[Coco] [COCO] Raspberry PI and CoCo Cart

[Aaron Wolfe]

On Fri, Oct 5, 2012 at 7:54 PM, David Ladd <drencor at gamepixel.net> wrote:
> Well my idea is is just addressing two FIFI buffer's at the Boisy
> Interface/Becker ports and then maybe modify the Drivewire 4 server to
> talk via parallel rather than serial using the Raspberry Pi's
> interface.  I need to find a parallel FIFO chips that can do this.  I
> figured a parallel interface would be a bit faster than serial plus
> two FIFO chips would provide at least a good buffer to allow the CoCo
> to catch up under OS-9 if other processes are going on between
> packets.
>
> Though I don't have a Pi yet and I figured since you have a Pi and
> wrote DW4 server that could come up with the required hardware to
> combine the Pi into a cart to basically be the CoCo's version of the
> Boisy Interface/Becker port.
>

Well.. I haven't the slightest clue about hardware but I have been
talking to those that do.
I've gotten the impression a FIFO/Boisy/Becker interface is very
possible for someone
with the knowledge to make things.  That would be cool, I just wonder
if we couldn't do
something a bit lower level and turn the Pi into a flexible I/O
controller rather than a
drivewire server on the end of a serial connection... there are some
pretty massively
nice things such a device could do.

> The only thing I know right now is that anyone who still uses a J&M
> Controller with a built in parallel port would have to switch to
> another floppy controller since it uses the same address ports.
>
> I still need to do more research into the FIFO chips and if 8bit ones
> are made.
>
> I figured using the becker port would make things a bit easier since
> NitrOS-9 already has the DW driver for it.  The only thing that would
> need to be done is for someone who has rights to update the source
> code to the HDBDOS project could update the DW read and write files to
> include the becker port and then add to the make file to produce
> another HDBDOS image that includes a becker compile of HDBDOS.
>

It's true that implementing the B/B interface means no additional software needs
to be written..  conversely it also means no additional functionality,
just much
better speed.  That's good, but I want to at least explore doing more.  It may
well end up that a fifo is the only practical way but I haven't given up yet.
If we had a way to communicate from coco to usb/ethernet/etc on the pi,
believe me I will write the software to use it.


> On Fri, Oct 5, 2012 at 6:24 PM, Aaron Wolfe <aawolfe at gmail.com> wrote:
>> I have been wanting/planning/trying to gather information on a "Pi in
>> a Pak" for some time now.
>>
>> Interfacing via a faster, less CPU intensive DriveWire connection is
>> interesting, and maybe that is the limit of what can be done.
>> However, I had hoped for something a bit different.  I've talked with
>> some experts (or at least they know a heck of
>> a lot more about it than I do) and gotten some ideas but maybe the
>> list has some too...
>>
>> It seems like the ideal (and maybe unobtainable) interface would be
>> one which can read and write directly to the
>> bus lines just like a regular ROM pak or drive controller.  Given this
>> ability, one could accurately simulate *any*
>> pak or peripheral (or several at a time?).  You could also create new
>> peripherals, such as interfaces to the various
>> I/O found on the Pi (USB, ethernet, audio/video, etc...).   Way better
>> than an embedded DW server imho.
>>
>> I don't know if its possible or practical, I just know that it would
>> be awesome.  From what I've gathered... it might be possible
>> but it's not a sure thing.  You have to dance pretty fast to play on
>> the bus.  The Pi does have an SPI interface capable of many
>> Mb/s throughput (one guy's testing says 16Mbps is a practical limit)
>> in addition to several GPIO pins.  John Kent showed me
>> a cheap I/O processing chip that was cheap, programmable, had SPI and
>> a whole bunch of latches, and could operate in the
>> multi Mhz range.  Can they be stuck together into something that can
>> participate on the coco bus?  I wouldn't even know how
>> to guess, but I'd surely like to find out.  Also interested in
>> alternate schemes that provide superior/more direct communication
>> than the DW style serial line.
>>
>> Can the Pi be turned into the ultimate virtual MPI?  $0.01 for your thoughts
>>
>>
>>
>>
>> On Fri, Oct 5, 2012 at 7:09 PM, David Ladd <drencor at gamepixel.net> wrote:
>>> One of the things I was thinking is with the cart side is having two
>>> FIFO chips to buffer data between the Raspberry Pi and the CoCo.
>>>
>>>
>>> On Sat, Sep 29, 2012 at 5:58 PM, Aaron Wolfe <aawolfe at gmail.com> wrote:
>>>> On Fri, Sep 28, 2012 at 10:42 AM, Brian <briang0671 at sbcglobal.net> wrote:
>>>>> On 09/28/2012 12:05 AM, David Ladd wrote:
>>>>>>
>>>>>> Here is a interesting idea I just had.
>>>>>>
>>>>>> What if we were to interface a Raspberry PI and have it connected to a
>>>>>> CoCo cart so you could take advantage of the Boisy Interface/Becker
>>>>>> port so that you could have faster access to drivewire 4 server?
>>>>>>
>>>>>
>>>>> What is the Boisy Interface/Becker port?
>>>>>
>>>>
>>>> This is a fifo buffer interface implemented initially in the CoCo3FPGA
>>>> project and since ported to some emulators.  It allows DW to operate
>>>> at very high speed/low cpu use compared to bitbanger on those
>>>> platforms.  OS9 modules to support the interface are in the NitrOS9
>>>> source.
>>>>
>>>> --
>>>> Coco mailing list
>>>> Coco at maltedmedia.com
>>>> http://five.pairlist.net/mailman/listinfo/coco
>>>
>>> --
>>> Coco mailing list
>>> Coco at maltedmedia.com
>>> http://five.pairlist.net/mailman/listinfo/coco
>>
>> --
>> Coco mailing list
>> Coco at maltedmedia.com
>> http://five.pairlist.net/mailman/listinfo/coco
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco