[Coco] Multipak redesign/replacement

Gene Heskett gheskett at wdtv.com
Tue Feb 24 09:11:39 EST 2015



On Tuesday 24 February 2015 00:54:17 RETRO Innovations wrote:
> On 2/23/2015 11:24 PM, Gene Heskett wrote:
> > If the cable is used, then height above the desk is up to the box maker. 
> > If thats built so the pcb is, say 5/8" above the desk, you could lower
> > the tops of the carts sticking out by close to 3/4" from what it is now. 
> > Exactness in that case is very nearly a moot point.
>
> True, but I'll probably design the PCB so that the cable is an optional
> thing (by having the board house make the PCBs as one combined PCB (they
> charge per design) with a way to snap the PCB if needed.  I've done
> similar on a few designs.

Sounds like a win-win.

> > Which as I said before could be exceedingly useful if we can generate a
> > signal going back into the coco to disable its internal CS in the right
> > places. But I don't know if thats possible.  If not, we''l have to cobble
> > up a teeny board with additional decoding on it, and cut it into the
> > coco's own decoding logic.  Doing this will free up 7 additional 4 byte
> > wide i/o ports between $FF04 and $FF1F, and for another 7 from $FF24 to
> > $FF3F.  Even just one of those 7 port wide additions to the /CS decoding
> > would solve a huge number of our address conflicts we have now.
>
> Well, I think I have a solution for that, that requires no additional
> logic.  Can you elaborate on the address conflicts folks see now?

Working from memory, no.  Because I hacked the decoding in the packs such that 
my normally loaded packs, an fdc, a tc^3 and a dual port rs232 pack can all 
live together. And the wp-rs that is not now in the system since Boisy threw 
away my special drivers very early in his sysram wasting cutting up of 
drivers to make them one function drivers.  But he was running the show and 
my messages to him went into a black hole at the time.  The oem version of 
the wp-rs used 2 address in one 4 byte wide reservation and 2 addresses in 
the next one up, so that decoding and the driver got fixed in my own setup.

So if the contents of the mpi are relatively fixed, any addressing clashes can 
be fixed in the pack itself.  But this of course works only for me and other 
hardware savvy folks, and is a dismal failure for someone who doesn't even 
know which end of a soldering iron gets hot, let alone follow the logic of a 
74LS138. :(

The ideal situation, but a non-starter, would be to put the address of the 
pack in ALL descriptors and decode on the mpi board to select which slot 
based entirely on the 4 byte wide i/o address in the descriptor.  That of 
course is a non-starter because the /CS logic in the packs would have to be 
bypassed and thats still a non-starter for the hardware compromised user who 
can't fix it on his own kitchen table.  All of the current fdc's waste a $10 
byte wide i/o space that with the full decoding could be limited to $08 
bytes, freeing 1 more address slot of 4 bytes,  And just that one additional 
space would go a hell of a long ways toward solving the I/O space problems 
the coco3's gime throws in our laps like a cup of McDonalds coffee.  Of 
coarse this doesn't fly if there is a real time clock. Mine had 2 but its 
been removed from the SC-II 4n1 for a while now. Running  a lot better on the 
tc^3 clock.

> > Fine if there is room. In my midden heap theres about  a foot of paper
> > mixed with old 3.5" drives and what have you piled up against the right
> > side of mine.  I think I have mentioned it once or twice that I am a
> > packrat, married to a packrat (which ought to be illegal) .  ;-)
>
> Well, I can bring the horiz connector out to the back, if that's of
> interest.  Question is whether a 1x8 or 2x4 arrangement is preferred.
> The schematic won't care, but the PCB will...

Personally, I'd just as soon they all lined up in a row, left to right. But 
thats just me, and its' worth what it cost you to read it. :) zip.

The one thing I would love to see, is to get rid of the MPI's front belly, 
hanging out and severely restricting my table space to run my mouse on. 
Another 1.5" cleared would make a heck of a difference.  The extra logic, and 
the psu connector all need to be on the back side of the board if it is to 
stick out horizontally as one piece. The midden heap of paper & what have you 
is 100% portable, some of it to the trash can if I get off my duff. :)  My 
mouse, a serial mouse, has had its cable rerouted out the rear of the mouse 
so it wasn't beat to death against the MPI. North of 20 years ago I did that, 
and as the cable then hangs over the front edge of the desk under my wrist, I 
am surprised I haven't had a broken cable failure, but its still marching 
along like a good soldier.

The construction of my "computer desk" is such that while there is room for 
cabling and Roys video convertor behind and under the monitor shelf, there 
isn't room for a pack turned on edge. I've no clue how many cocoists have a 
similar limitation.  Since I'm decent with wood carving too, that is not an 
unsolvable problem as the monitor shelf, IIRC is subject to being raised or 
lowered in 1" increments. I think, haven't checked in the years since I 
retired, and Dave needed the room in his postage stamp sized office, so he 
gave me the whole maryann, back tower style desk, coco3, monitor, drives and 
all. He's a pc guy, what can I say.  So I am not completely out of the coco3 
business if my main machine has a heart failure. ;-)  And I have one spare 
6309 to put in it yet, I bought 2 from Chris Burke waaaaaay back when... :)

I'll get you the rest of those measurements yet today. ATM working on first 
cuppa cold caffeine, leftover from yesterday, trying to get both eyes open at 
the same time...  And still configuring mail filters as I gave up on wheezy's 
font rendering errors and installed a fork of the kde3 tree called TDE, just 
incompatible enough I had to re-invent all the wheels I've been useing for 
years.  Working great so far except for printing.  But thats 100% a wheezy 
problem.

When I get caught up on the other stuffs, I'll go file a bug, the poppler 
library debian 7.8 uses is so old it cannot do duplex!  And my color laser 
can .

Take care Jim.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>


More information about the Coco mailing list