Re: 3B2 Disks

On 2009-01-13, Bill Gunshannon <billg999@xxxxxxxxxxx> wrote:
In article <slrngmo6kh.ke5.dnichols@xxxxxxxxxxxxxxxxxx>,
"DoN. Nichols" <dnichols@xxxxxxxxxxx> writes:
On 2009-01-12, Bill Gunshannon <billg999@xxxxxxxxxxx> wrote:

[ ... ]

Mine were, 2 MFM drives on a custom (Western Electric) controller.

O.K. Any clues as to whether it was implemented as a
MFM<-->SCSI adaptor and a SCSI controller? Or did it plug directly into
the system bus?

Pure MFM. SCSI came much later as an add on card. Doesn't mean people
didn't then put MFM on the SCSI controller but it sure wouldn't make
much sense to do so. Biggest MFM disk was ~70 meg.

Actually, the biggest MFM 5.25" full height drive was the Maxtor
and Priam 190 MB (raw), which formatted to 160 MB on the 3B1.

Not much space,
even in those days. I saw a 3B2 running as a USENET News Server. Not
going to do that long with only 2 70meg disks when the OS easily takes
up one whole disk. :-)

Well ... you could strip out a lot of the OS if it were only to
be a news server. You would need to keep uucp (unless you were running
NNTP instead), and you would need to keep at least one editor, but a lot
of the application programs (including the Documenter's Workbench) could
be stripped out. And blowing away the Man pages would make a lot of
room.

he 3B1 never had on-line manuals. Add to that the fact that the
default editor if you clicked on a file was "ed", and the only
documentation for ed was in the development set, which also got you vi
so you would not use ed by choice at all. :-) Of course, the assumption
was that Mr. Businessman, who had it on his desk, would have installed a
word processing package which would replace the ed as the default
editor. I can just imagine a businessman trying to find out how to get
out of a program which simply prompts with a '?'. :-) I had other
manuals from other versions when I got my first 7300, so I had a chance
to deal with that.

There was *one* SCSI controller for the 3B1 which a member of
this newsgroup used to have (and may still have). It was being
developed by his company just about the time the bottom dropped out of
the 3B1 market. So -- he got to take it home and play with it.

I heard about this, but never saw one available.

Not one *model*. One physical controller, no more ever made.
:-)

Of course, this would
not have helped the 3B2 as they shared no common bus.

Agreed.

The SCSI card
for the 3B2 was definitely a commercial offering. I just can't remember
if AT&T was the maker or even the marketer.

O.K.


I actually used my first MFM drives (a pair of Ampex 27 MB
drives) using a MFM<-->SCSI adaptor and a hand wire-wrapped host adaptor

I thought you had been doing this a lot longer than that.

I was. I started with a pair of 5.6 MB drives by IMI (IIRC),
which used a weird controller card to talk to up to two drives. The
drives were not MFM (or at least not ST506 interface). When I
overflowed those, I got the first Ampex 27 MB MFM drive to add one, with
a SCSI controller card, and when that got near full, I picked up another
Ampex 27 MB drive to fill up that controller. Lots of wire-wrapped
interfaces and home brewed drivers. I first wrote and tested a driver
on the SSB DOS-69, and once that was known to work, I attacked writing
the driver for OS-9, and forgot about using 5.6 MB on a SSB OS, let
alone using a 27MB one on there. :-)

I still have
(and use) my first MFM disk. A whopping 12M which was actually rather
large at the time, 5 meg being much more common. Luckily, I didn't pay
for it. It easily cost more than the computer it was connected to.

I still have both the 5.6 MB drives and the 27 MB drives, and
may someday fire that system up again.

on a system which was dual-boot -- SSB's DOS-69 and Microware's OS-9
(the *real* one, before Apple claimed the name for the Mac. :-)

I thought Apple went straight from SYSTEM7 to OS-X skipping over things
like OS-9 (which was already trademarked).

Nope -- they had OS-8 (which did not bother me) and OS-9
(different punctuation between "OS" and "9" than Microware used, and
they and the lawyers, so they could get away with it. :-( This did lead
to a lot of debates with a friend in Minnesotta (never actually saw him,
just e-mail exchanges). I insisted on calling the Apple version OS-IX.

Of course, Microware is all
the way up to OS-9000 now. :-) I still have systems running OS9. I
just wish they would open source OS9 and OS9-68K, but I doubt it will
ever happen.

That would be nice. It would give me something to run on a set
of VME cards which include a 68K CPU (68020 IIRC) and a floppy
interface. Until I get an OS, I won't bother hanging power supplies on
the card cage. :-)

OS-9 handled it a lot more gracefully than DOS-69 did, because
when you blend a 6.3 filename format fixed upper case with a lack of
subdirectories, 27 MB becomes *very* unwieldy. :-)

How about early MSDOS?

I didn't touch a MS-DOS machine until they had subdirectories.
And then I would still have preferred to not do so, but work insisted.
(I later wound up on the unix/network admin team and got plenty of unix
then. :-)

I had a NEC APC. Had 2 quad-density 8" drives.
1.25M each. Ever heard of "Extended Directory Entries"? You type DIR.
The system prints out all the entries in the normal directory. Then
comes "The Extended Directory Entries". Move head to directory. Pointer
extended directory. Move head to extended directory. print one entry.
Repeat until all entries printed. Takes about 1 minute per entry. :-)
Makes CP/M and RT-11 look like real screamers.

:-)

Those systems shared four 8" floppys and four 5.26" floppies as
well.

OS/9 always did a good job of disk handling. Could easily handle many drives
all in different formats. A really good OS.

Right. I even wrote a modified floppy driver for it to handle
the Radio Shack Color Computer OS-9 format -- which had to skip over the
Radio Shack directory which was always in the middle tracks. :-) I did
this so I could use the more affordable C compiler from them instead of
paying Microware the extra bucks for normal floppys. (I also had to do a
little tweaking to the binaries to keep it from forcing the source to
live on a different disk than the compiler did. :-)

[ ... ]

a real password, or a
redirection to a shadow password, depending on the age of the system.

Never saw shadow passwords on an 3B I worked with. I don;t think they
were around long enough for that to become the practice.

I remember having an open source implementation of shadow
passwords,

Yeah, was passed out in comp.sources.misc.

Yep. When it came out, I was using (remotely) a BBN C70
which was the e-mail system for the area where I worked. 10-bit bytes,
20-bit words, 40-bit longs. :-) And some of them had no unsigned (or was
it no signed) version. :-)

It really blew the normal "compress" program out of the water,
but the configuration for another compression program (which I got from
the OS-9 User's Group Library) could be tuned to deal with that, so it
would work happily.

Actually, the normal LZW "compress" gave no trouble -- until it
was time to *uncompress* a binary. It could fill several disks.

But we didn't call it
"Open Source" in those days. We just wrote code and gave it away.
Everyone knows Stallman invented Open Source.

:-)

but I never dared to install it -- too much chance of not
being able to access the system if something screwed up. Kind of like
some company which suggested removing the passwords from the /etc/passwd
file and putting them in a database. That scared the others in
comp.unix.wizards at the time -- though now that is just what the unix
under Mac's OS-X does.

Well, the first shadow password suite was in comp.sources.misc volume 26
which was lat 1991 and the 3B line died in 1990 so it is pretty safe bet
that AT&T never officially adopted it.

O.K. That does not mean that someone has not implemented it
from those sources -- perhaps even in the machine which started this
thread. :-)

[ ... ]

OK, I'll see your Intergraph and raise you a Tektronix 4300 family
graphics station.

Was that the one with a QIC tape drive in the right pedestal? I
think that I have the tapes from one hiding somewhere around. :-)

Nope. All the software it needed as a graphics design station was in
firmware. But you could boot it out of this with CPM-86 and that even
came with Fortran. Had a pair of 8" (big clunky Shugart's) disks.

O.K. The SA8002 and SA8004 (two and four surfaces, resulting in
10 MB or 20MB? My first machine actually came with the drivers for that,
but with a 20 MB drive you could not put the man pages on. :-)


The Intergraph was a significantly larger and heavier beastie.
I had to run it up two deck plank boards to get it into the bed of my
pickup, and those boards were sagging about 6-8" below the centerline
between the ends. :-) I kept expecting them to break. I would guess
about 600 pounds. I've got machine tools which weigh a lot more, and
I'm not sure what the weight of the 6' rack filled with a Sun 3/260 and
three Fujitsu Eagles was. A lot, for sure. :-)

I have a real truck now, but when I thin about what I moved using my
old Volvo 200 series station wagon.... Once made three trips to Jersey
and came home with the car completely jammed to the ceiling (even had
one on the pasengers seat) with RA80/RA81 disk drives.

Was there any clearance between the frame bumpers and the
axle? :-)

Ran not only thier custom stuff but also CPM-86.

Ouch! I never knew what the CPU was in the beastie we had
hanging around in the hall for a year or two after we moved into that
building. It finally found an owner to turn it in during one of the
regular property roundups -- someone got tired of having to travel to
our building to verify that it was still there. :-)

I have picked up a lot of gear that way. I got the Tek when we had a
visit from and auditor who wanted to know why we were paying to store
it. He said dump it and we did. Right into the back of that infamous
Volvo station wagon. :-) That also how I got all my Teraks.

I couldn't get things that way. This was the government, and we
were forced to do paperwork to turn it in so it could be sold for $5.00
at a surplus sale. :-) The paperwork cost more than the income from
it. :-)

I still keep a couple of the magnets fopr biasing the digitizing
tablets around for erasing disks. The magnet was so powerful they
came packed inside a steel pipe when you bought a terminal. :-)

Our computer center had a color laser copier/printer (full
office sized one, not a desktop machine) and someone zapped the boot
floppy with one of those bias magnets. :-)

Works good, doesn't it? :-)

Yep!

And then, I used to have an Apollo Workstation. Now there's a boat
anchor.

O.K. Not sure of the weight or size for those. I'm impressed
enough with the 60+ pounds of my Sun Blade 2000 machines, and the
somewhat heavier rack-mount version (the Sun Fire 280R).

I would have put it in the 500-600 lb catagory. I had to dismantle it
to bring it home. Made a great little space heater int he winter.
Couldn't run it in the summer. Had an HDA even bigger than an RA80.

O.K. that sounds similar to the Intergraph -- except that I
could drive it home -- with a 3/4 ton pickup truck. :-)

NCR "upgrade" for the 3B2 computer systems? :-)

No -- how bad was it?

As you may or may not know, after the big court fight when AT&T actually
got permission to be in the computer business commercially, they decided
they didn't want to. So they brought in NCR to run that division. NCR's
first move was to "upgrade" the 3B2. Simple upgrade, really. Open case.
dump all 3B2 related parts into garbage can. Insert NCR designed and
manufactured M68K based computer into case. Upgrade completed. :-)

Ouch! The name is the same -- but it won't run the same
software at all. :-(

Well, NCR already had a line of computers and when AT&T acquired them they
became the AT&T Computer Division so it was obvious which way they would
go. That, and at that time the small unix system market was pretty much
M68K based. Unisys, Prime, NCR, Convergent, Tandy, all the real players.

Was Sun in business by then? They started with the 68K
processors. My 1/120 had a 10 MHz 68010 (same as the 3B1, but BSD
flavor). I *think* that the original SUN had the 68000 plain vanilla,
like my Coamos CMS-16/UNX.

That's what the IBM PC was supposed to be, but that's another story.

That would have been an interesting twist to history. I wonder
where things would be if that had happened?

[ ... ]

I got rid of a pair of Indigos a while ago. I do still have an O2.
Also SPARCStation-II, Ultra Sparc, a couple of original MAC's, a bunch
of newer M68K MAC's, a bunch of PDP-11's and VAXen. Also a bunch of
APPLE ]['s and Tandy computers. And I almost forgot all my 3B1's. One
of these days I hope to actually set up amd operate a computer museum.

O.K. List time:

1) Altair 680b (raised from a kit). This was the Motorla 6800
CPU version, not the Altair 8800 which used the Intel 8080.

[ ... ]

22 Mac Mini (Intel based)

23 ... Not counting the various MS-DOS and Windows machines over time.

I've mostly only listed the machines which I've actually used
for one thing or another.

And I missed the Emco-Maier Compact-5/CNC -- a little desktop
CNC lathe -- controlled by a 6502. :-)

[ ... ]

They maintained very good compatability along the whole line. But then,
they had to, they were running the phone system with them and that was
their bread and butter.

I wonder what they are using today?

Silly question. I have no doubt that the phone system is PC based.
At least that which isn't running embeded CPU's.

Ouch!

[ ... ]

Hmm ... The Tandy model 16 was a 68000 based system IIRC. I
wonder who did the port of unix to that one?

Microsoft. It was originally SYS III based Xenix. The original
documentation I got (because it was pre-release machine) was a mixture
of original AT&T Unix docs and Xeroxed MS Xenix docs.

O.K.

[ ... ]

But, I do know
there is a version of mini-Unix that runs on the TERAK 8510. LSI-11/02,
28K Words of memory and an 8" floppy. :-) There was a time when Unix
was not nearly as bloated as it later became.

O.K I have the LSI-11 two-wide card cage full of cards
(including the floppy interface) and a single matching floppy drive to
build one.

To build one what? Terak requires a quad wide backplane as the graphics
and memory all on one piggy-backed double board quad wide module.

O.K. Out of reach of the hardware which I have, then.


I also have a Bridgeport BOSS-3 CNC mill with a serious case of
electronics Altzheimer's built on a quad-wide LSI-11. That is coming
out, and being replaced with a linux-based EMC package.

Not surprised. I once sent a couple of 11/02 modules to a guy out west
to fix his Bridgeport. I'll bet there are a bunch of them running in
all the speed shops we have around here.

As far as I can tell, they had custom firmware on the chips as
well.

All the best.

And to you,
DoN.

--
Email: <dnichols@xxxxxxxxxxx> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
.