent-looking chick doesn't
hurt either), but that was years ago and what she achieved is being
replicated by dozens if not more people every day.
More information about the Coco