[Coco] Multipak redesign/replacement / New CoCo...

[RETRO Innovations]

On 2/23/2015 12:50 PM, Louis Ciotti wrote:
> This may have been brought up already, but has anyone considered this
> as a starting point:
>
> https://sites.google.com/site/tgmaxx/tandy-color-computer/expansion-interface
>
I did.  I looked at this design, the 8 slot design mentioned in the 
thread, the original MPI schematic.  All of the designs are very similar:

1 bidir buffer for the data bus (u3)
3-4 unidir buffers for the address lines and some of the select lines 
(u1,u2,u4)
latch for SW control (u6)
buffer to read latch/switches (u7)
3 4->1 multiplexors (u9, u10,u20)
some logic to trigger on ff7f (u17/u18/etc)

I did not choose this design, since I feel it used too many TTL ICs to 
accomplish the goal. The design converts the selector switch into 1 of 4 
discrete selection signals, but the original MPI used a 2 bit binary 
selector, using a DP4T switch, which this design has as well.  Using 2 
bit binary in this design would eliminate u12/u13/u20, and u5/u4 appear 
to buffer too many signals (halt, nmi, slenb, you don't need to buffer 
them), and should be able to be reduced to 1 ls244.  A 2->4 decoder 
would be needed for the LED, so that adds an IC, but using 2 of them 
(which come in the TTL package) and 4 more LEDs one could track SCS and 
CTS mappings, which might be of interest)

As for the other designs, the 8 slot unit was gargantuan, and I wanted 
something I thought would be of interest, challenging enough to allow me 
to learn this platform, but not so large as to be a debug nightmare.  
The MPI design looked fine, but I felt I could minimize TTL count by 
using a CPLD, and still provide the same functionality.

Jim




-- 
RETRO Innovations, Contemporary Gear for Classic Systems
www.go4retro.com
store.go4retro.com