CUBIX/6809 updates - test-drb@ccmp.vtda.org

13 Dec 2005

...

Subject: Re: CUBIX/6809 updates
   From: Roger Merchberger <zmerch at 30below.com>
   Date: Tue, 13 Dec 2005 13:03:10 -0500
     To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>

Warning: I did not top post! There are further responses inline with the 
reply. ;-)

=-=-=-=-=

Firstly, I'd like to mention that I've always wanted to build my own 
homebrew 6809 system... altho I've been thinking of designing my own buss 
system (around 80 pins or so) so other CPUs could be used if desired. I was 
thinking of a multi-board system with a 6-7 socket backplane. I want it to 
be educational to others (read: see-thru Lucite case & lotsa Blinkenlights! 
;-) at the same time that I learn from it, too.

For a few reasons, S100 is not an option: I want the system smaller / 
easily portable (I'm guesstimating in inches: 6x6x9, with an external 5V 
only switcher), and I also want it to be easily buildable by a hobbyist. 
Trying to dremel an S100 cardedge would seem to be quite a bit of work for 
not a lot of gain - but square boards with a pair of 40-pin IDE connector 
(for example) edge-soldered on would be pretty easy to do at home.

I do have a question:

Are there any 80-100 pin buss structures already in use I could copy that 
might fit my bill? I'd rather do something that might be compatible with 
something else out there if it's similar enough to what I want to accomplish. 
Not that some busses are processor specific so if you using CPU XXXX busses
X,Z and Q may not be usefully available to you.

On the other hand if your going small, at most you need an addres, data and
control busses and for most 8bitters thats easily fit into less than 40 pins.
I've done this before and a 8+ pin connector with odd or even pins all ground
makes for a simple blackplane (IDC connectors on a wide peice of cable).

...
 I'm hoping to take lots of pictures & basically
make the design free on the 
web as I do this, mainly to show people if an idiot like me can make a 
computer, so can you! ;-)

Oh, and for the record: Cubix looks *kewl*. ;-)

Rumor has it that Scott Stevens may have mentioned these words:
 On Sun, 11 Dec 2005 20:15:48 +0000
"Dave Dunfield" <dave04a at dunfield.com> wrote:
 > > I think cubix was a good idea, but this 15 years too late for
 > > me as I
 >
 > What a co-incidence ... CUBIX is 20+ years old, so it should
 > have been perfect :-) 
And it's never too late for good ideas. ;-)

  > > realize in hindsight that 128k of
 > > memory - split code and data is needed for any real work. 
Them's fightin' words. 64K w/OS-9 got me thru High school & my first year 
of college, before I got my CoCo3 & 128K (later to 512K) but I still had a 
maximum 64K code space - and that kept me working until the mid-90's. Other 
than running (crawling) Autocad, my '386 was the 'toy' and my CoCo was the 
workhorse. Once I got an EISA 486-66 server from my (then current) 
employer, did I consider the CoCo my secondary machine. 
I didn't cross the 64k barrier until 1981.  Even then I rarely use it all.

...
   > > This
the  crummy 8088 has but not the 6809.
 >
 > Funny, I've done LOTS of "real work" in <64k 8-bit CPUs. Even
 > now a lot of my command line utilities are compiled in 64k
 > "tiny" model (Referencing stuff known here, Anyone notice that
 > ImageDisk, my Simulators and the various other transfer
 > utilities that I've done are all .COM files) - I used to think
 > 64k was lots of memory... and I still do! 
Unless you're running Windows. ;-) 
Here's the kicker.  Winders sucks up a whole hell of a lot of 
space but an App that uses it wisely can be trivial in size.

...
  64K is a HECK
of a lot of memory if your code is all in assembly.
I've worked on projects where the limited program memory in the
micro, i.e. the 16K of program memory available on-chip, was a
godsend- it served as a brake on futher 'feature creep' requests
from the folks in marketing.  ("yes, we can include new feature
'x' but it means doing away with the lookup table that feature 'j'
you requested last month uses.") 
A few of those microcontrollers have 128K of flash on 'em now.

I want my CoCo on a chip! ;-) 
I sure a Xylinx mumblefrap with all the right goop could get you 
that.  However with 8k srams, 8keproms and a amazingly small 
handfull of parts.. As drawn by Dave 27 chips gets you 48k ram,
OS in 8k Eprom, 2 serial ports, FDC that can handle 4 disks 
(360/720 5.25 or 720k 3.5" and 8" is basicaly there).  If you
recognize that the six 8Ksrams can be replaced with one 64k sram
you knock off 5 chips and the FDC portion is 10-11 chips alone 
(basic 765 with ttl around it). If one could find a smc9229 FDI 
or mod the circuit for 37C65 that chip count gets much smaller.
I happen to have some 64k srams and the majikal super rare 
custom FDI so the chip count will be under 20.  As small single
board systems go that is a very small chip count with out using 
PALS/GALS. 

Allison

...
 Laterz,
Roger "Merch" Merchberger

--
Roger "Merch" Merchberger   | "Profile, don't speculate."
SysAdmin, Iceberg Computers |     Daniel J. Bernstein
zmerch at 30below.com          | 