[Coco] Multipak redesign/replacement

[Gene Heskett]

On Saturday, February 21, 2015 10:17:13 PM Gene Heskett wrote:
> On Saturday, February 21, 2015 09:16:47 PM Andrew wrote:
> > While I can't help with the design or such, I would be willing to
> > help test any resulting system, if needed.
> > 
> > That said, what I would like to see for a new MPI would be something
> > more akin to the few "backplane" systems that were available back in
> > the day in the pages of the Rainbow and elsewhere (alas, I was a kid,
> > and such a backplane was out of the question to my parents). I would,
> > though, change things up slightly.
> > 
> > Instead of orienting the paks vertically, as in a standard MPI, I
> > would rather that they were oriented on their sides; think of the
> > current MPI, with the paks vertical, then rotate the packs 90
> > degrees clockwise, edge connector toward the cartridge slot.
> > 
> > There would have to be some kind of vertical interface plane for the
> > slot connectors; probably all the needed electronics could be fit on
> > a 2-sided board (heck, maybe on one side using SMT components).
> > There would have to be some kind of extra mechanical support or such
> > for the vertical board, but I think it would be doable. An
> > alternative arrangement might be to have the same orientation, but
> > the edge connectors facing toward the rear of the CoCo (label sides
> > of the paks to the front).
> > 
> > The nice thing about such an arrangement would be that you wouldn't
> > have tall packs sticking up so far (like the original RS-232 pak - or
> > the older floppy drive controllers); in fact, since all paks are the
> > same width, that's the maximum height they would be above the level
> > of the CoCo. If a cable connection is doable with such an
> > arrangement, all the better.
> > 
> > Finally - a potential thing to do might be to put a micro controller
> > (or something) on each end to buffer the I/O lines and translate the
> > signals to high-speed serial (maybe USB?) - so that a thinner (and
> > longer) cable between the CoCo and new MPI could be used (though I
> > realize that such a solution likely has major pitfalls of its own).
> 
> Something along those lines has been on my mind since forever I think,
> but with a 80 in SCSI-III sort of an interconnect, between the board
> design we have now in the part 1 of 2 file from little John.  That,
> IIRC uses the same cabling as the 80 wire IDE, but would give us 40
> signal wires and a rock solid ground.
> 
> I think that is just a 2 layer board that we have the eagle files for,
> but I am not sure how he intended to do the interconnection between
> the slot carrier board and the logic board we already have, which has
> an 8 slot capability.  I should load that up into eagle and run it
> thru pcb2gcode to get an idea of the logic board size when laying in
> the mill table as my y motion is all used up at about 5".  It may be
> small enough I could make the first copy on my mill.  But that will
> wait till the local deep freeze is over.  Way too cold now even with
> the keep it dry heat thats on out there.
> 
> Its been so long since I designed the encoder board for my lathes
> spindle that I may as well go get the latest free version of eagle and
> install it. And probably have to learn how to do it all over again.
> 
> I'll report, when I have something to report.
> 
> Cheers, Gene Heskett

A PS as it were. I did find part 2 which is a web page describing it, with 
no links to get the slot socket carrier.  AND a note from little John 
about a couple errors plus the fact that it didn't really need 16 
positions of dip switches.

So the Part 1 eagle files and .brd drawings were obsolete before being 
committed to copper. Part 3 was supposed to be the cpld recipe's.  Neither 
happened.  Since these drawings carry a copyright date of 2010, its 
apparent that they will not be corrected.  And I am not good enough to 
even get eagle to talk to pcb2gcode.  Fortunately there is the even newer 
pcb-gcode that I have used before, and will again if I can get the paths 
set.

The slot carrier interface is a 40 pin dual row header, but to complete 
that we would have to consult the schematic and put some sort of a naming 
convention on the pins so we could carry the net to the 2nd board a lot 
easier.

What we have has 67 errors found by DRC (design rules checker), all of 
which can be safely processed and ok'd. 40 of them are at the ends of the 
finger traces where it _thinks_ the spacings are too narrow but in fact 
are fine.

And I'm still moving files around so that pcb-gcode can find them.

This board is complex enough that I expect each side will be close to a 
full days work for the mill as it removes copper, using the etching bits I 
have, in cuts about .002" wide, so a 16 mill spacing means it has to make 
8 passes to clear all the copper away between 2 traces.  Its also big 
enough that any un-flatness of the board will need to be either held flat 
by applying a serious vacuum, or the board will need to be mapped so that 
the machine can follow the warpage. Each piece individually of course.  
I'm in favor of the vacuum but I'd have to rig up a vacuum storage tank 
since the pump I have would probably overheat if it runs for 8 hours 
straight.  If edge of board holder palette leakage can be held low enough, 
the pump could run and pull a good vacuum in 5 minutes, then take a 30 
minute rest before it had to pull it down to 29.5" hg again.  Running full 
time it will also slowly vaporize its lubricating oil, filling the air in 
the shop with a glasses fogging mist of its fancy oil.

This, serial number one unit is strictly for debugging as the board houses 
can do it at 10% of my cost, or less.  I don't think much of the atx power 
supply, but elderly XT's would work as well & probably 10x more dependable 
because a new atx supply today will last only long enough for the warranty 
to expire. The name brand stuff like Antec, is very carefully designed to 
last 2 weeks past the warranty.

However, I won't pursue it any farther unless I see the potential market 
is at least 50 units. If someone else can work up a B.o.M. so we can get 
an idea of chip, sockets, resistors, caps & 40 pin connectors and ribbon 
cable, then I figure out a suitable container of lexan, we might have a 
SWAG about the cost of a kit or assembled & boxed ready to use version.

Realistically, is there a market for 50 of these if the kit was $60 and 
the boxed one was $120?  IDK.  You tell me IOW.

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>