[Coco] grey MPI 26-3024 - how to tell if upgraded

[Darren A]

On Mon, Feb 9, 2015 at 11:20 PM, RETRO Innovations wrote:

> Is there anyone who can confirm the below?  THere are a few discrepencies
> (on CTS, for example, and no mention is made if SLENB writes are allowed)
>
>  * If SLENB active
>      o allow reads from any location
>      o allow writes from any location?
>  * If SCS active (Coco3SM says this is only active for both reads and
>    writes, but only for addresses ff40 to ff5f)
>      o allow reads from ff40 to ff5f
>      o allow writes from ff40 to ff5f
>  * if CTS active (Coco3SM says this signal is only asserted on reads
>    from c000 to dfff, but Gunnison's doc says c000 to feff)
>      o allow reads from c000 to dff (feff?)
>  * If no additions lines active:
>      o allow read from ff40-ff7f
>      o allow writes from ff40-ff7f
>
> The PLD equations (http://www.coco3.com/community/2010/06/26-3124-mpi-
> coco-3-upgrade-2/) seem to say:
>
> enable buffer if:
>
>  * address is $ff40-$ff7f or (regardless of whether SCS is active or not)
>  * SLENB is active (any address) or
>  * CTS is active (any address)
>
>
> What about ff80-ff8f?  Why can't it be used?
>
>

The SCS and CTS signals are only asserted on valid cartridge access cycles.
The data transceiver should be activated (in the appropriate direction) for
both of those. SCS is just a convenience signal for decoding the range of
FF40-FF5F which is also within the legal external range (FF40-FF7F). If you
decode that entire range then there isn't any need to consider SCS when
deciding if the transceiver should be activated.

The SLENB signal is not asserted by the CoCo but rather by cartridges
themselves. This allows a cartridge to override the standard decoding. The
MPI should always activate the transceiver in this case.

The FF80-FF8F address range is excluded because the CoCo 3 GIME chip drives
the data bus when that area is read, even though no registers are actually
implemented at those locations.

- Darren