[Coco] Future of the CoCo and the 6309

[jdaggett at gate.net]

On 8 Jun 2006 at 9:30, Fedor Steeman wrote:

> How many 6309 CPUs are there still left? 

Good Question? No real definite answer. Could be over ten thousand in the market 
left unused. 

> I know Cloud-9 must have a stockpile, but are there more besides  > those? 

Yes there are stocks left to buy with dealers who buy up surplus inventory and offer 
for sale. The internet is full of these vendors. Often most require a $100 or more 
minimum order though.

> Is there the tiniest chance they will be manufactured again? 

Fat chance by Hitachi!!!!

> Can they be cloned? 

Yes. The patents have long ran out on the 6309. Also there is most likely very little 
IP left in the device since the technology to fabricate the IC is over 20 yrs old. 

> Does Hitachi still have the blueprints? 

Since they have entered into an agreement with Rochester Semiconductor,  I doubt 
they have the masks to fabricate the IC again. Rochester is a fabrication house for 
what is now called obsolete devices. They are in business to support 
semiconductors for older designs that are not worth redesigning. Chances are that 
there is a minimum wafer lot required. A wafer lot can be up to 24 wafers with about 
300 IC per wafer at a cost of around $1500 a wafer. 

> How difficult is it to design a CPU like that from scratch?

Not so much how difficult it is but how much money one has to do it. To design and 
get the first 40 pieces can be up to $500,000. But man would it be nice to see a 
6309 on the IBM 130nM copper process. Could probably get buss speeds well in 
excess of 250 MHz. To spend that kind of money you need a market to do it. 
Today's world is to SoC ( System  on a Chip). That is integrateing more functions 
into the chip than just the MPU section alone. The ultimate design would be to put 
the whole Coco 3 on a 130nM chip. That would include the MPU, a RTC,  two serial 
ports, a parallel port, IDE  interface,  MMU  and 32K of flash. I would imagine that 
all that can easily fit on a die that is about 8mm by 8mm. 

> Can programmable chips be made to mimic the 6309?

Yes. In fact there are at least four HDL versions of the 6809 that are available. Two 
commercially and two under public license. One under public license is not 100% 
fully functional. Two instructions are not supoorted. The second one under public 
license is now somewhat stable and bug free. That version will fit into a 
Spartan2/2E/3/3E device of 200K gaes or more capacity. The FOGA will run you 
about $20 to $40 range depending on make and density. 

To make the FPGA from 6809 to 6309 yo ujust have to add the specific instructions 
to the current 6809 HDL code, recompile, synthesize and create the binary file to 
program with. 

> Just a few thoughts that have been circling through my mind this morning.
> 
> Fedor
****************

No problem. I currently have the two HDL source code for the public license 6809 
FPGA versions. Both would take an effort to add code for the 6309 instructions. 
More time tha anything to debug the new code and make sure the new code does 
not impact older code. 

james