[comp.sys.m68k] 680X0 dream machine

andrew@motto.UUCP (Andrew Walduck) (04/12/90)

I'm looking for information about homebrew systems that people are using
to run MINIX that are based around the 680[2,3,4]0 processors.

I'm looking for stories, advice, schematics, how-to, dreams etc.

Why??
Does a 68030 system running Minix appeal to you?? Especially if it can
be done cheaper than a PC?  It could be a "generic" (if there can be
such a thing ;-) MINIX machine. Totally non-proprietary, totally
public-domain. 

If you're interested, send mail, or post.
Thanx.
Andrew Walduck
andrew@motto.UUCP

meo@stiatl.UUCP (Miles O'Neal) (04/13/90)

andrew@motto.UUCP (Andrew Walduck) writes:
|
|Does a 68030 system running Minix appeal to you?? Especially if it can
|be done cheaper than a PC?  It could be a "generic" (if there can be
|such a thing ;-) MINIX machine. Totally non-proprietary, totally
|public-domain. 

Don't run off just because I say the N-word - this may still be applicable.
There's a group doing just that, except it's with a NS532 chip set. They
are now running the second set of production boards - set up for 6 serial
ports, *2* SCSI channels, and room for 4 to 32 (I think) MB of RAM, with
or without parity (board-selectable).

Even though the 2 chips are somewhat different, it might make a good
starting point. In fact, if the two groups worked together (C part of
the drivers and such), things could move even quicker!

They also have some good lines on all the various parts, including SCSI
drives you would need.

-Miles O'Neal
meo@SalesTech.com
emory!stiatl!meo


   This posting involves no company proprietary information.
My employer & I speak for ourselves, not each other. Trust me.

zaft@suned1.Navy.MIL (Gordon C Zaft) (04/13/90)

In article <105@motto.UUCP> andrew@motto.UUCP (Andrew Walduck) writes:
>I'm looking for information about homebrew systems that people are using
>to run MINIX that are based around the 680[2,3,4]0 processors.
>Andrew Walduck
>andrew@motto.UUCP

What's wrong with the Unix PC (TM) ?  It's a 68010, it's already out
there, and it's running a REAL operating system.  Furthermore, it
can be had for not too much money.

-- 
+  Gordon Zaft                        |  zaft@suned1.nswses.navy.mil         +
+  NSWSES, Code 4Y33                  |  suned1!zaft@elroy.jpl.nasa.gov      +
+  Port Hueneme, CA                   |  Phone: (805) 982-0684               +
+  These opinions are mine, the Navy doesn't pay me to have opinions for it. +

rainer@boulder.Colorado.EDU (Rainer Malzbender) (04/13/90)

The response about the Unix PC was appropriate, but having built several
homebrew 68000 machines I for one would love to see a public-domain
hardware project. Schematics would be available (in different formats) on
the net, pcb's would be available at cost, and it would run something
like Minix. It would be nice to use available peripherals (multisync
monitors, SCSI drives, etc.) I know some outfit sells 68K motherboards
that then use the PC bus and some of its controller cards, but this seems
unnecessary. You could pretty easily do a motherboard with an '030, SCSI
controller, 1024 x 768 (say) color graphics (256/16M), built in
ethernet, wads of ram using SIMM/SIP's. One issue that needs careful
thought is the bus - the PC bus is cheap, but inadequate, and better stuff
like VME, NuBus, etc. is probably too expensive (ever buy a Mupac VME
chassis ???) Oh yeah, it's got to have CD-quality sound and a DSP on board,
and a transputer interface. Then we'll spend the next five years writing
code for the beast, at which time HAL Systems will come out with the $1000
1000 mips, 100 mflops pocket PC and make us gnash our teeth.
			Free Hardware Foundation :-)
--
Rainer M. Malzbender                           Just another Unix/C demagogue.
Dept. of Physics            (303)492-6829         rainer@hibachi.colorado.edu 
U. of Colorado, Boulder, USA                malzbender@opus.vaxf.colorado.edu

mikec@wheaties.ai.mit.edu (Mike E. Ciholas) (04/13/90)

In article <19620@boulder.Colorado.EDU> rainer@boulder.Colorado.EDU (Rainer Malzbender) writes:
>The response about the Unix PC was appropriate, but having built several
>homebrew 68000 machines I for one would love to see a public-domain
>hardware project.
> [more stuff about i/o, display, etc.]
>			Free Hardware Foundation :-)
>Rainer M. Malzbender                           Just another Unix/C demagogue.
>Dept. of Physics            (303)492-6829         rainer@hibachi.colorado.edu 
>U. of Colorado, Boulder, USA                malzbender@opus.vaxf.colorado.edu

Sure, I've always thought that the electronics in a PC, Sun, or MAC were
dead simple (and very cheap for what you pay for it).  I a hardware type
(I hand layout 4 layer boards, for example) but I couldn't imagine
myself doing the system software (do I hear voices of those who could?).

I am willing to bash this idea around for awhile.  I'll offer my expertise
(ha!) on hardware (and I can get PCB made, also).  I'd be willing to 
layout someones design and sell the PCB for double cost (that would be
about $100 for an average PC 4 layer motherboard).  If we set our own
standard, and give it away (like unix, X, etc.), we can finally do it
right without all those silly no cost compromises that other companies
make.

I'd like to see a wide bus (32 bits), probably standardize on the 680x0
series (maybe 88000 support?).  Plug in cards for I/O, scsi, video, etc.

Okay people, let me hear from you.

Mike Ciholas

email:  mikec@ai.mit.edu
snail:  289 Highland Ave #108/Somerville, MA 02144
phone:  (617) 623 3563
air:    N1909C, 1954 Cessna 170B

garlick@csuchico.edu (Jim Garlick) (04/13/90)

This is wierd, but aren't there literally tons of S100 IO cards and
busses out there that can be had for next-to-nothing?  And wouldn't
it be possible to put a 68030, SCSI interface, and SIMM sockets on
one board, so the slow bus bandwidth need only be used for talking to
serial cards, eprom programers, fdc's, and all that other old junk that
everybody's throwing away?  Remember those old Northstar Horizon's?
Didn't they have nice boxes?

We could port gcc, mach...

cleeland@rex.cs.tulane.edu (Chris Cleeland) (04/13/90)

In article <7835@wheat-chex.ai.mit.edu> mikec@wheat-chex.ai.mit.edu (Mike E. Ciholas) writes:
>Sure, I've always thought that the electronics in a PC, Sun, or MAC were
>dead simple (and very cheap for what you pay for it).  I a hardware type
>(I hand layout 4 layer boards, for example) but I couldn't imagine
>myself doing the system software (do I hear voices of those who could?).
>
>If we set our own
>standard, and give it away (like unix, X, etc.), we can finally do it
>right without all those silly no cost compromises that other companies
>make.
>
>Mike Ciholas
>
>email:  mikec@ai.mit.edu
>snail:  289 Highland Ave #108/Somerville, MA 02144
>phone:  (617) 623 3563
>air:    N1909C, 1954 Cessna 170B

You've certainly got me!  I'm sitting here at a "vintage" Macintosh,
wondering what in the H*ll I'm going to do with it now that it's *so*
out-of-date.  It certainly would be nice to have a machine that one
could put together for about $1000 that would last you more than a
few years!

I'm no expert in any of these categories, but I feel competent in the
software area.  I'm willing to give it a go as member of a team.

How about it?  Somebody else out there must want something like this
and have some [free?] time!
-- 
--------------------------------------+--------------------------------------
Chris Cleeland                        | BELL:   (504) 866-8186
INET:   cleeland@rex.cs.tulane.edu    | USnail: 1320 Lowerline, Apt. E
UUCP:   cleeland@rex.UUCP             |         New Orleans, LA  70118
Disclaimer:  "I'm a student -- I can't afford one!"

marc@lakesys.lakesys.com (Marc Rassbach) (04/15/90)

If people are really intereded in this type of project, might I suggest
you contact Lee Felsenstein at lee@Well.  He had a similar project back
in 1985.

The s-100 idea for the serial/parallel/io is a good idea.  But why not
2 buses?  One for the 68x00, one for the I/O (like serial)

-- 
Marc Rassbach     marc@lakesys	              If you take my advice, that
"I can't C with my AI closed"                 is your problem, not mine!
              If it was said on UseNet, it must be true.
   Unix - It's a nice place to live, but you don't want to visit there.

pa1562@sdcc13.ucsd.edu (C. SQuibby Breyman) (04/15/90)

In article <1990Apr12.233111.3455@csuchico.edu> garlick@csuchico.edu (Jim Garlick) writes:
>This is wierd, but aren't there literally tons of S100 IO cards and
>busses out there that can be had for next-to-nothing? [...]
Yes, but do you really want to put the effort into designing
something with such a limited future. Additionally, the procurement
of replacements (or even standard equipment) would dwarf the present
hassles with cross-preiph compatibility. Do you like writing device
drivers that much? ISA seems that it is fairly easy to get info on,
it has a big user & developer base. Isn't volume production rather
than obsolecence a better route to lower cost?
Sinc,
C. Squibby Breyman

rainer@hibachi.colorado.edu (Rainer Malzbender) (04/15/90)

In article <1990Apr12.233111.3455@csuchico.edu> garlick@csuchico.edu (Jim Garlick) writes:
>This is wierd, but aren't there literally tons of S100 IO cards and
>busses out there that can be had for next-to-nothing? 
>	...

I've actually built up a 68K system doing this, except I modified the
S100 bus assignments and just used the available cheap wire-wrap boards
and backplanes (got 'em for free, that's why I did this, it was supposed to
be the world's greatest synthesizer). Anyway, for *this* type of project
(homebrew high-end graphics Unix workstation) the S100 bus is woefully
inadequate. First of all, all the stuff we (I ?) want on the motherboard
wouldn't fit on an S100 card. Secondly, the S100 is not a 32-bit bus.
Thirdly, it uses cards with on-board regulators, requiring an 8 volt supply
whereas we can buy PC-style +5,+-12 supplies more cheaply. Fourthly, my
experience with S100 goes back to the late seventies, so I don't know what
kinds of boards are still being made, but I doubt you'd find anything
all that useful for this kind of project. Fifth, I hate the S100 form factor,
since it really requires a bulky chassis.
	Having as much circuitry as possible on the motherboard will be
cheaper than having to buy a bunch of cards. Peripheral Technology's
PT68K-2 (see MicroCornucopia, Nov-Dec '88) was a nice idea for a 68000
motherboard using the PC (not AT) bus, but it's a little wimpy now. Using
the AT bus might be OK, but if everything is on the motherboard (I mean
everything - GPIB, Midi, audio, SCSI, graphics, memory, floppy controller,
transputer link, Ethernet, for starters :-) what kind of AT boards do you
need ? I see the bus as a way to add more homebrew stuff later, as people
come up with ideas and needs, so we might not need to use a standard bus.
Lots of people use stock Amigas and ST's without using the bus, but I think
there should at least be one. I guess the other alternative is to have a
very simple and cheap motherboard and put everything on cards and let people
buy what they needed.
	Would someone who knows about the NS532 project please post
something ? I hate reinventing the wheel (although I do it all the time :-)
--
Rainer 'rhymes with miner' Malzbender          Just another Unix/C demagogue.
Dept. of Physics              (303)492-6829       rainer@hibachi.colorado.edu 
U. of Colorado, Boulder, USA                malzbender@opus.vaxf.colorado.edu

jac@paul.rutgers.edu (Jonathan A. Chandross) (04/15/90)

rainer@hibachi.colorado.edu (Rainer Malzbender)
> Secondly, the S100 is not a 32-bit bus.

Both Cromenco and CompuPro sell 68020 based systems that use the S-100 bus.
Cromenco used two of the reserved pins to implement special 32 bit memory
signals.  During phase 0 of a bus cycle you request a 32-bit access with
one of the wires.  During phase 1 you get back 16 bits and the MUACK if
the board can deliver 16 bits in the next cycle.  During phase 2 you get
the next 16 bits.  Thus, you can get 32 bits over a 16 bus using 2 of the
3 bus cycles.  (I may have butchered the explanation; this is from memory.)
I don't know how Macrotech's 386 board deals with things.

> Thirdly, it uses cards with on-board regulators, requiring an 8 volt supply
> whereas we can buy PC-style +5,+-12 supplies more cheaply.

There is now a standard for using regulated supplies; CompuPro and Cromenco
use it.  The standard basically says that you have to supply jumpers so the
board can be used in an unregulated system and that all boards modified to
work in regulated systems have to be labelled as such.

> Fourthly, my
> experience with S100 goes back to the late seventies, so I don't know what
> kinds of boards are still being made, but I doubt you'd find anything
> all that useful for this kind of project.

You would be surprised -- you can get a SCSI board with a time of day
clock, and four 38k serial ports (might be 19.2k) for 300-400 bucks from
a number of companies.  CompuPro, Cromenco, and Macrotech still sell
quality high performance stuff.  Both CompuPro & Cromenco sell 68020
systems, Macrotech sells a 386 board, you can get a PC compatable video
board (will run flight simulator), SemiDisk sells 2 Megabyte solid state
disk emulators, etc.  The stuff is definitely not cheap, though.


Jonathan A. Chandross
Internet: jac@paul.rutgers.edu
UUCP: rutgers!paul.rutgers.edu!jac